Ultra low power high-performance amplifier

Information

  • Patent Grant
  • 10177713
  • Patent Number
    10,177,713
  • Date Filed
    Monday, March 6, 2017
    7 years ago
  • Date Issued
    Tuesday, January 8, 2019
    5 years ago
Abstract
Methods, circuits, and apparatuses that provide Buffer Amplifier, containing Amplifiers and Buffer Drivers, one or more of the following: ultra low power Buffer Amplifier, capable of having high gain, low noise, high speed, near rail-to-rail input-output voltage span, high sink-source current drive capability for an external load, and able to operate at low power supply voltages. Methods, circuits, and apparatuses that provide regulated cascode (RGC) current mirrors (CM) capable of operating at low power supply and having wide input-output voltage spans.
Description
FIELD OF DISCLOSURE

This disclosure related to improvements in current mirrors, current sources, amplifiers, and output buffer drives for use in integrated circuits (ICs).


BACKGROUND

Operating ICs under ultra low currents and low power supplies, in complementary metal-oxide semiconductor (CMOS) technology, pose serious challenges in the design of integrated circuits. Low operating currents cause lower speeds and lower gain and higher noise in an IC. Also, rail-to-rail operations for ICs becomes a necessity given that signal-to-noise requirements at low power supplies demand input and output terminals of ICs to get as close as possible to the power supplies.


SUMMARY OF THE DISCLOSURE

An aspect of the present disclosure is to make small and low cost current sources, current mirrors, amplifiers, buffers drivers, and buffer amplifiers that can operate with one or more of the following characteristics: (a) wide input-output voltage span, (b) low power supply voltage, (c) low power consumption, (d) low noise, (e) fast dynamic response, (f) symmetric design to minimize systematic errors, (g) use simple design that generally improves performance to specifications over operating and process variation, and/or (h) make the IC rugged for long term manufacturing using standard CMOS fabrication process that is inexpensive, and readily available at multiple fabrication factories, thus easing process node portability.


Another aspect of the disclosure includes a current mirror with high impedance, wide input-output span, and low drain voltage (VDDVDD). Another aspect of this disclosure is to make current sources and current mirrors with high output impedance, wide input-output voltage range, and operating with low power supply voltage. This is accomplished by a method of making a ‘current source’ or a ‘current mirror’ comprising regulated cascode current mirror (RGC-CM) coupled with diode connected self cascode (DCSC). This may also be accomplished by another method of making a ‘current source’ or a ‘current mirror’ comprising RGC-CM couple with inverting current mirror amplifier (ICMA). Moreover, this goal is met by another method of making a ‘current source’ or a ‘current mirror’ comprising RGC-CM coupled with a composite amplifier (CSGA) that contains common source armplifier (CSA) and common gate amplifier (CGA) with the gate of CGA connected to source of CSA.


Another aspect of the disclosure further includes an amplifier using the disclosed current mirrors discussed above to make amplifiers with high gain and wide input-output range, and operating with low power supply voltage. This is accomplished by a method of making an ‘amplifier’ comprising plurality of the disclosed RGC-CM couple with DCSC. This may also be accomplished by another method of making an ‘amplifier’ comprising plurality of the disclosed RGC-CM coupled with ICMA. Moreover, this goal is also met by method of making an ‘amplifier’ utilizing plurality of the disclosed RGC-CM coupled with a composite amplifier (CSGA).


Another aspect of the disclosure further includes a floating current source with low VDDVDD that also operates fast. Another aspect of this disclosure is to emulate the function of a floating current sources (FCS) that equalizes upper and lower current sources and is capable of operating with low power supply voltages. This may be accomplished by a method utilizing cascoded PMOSFETs (e.g., one PMOSFET on top and second PMOSFET in the middle), and cascoded NMOSFETs (e.g., one NMOSFET on top and second NMOSFET in the middle). The middle PMOSFET and the middle NMOSFET drain and source currents are crisscrossed and fed to one another. A lower regulating circuit holds the VGS of the middle NMOSFET constant by regulating the gate voltage of the lower NMOSFET. An upper regulating circuit holds the VGS of the middle PMOSFET constant by regulating the gate voltage of the upper PMOSFET. As a result the net sum of current in the upper PMOSFET and lower NMOSFET is equalized.


A further aspect of this disclosure is an amplifier using the disclosed FCS so that the amplifier's upper and lower current sources are equalized in order to improve amplifier's performance. Some of such improvements are in reducing the amplifier's offsets due to current asymmetries when the amplifier's input has a wide common mode span, while enabling the amplifier to operate with low power supply voltage.


A further aspect of this disclosure is reducing an amplifier's output noise, when the amplifier consumes low currents, operates at low VDDVDD, while the amplifier has wide input-output voltage span, and it is fast. Another aspect of this disclosure is to reduce an amplifier's noise by narrow banding it. When the amplifier's inputs are imbalanced in response to a large transient input signal, to make up for lost speed due to narrow-banding the amplifiers, the operating current of the amplifiers is dynamically and rapidly boosted to IPeak=h×IQ (in the current boost-on phase) where IQ is the steady state operating current of the amplifier set at low current levels to save power consumption. After the amplifier's inputs approach steady state and are near balance (not precisely equal, just roughly in balance), the amplifier's operating current is dynamically and slowly decreased (current boost-off phase) back to IQ levels. The amplifier dynamic response is improved by enhancing both the Slew Rate (SR) and the settling time (τS) of the amplifier. In summary, to rejovinate the (narrow banded) amplifier's SR while its' inputs are imbalanced due to receiving a large-signal transient, the amplifiers' operating current receive a very fast IPeak pulse during the current boost-on phase. Moreover, to rejovinate the (narrow banded) amplifier's τS when the amplifier's inputs approach balance and amplifiers enters the boost-off phase, instead of shutting off Ipeak very fast, the amplifier's operating current starts decaying slowly (following a time constant that tracks the amplifier's AC response) from Ipeak towards IQ. Accordingly, when the amplifier operating current is at IQ equilibrium or steady state, the amplifier's noise is reduced since it is narrow banded.


Another aspect of the disclosure herein is a buffer driver that operates with low VDDVDD, has wide input output voltage span, and is fast by operating chiefly in current mode. A further aspect of this disclosure is an amplifier using the disclosed buffer driver that can operate with low VDD, have near rail-to-rail input-outputs spans, and mostly operate in current mode which improve speed at low currents. This is accomplished by a method of making an ‘amplifier’ comprising plurality of minimum current selectors (MCS) or loser take all (LTA) which, directly or indirectly, monitor the sink-source currents of the buffer driver's output field effect transistors (FETs). Concurrently, a non-inverting current mirror amplifier (NICMA) or inverting current mirror amplifier (ICM) or inverting current feedback amplifiers (ICFA) would regulate the minimum stand-by currents for either the inactive sink output transistor or the inactive source output transistors. Also, in order to lower the current consumption associated with monitoring the sink source output transistor currents, a complementary non-inverting current mirror (CNICM) is utilized to (rectify) curb the sink-source signals before they are fed to the MCS or LTA.


Aspects of the embodiments disclosed herein further include that they can often be fabricated in standard digital CMOS; and embodiments have small size (e.g., for low cost and high volume applications); and the embodiments typically operate MOSFETs in subthreshold so they can operate at ultra low currents and low power supply voltages needed in particular to emerging wireless and battery less applications.


Aspects of the embodiments disclosed herein further include a method of operating an ultra low power Buffer Amplifier (BA), containing an Amplifier coupled with a Buffer Driver (BD), comprising: increasing gain and widening the input-output voltage span of the Amplifier utilizing a plurality of regulated cascode (RGC) current mirrors (RGC-CM) where each RGC-CM is made of at least one of the following three circuits: 1) a diode connected self cascode (DCSC) coupled with a common source amplifier (CSA); 2) a current mirror amplifier (CMA), inverting or non-inverting type, that contains a common source amplifier (CSA); and 3) a CSA coupled with one common gate amplifier (CGA) wherein the common gate terminal of the CGA is connected to the common source terminal of the CSA; lowering the minimum operating power supply and reducing offset of the Amplifier by utilizing a current equalizer circuit that emulates the function of a floating current source (FCS) having at least two complementary cascoded current sources made of field effect transistors (FETs), wherein middle cascoded FET's gate to source voltages (VGS) are held constant by regulating the VGS of the lower FETs, whose currents are equalized and mirrored into the Amplifier's bias network, where the lower FET's source terminals are connected to the power supplies; lowering output noise of the Amplifier, by narrow banding the Amplifier's high gain node, while concurrently rejuvenating the narrow banded Amplifier's speed by utilizing one of the following circuits: 1) minimum current select (MCS); and 2) loser take all (LTA) circuits to generate dynamically boosted operating current when the Amplifier's inputs are intermittently imbalanced; and lowering the minimum operating power supply, having near rail-to-rail input output voltage span, having high-speed, providing large sink-source current for output load, while regulating the operating current in the inactive sink-source transistor of the Buffer Amplifier (BA) by utilizing a Buffer Driver (BD) that contains at least one of the following circuits: 1) MCS; 2) LTA; 3) current mirror amplification (CMA); and 4) complementary current mirror that enable the Buffer Driver (BD) to chiefly operate and process signals in current mode.


Aspects of the embodiments disclosed herein further include a method of reducing output noise in an amplifier comprising: narrow banding the amplifier to keep the amplifier's static current consumption low; rejuvenating the dynamic response of the narrow banded amplifier by dynamically boosting the amplifier's operating current when the amplifier's inputs receive a large transient signal that cause an imbalance at the amplifier's inputs; and returning back to the steady state conditions when the amplifiers inputs are substantially equalized and shutting off the dynamic boosting of the amplifier's operating current when the amplifier's current consumption returns to low levels and the amplifier's output noise is reduced. The method further comprising: connecting a first capacitor to the high impedance or high gain node of the amplifier to narrow band the amplifier; and wherein the first capacitor can be an active or a passive capacitor that is intrinsic or extrinsic at the high impedance or high gain node of the amplifier. The method further comprising: rejuvenating the dynamic response of the narrow banded amplifier by speeding up the amplifier's slew rate and settling time upon detecting an imbalance at the amplifier's inputs by utilizing one of a: 1) a loser take all (LTA) circuit; and 2) minimum current selector (MCS) circuit; applying the output of the LTA circuit or MCS circuit to generate a ‘boost on’ or ‘boost off’ signal; using the ‘boost on’ signal to boost the operating current of the amplifier; and using the ‘boost off’ signal to return the amplifier's operating current back to the low static current at steady state condition and attaining lower output noise for the amplifier at steady state conditions. The method further comprising: using the ‘boost on’ signal to rapidly boost the operating current of the amplifier to increase the amplifier's slew rate (SR); using the ‘boost off’ signal to generate a slow declining current, with a slow decay to zero to speed up the settling time (τs) of the amplifier; improving the dynamic response of the amplifier by optimizing both its SR and τs; and using the ‘boost off’ signal to return the amplifier's operating current back to the low static current at steady state condition and attaining low output noise for the amplifier during steady state conditions.


Aspects of the embodiments disclosed herein further include a buffer driver circuit comprising: first output driver field effect transistors (FETs) having the function of sinking and sourcing currents for an external load; at least one of a minimum current selector (MCS) signal or at least one loser take call (LTA) signal having the function of monitoring and processing the sink-source currents of the first output driver FETs; at least one of a group consisting of a: 1) current mirror amplifier (CMA); 2) an inverting CMA (ICMA); and 3) non-inverting CMA (NICMA) having the function of receiving the MCS or LTA signals, and utilizing the MCS or LTA signals for regulating and controlling the current in the inactive sink-source FET; a first buffer driver that utilizes at least one MCS or LTA signals and at least one of the CMA, ICMA and NICMA; and wherein the buffer driver has the function of sinking and sourcing current for external loads and regulating a minimum operating current in the inactive sinking or sourcing FET. The circuit further comprising: one of the group consisting of: 1) a first complementary non-inverting current mirror (CNICM); and 2) a first complementary inverting current mirror (CICM) having a function of monitoring the sink-source output driver FETs current and generating rectified sink-source signals; and providing the rectified sink-source signals to the minimum current selector (MCS) or loser take call (LTA) circuits to process the sink-source output driver FETs signals. The circuit capable of being used in a first amplifier further comprising: wherein the first amplifier utilizes the first buffer driver circuit in order for the first amplifier to be coupled with the buffer driver to be capable of sinking and sourcing current for external loads and regulating a minimum operating current in the inactive sinking or sourcing FET.


Aspects of the embodiments disclosed herein further include a method of operating at least one regulated cascode (RGC) current mirror (RGC-CM) comprising: supplying voltage to the at least one RGC-CM from a positive supply voltage (VDD) and a negative supply voltage (VSS); increasing the output resistance of the at least one RGC-CM by a first auxiliary amplifier; and widening the voltage span of the input-output terminals of the at least one RGC-CM by generating a direct current (DC) voltage shift from at least one diode connected self cascode (DCSC) coupled to the first auxiliary amplifier. The method further comprising: utilizing a first amplifier that contains a plurality of RGC-CMs, wherein each of the plurality of RGC-CMs are utilized in the first amplifier to function as current mirrors; delivering power to the first amplifier by a positive supply voltage (VDD) and a negative supply voltage (VSS); increasing the gain of the first amplifier by utilizing the plurality of RGC-CMs; and widening the input-output span of the first amplifier by utilizing the plurality of RGC-CMs.


Aspects of the embodiments disclosed herein further include at least one regulated cascode (RGC) current mirror (RGC-CM) circuit comprising: cascoded transistors having a first transistor placed in series at a first node with a second transistor wherein the output of the RGC-CM is the drain terminal of the second transistor; a first diode connected self cascode (DCSC) having a diode connected third transistor in series with a fourth transistor, wherein the gate terminals of the third transistor and the fourth transistor are connected together and wherein the source of the third transistor is connected to the drain of the fourth transistor at the second node, wherein the source of the fourth transistor is connected to the first node; and a first auxiliary amplifier (AA) whose input is connected to the second node, and wherein the output of the first AA is connected to the gate terminal of the second transistor. The circuit utilized in a first amplifier further comprising: a plurality of the at least one regulated cascode (RGC) current mirrors (RGC-CMs); wherein each of the plurality of RGC-CMs function as current mirrors; wherein the gain of the first amplifier is increased by utilizing the plurality RGC-CMs; and wherein the input-output span of the first amplifier is widened by utilizing the plurality of RGC-CMs.


Aspects of the embodiments disclosed herein further include a method to operate a regulated cascode (RGC) current mirror (RGC-CM) comprising: providing power to the RGC-CM by a positive supply voltage (VDD) and a negative supply voltage (VSS); operating input and output signals in a current mode by utilizing a first inverting current mode amplifier (ICMA) that includes a first inverting current mirror (ICM); generating amplification through the first ICMA to perform the function of a first auxiliary amplifier (AA); increasing the output resistance of the RGC-CM by utilizing the first ICMA; and widening the input-output terminal spans of the RGC-CM by utilizing the first ICMA. The method of comprising: utilizing a first amplifier having a plurality of regulated cascode (RGC) current mirrors (RGC-CMs) wherein each of the plurality of RGC-CMs are utilized in the first amplifier function as current mirrors.


A regulated cascode (RGC) current mirror (RGC-CM) circuit comprising: cascoded transistors having a first transistor placed in series at a first node with a second transistor wherein the output of the RGC-CM is the drain terminal of the second transistor; a first current mirror (CM) with a diode connected third transistor and a fourth transistor; a first current mirror amplifier (CMA) having the first current mirror, and a third transistor whose source terminal, which is the input of CMA, is connected to the first node and its drain terminal is connected to the gate and drain terminals of the third transistor; wherein the output of CMA, which is the drain terminal of the fourth transistor, is connected to the gate terminal of second transistor; and wherein the CMA functions as the auxiliary amplifier for the RGC-CM in order to increase the output resistance of the RGC-CM. The circuit used in a first amplifier further comprising: wherein each of the plurality of RGC-CMs that are utilized in the first amplifier function as current mirrors; wherein the gain of the first amplifier is increased by utilizing the plurality RGC-CMs; and wherein the input-output span of the first amplifier is widened by utilizing the plurality of RGC-CM.


Aspects of the embodiments disclosed herein further include a regulated cascode (RGC) current mirror (RGC-CM) circuit comprising: cascoded transistors having a first transistor placed in series at a first node with a second transistor wherein the output of the RGC-CM is the drain terminal of the second transistor; a first common source amplifier (CSA); a first common gate amplifier (GGA); wherein the inputs of the first CSA are connected to the source terminal of the first transistor and the first node; wherein the output of the first CSA is connected to the input of the first CGA; wherein the common gate terminal of the first CGA is connected to the common source terminal of the first CSA; and wherein the first CGA's output terminal is connected to the gate of the second transistor. The circuit used in a first amplifier further comprising: a plurality of regulated cascode (RGC) current mirrors (RGC-CMs); and wherein each of the plurality of RGC-CMs that are utilized in the first amplifier function as current mirrors; wherein the gain of the first amplifier is increased by utilizing the plurality RGC-CMs; and wherein the input-output span of the first amplifier is widened by utilizing the plurality of RGC-CM.


Aspects of the embodiments disclosed herein further include a regulated cascode (RGC) current mirror (RGC-CM) circuit comprising: cascoded transistors having a first transistor placed in series at a first node with a second transistor wherein the output of the RGC-CM is the drain terminal of the second transistor; a first common source amplifier (CSA), with a built in offset, having a third transistor and a fourth transistor, whose source terminals are connected to the second node, wherein the gate of the third transistor which is one of the inputs of CSA is connected to the first node, wherein the other terminal of the CSA, which is the gate and drain terminals of the fourth transistor are connected together and are connected to the source terminal of the first transistor; and a first common gate amplifier (CGA), comprising a fifth transistor whose gate terminal is connected to the second node, where the input of the CGA which is the source terminal of is connected to the first node, and the output of the CGA which is the drain terminal of the fifth transistor is connected to the gate terminal of the second transistor. The circuit used in a first amplifier, further comprising: a plurality of regulated cascode (RGC) current mirrors (RGC-CMs), wherein each of the plurality of RGC-CMs that are utilized in the first amplifier function as current mirrors; wherein the gain of the first amplifier is increased by utilizing the plurality RGC-CMs; and wherein the input-output span of the first amplifier is widened by utilizing the plurality of RGC-CMs.


Aspects of the embodiments disclosed herein further include a current equalizing circuit comprising: a positive supply voltage (VDD) and a negative supply voltage (VSS); a first Positive Metal Oxide Silicon Field Effect Transistor (PMOSFET) and second PMOSFET forming a cascoded current source, wherein the drain of the first PMOSFET is connected to the source of the second PMOSFET at the first node; a first Negative Metal Oxide Silicon Field Effect Transistor (NMOSFET) and second NMOSFET forming a cascoded current source, wherein the drain of the first NMOSFET is connected to the source of the second NMOSFET at the second node; wherein the second PMOSFET drain terminal is connected to the source terminal of and second NMOSFET at the second node; wherein the second NMOSFET drain terminal is connected to the source terminal of and second PMOSFET at the first node; wherein a first regulating circuit keeps the gate-to-source voltage of the second PMOSFET substantially constant by regulating the gate-to-source voltage of the first PMOSFET; wherein a second regulating circuit keeps the gate-to-source voltage of the second NMOSFET substantially constant by regulating the gate-to-source voltage of the first NMOSFET; and wherein the current in the first PMOSFET and first NMOSFET are substantially equalized. The circuit used in a first amplifier further comprising: the first amplifier including the first current equalizing circuit of claim 22.


Aspects of the embodiments disclosed herein further include a current equalizing method comprising: delivering power to the current equalizer by a positive supply voltage (VDD) and a negative supply voltage (VSS); lowering the minimum operating power supply and reducing offset of the Amplifier (A) by utilizing a current equalizer circuit that emulates the function of a floating current source (FCS) containing two complementary cascode current sources, wherein the cascoded field effect transistors (FETs) gate to source voltages (VGS) are held constant by regulating the VGS of the lower FETs, whose currents are equalized and mirrored into the amplifier's bias network, and whose source terminals are connected to the power supplies.





BRIEF DESCRIPTION OF THE DRAWINGS


FIG. 1A is a schematic circuit diagram of the embodiment illustrating an amplifier utilizing FIG. 1B.



FIG. 1B is a schematic circuit diagram of the embodiment illustrating a current mirror utilizing a regulated cascode current mirror (RGC-CM) coupled with diode connected self cascode (DCSC)



FIG. 1C is a schematic circuit diagram of the embodiment illustrating an amplifier utilizing FIG. 1D.



FIG. 1D is a schematic circuit diagram of the embodiment illustrating a current mirror utilizing a RGC-CM coupled with an inverting current mirror amplifier (ICMA).



FIG. 1E is a schematic circuit diagram of the embodiment illustrating an amplifier utilizing FIG. 1F.



FIG. 1F is a schematic circuit diagram of the embodiment illustrating a current mirror utilizing a RGC-CM coupled with a composite amplifier (CSGA).



FIG. 1G is a schematic circuit diagram of an embodiment illustrating a prior art RGC-CM.



FIG. 1H is a circuit simulation showing a frequency response of the amplifier illustrated FIG. 1A.



FIG. 1I is a circuit simulation showing a transient response and current consumption of the amplifier illustrated FIG. 1A.



FIG. 1J is a circuit simulation showing a frequency response of the amplifier illustrated FIG. 1C.



FIG. 1K is a circuit simulation showing a transient response and current consumption of the amplifier illustrated FIG. 1C.



FIG. 1L is a circuit simulation showing a frequency response of the amplifier illustrated FIG. 1E.



FIG. 1M is a circuit simulation showing a transient response and current consumption of the amplifier illustrated FIG. 1E.



FIG. 2A is a schematic circuit diagram of the embodiment illustrating an amplifier utilizing the first floating current source (FCS200A) embodiment



FIG. 2B is a schematic circuit diagram of the embodiment illustrating an amplifier utilizing the second floating current source (FCS200B) embodiment



FIG. 2C is a schematic circuit diagram of the embodiment illustrating an amplifier utilizing the third floating current source (FCS200C) embodiment



FIG. 2D is a schematic circuit diagram of a prior art floating current source (FCS) embodiment



FIG. 2E is a circuit simulation showing transient response, current consumption, and change in the FCS200A current in response to change in input-output voltage of the amplifier illustrated FIG. 2A.



FIG. 2F is a circuit simulation showing transient response, current consumption, and change in the FCS200B current in response to change in input-output voltage of the amplifier illustrated FIG. 2B.



FIG. 2G is a circuit simulation showing transient response, current consumption, and change in the FCS current in response to change in input-output voltage of the amplifier illustrated FIG. 2C.



FIG. 3A is a schematic circuit diagram of the embodiment illustrating an amplifier utilizing a first noise reduction and speed boost circuit.



FIG. 3B is a schematic circuit diagram of the embodiment illustrating an amplifier utilizing a second noise reduction and speed boost circuit.



FIG. 3C is a circuit simulation showing output noise of the amplifier illustrated in FIG. 3A with and without the noise reduction & speed boost.



FIG. 3D is a circuit simulation showing transient response, current consumption of the amplifier illustrated in FIG. 3A, with and without the noise reduction & speed boost.



FIG. 4A is a schematic circuit diagram of a prior amplifier's gain stage with near rail-to-rail input and output voltage span, without a buffer driver.



FIG. 4B is a schematic circuit diagram of the embodiment illustrating a buffer driver utilizing complementary non-inverting current mirror (CNICM), minimum current selector (MCS), and inverting current mirror amplifier (ICMA).



FIG. 4C is a schematic circuit diagram of the embodiment illustrating a buffer driver utilizing complementary non-inverting current mirror (CNICM), loser take all (LTA), and non-inverting current mirror amplifier (NICMA).



FIG. 4D is a schematic circuit diagram of the embodiment illustrating a buffer driver utilizing minimum current selector (MCS), and inverting current mirror amplifier (ICMA).



FIG. 4E is a circuit simulation showing the sink-source output FET currents and IDD as a function of external resistive load for an amplifier containing the gain stage of FIG. 4A coupled with buffer driver illustrated in FIG. 4B.



FIG. 4F is a circuit simulation showing sink-source output FET currents and IDD as a function of external resistive load for an amplifier containing the gain stage of FIG. 4A coupled with buffer driver illustrated in FIG. 4C.



FIG. 4G is a circuit simulation showing sink-source output FET currents and IDD as a function of external resistive load for amplifier containing the gain stage of FIG. 4A coupled with buffer driver illustrated in FIG. 4D.



FIG. 5 is a block level diagram showing the Buffer Amplifier having the Amplifier coupled with the Buffer Driver section.





DETAILED DESCRIPTION
Definitions, Acronyms and Abbreviations

The following terms, definitions, acronyms, term usages and abbreviations are explained below and used throughout this application:













TERMS
DESCRIPTION







IC or chips
Integrated circuits


Amplifier
Has high input and high output impedance and the



amplifier amplifies signals applied to it its input, unless



otherwise specified


buffer driver
Has high input impedance and low output impedance with



high current drive capability so the buffer driver can sink



and source currents for an external load, unless



otherwise specified


buffer amplifier
Includes an amplifier coupled with a buffer driver,



combination of which can amplify signals applied to the



buffer amplifier inputs. Buffer amplifier has high input



impedance, and low output impedance with high output



current drive capability so it can sink and source currents



for an external load, unless otherwise specified


VDD or positive rail
Positive power supply voltage


VSS or negative rail
Negative power supply voltage


Rail-to-rail input
Input and output terminals spanning (substantially near)


and output
to VDD and VSS


GND
ground voltage, which can be the same as VSS


VIN
Input voltage containing the negative terminal VIN− and



the positive terminal VIN−


VINCM
Common mode input voltage


VOUT
output voltage


Bias current or reference
Used interchangeably


current or current source



IDD
Current flowing through VDD or current consumption


FET
Field Effect Transistors


JFET
Junction Field Effect Transistors


BJT
Bipolar Junction Transistors


CMOS
Complementary Metal Oxide Semiconductor (may be



operated in subthreshold or normal regions)


BiCMOS
BJT and CMOS transistors fabricated on the same water


MOSFET
Metal Oxide Semiconductor Field Effect Transistors (may



be operated in subthreshold or normal regions)


MOSFET's linear or
Used interchangeably


triode or resistive



regions of operations



Subthreshold
MOSFET's Subthreshold region of operation


CMOS, PMOS, and
Used interchangeably


NMOS or CMOSFET.



PMOSFET, and



NMOSFET, respectively



Gate, source, and drain
Used interchangeably


or gate terminals,



source terrninal, and



drain terminals



respectively



W/L of a MOSFET
Width over length ratio (aspect ratio) of MOSFETs


W/L of a MOSFET or
Used interchangeably


aspect ratios



Ohms per square or
Resistivity of material per square area


Ω/square



Terms applied to the
Used interchangeably


W/L of MOSFETs such



as ‘predetermined’ versus



‘programmed’ versus ‘set’



C or ° C.
Used interchangeably and shows unit of temperature in



Celsius


Ω
Ohms, unit of measurement for resistivity


A
Ampere


V
Volt


Mm
Micro-meter or 10−6 meter (e.g., W = 4 μm or W = 4 × 10−6



meter


N
Nano or 10−9 (e.g., nA = nano ampere or 10−9A)


P
Pico or 10−12(e.g., pF = pico Farad or 10−12F


M
MOSFET carrier mobility


VT
Thermal voltage


COX
Gate oxide capacitance of a MOSFET


Ce
Effective capacitance


CeMXXXX
Effective input gate terminal capacitance of a MOSFET



(e.g., CeM316A is the effective capacitance looking into the



gate of FET M316A)


Cexxxx
Effective capacitance at node xxxx (e.g.,



Ce314A is the effective capacitance at node 314A)


VTH
Threshold voltage of a MOSFET


VA or 1/λ
Used interchangeably to show MOSFET early voltage


β
Beta = β = μ × COX to show a MOSFET gain


M
To show MOSFET mobility exponent factor used in



device equations


H
To show MOSFET subthreshold slope factor used in



device equations


VOFS
Offset voltage between two MOSFETs, mostly referring to



a CMOS amplifier's input offset voltage


Vxxxx
Voltage at node xxxx (e.g.,V108A is the voltage at node



108A)


VDS or VDSsat or Von
Used interchangeably is the drain terminal voltage to



source terminal voltage of a MOSFET, including in the



saturation region


VGS
Gate terminal voltage to source terminal voltage of a



MOSFET


VGMxxx
Voltage at the gate terminal of a MOSFET (e.g.,VGM128D



is the voltage at the gate terminal of M128D)


VDMxxxx
Voltage at the drain terminal of a MOSFET (e.g.,VDM128D



is the voltage at the drain terminal of M128D)


VSMxxxx
Voltage at the source terminal of a MOSFET



(e.g.,VSM128D is the voltage at the source terminal of



M128D)


VGSMxxxx
Gate to source voltage of a MOSFET



Mxxxx (e.g.,VGSM128D is VGS of M128D)


VDSMxxxx
Drain to source voltage of a MOSFET



Mxxxx (e.g.,VDSM128D is VDS of M128D)


ΔVGS
Difference between the gate to source voltage of two



MOSFETs


ΔvO or ΔVOUT
Change in an output voltage of a function (e.g., amplifier's



output or current mirror's output)


ΔvX
Small change in a voltage


ΔiX
Small change in a current


IDS or ID or IDMxxxx
Used interchangeably and it is the drain to source current


or IMxxx
of a MOSFET (e.g., IM117D is IDS of M117D)


Ixxxx
It is a current source (e.g., I107B is the current source



#107B)


gm or gmMOSFET
To show transconductance of MOSFET where gm



ID/VT for MOSFET operating in subthreshold (ignoring



second order effects, such as η or



VTH or VA for clarity of description)


rO or rds
Output resistance of MOSFET or ∝ VA/ID for MOSFET



operating in subthreshold


ROUT or RO
Used interchangeably and shows the output resistance of



a function such as that of a current source, current mirror,



transconductance amplifier, or an auxiliary amplifier


Zexxxx
Effective impedance or effective resistance at node xxxx



(e.g., Ze314A is the effective impedance at node 314A)


TC
Temperature coefficient


AC response
Small signal frequency output versus input response of a



circuit


AV
Main amplifier's open loop gain at DC


FU
Unity gain frequency of amplifier (stated in Hz = Hertz)


PM
Phase margin of amplifier at FU (stated in ° = degrees)


SR
Slew rate of amplifier (stated in V/μS or volts per micro



seconds)


tS
Settling time of amplifier's VOUT to 5% of steady state



(stated in μS = micro seconds)


PSRR
Power supply rejection ratio of amplifier


‘a’ through ‘z’
To show the ratio of MOSFET W/Ls or their aspect ratios



used in a circuit design


~ or ≈
Approximately equal


<<
Significantly less than (e.g., IDM304 << 2i



Means IDM304 is significantly less than 2i and IDM304 can



be zero or substantially close to it



Approximately equal



As a function of or proportional to


⇒ or →
Implication, results


SC
MOSFET self-cascode


DCSC or DCSxxxx
MOSFET diode connected self cascode or DCSC number



xxxx


RGC or RGCxxxx
Regulated Cascode or RGC number xxxx


CM or CMxxxx
current mirror or CM number xxxx


RGC-CM or
Regulated Cascode current mirror or RGC-CM number


RGC − CMxxxx
xxx


RGC-CS
Regulated Cascode current source


RGC or RGC-CM or
Used interchangeably, unless otherwise noted


RGC-CS



Axxxxx
Amplifier used in a regulating circuit (e.g., AP200C is an



amplifier number P200C with PMOSFET inputs)


AAUX or AA
Auxiliary amplifier used for example in regulate cascode



current mirror (RGC-CM)


MCS or MCSxxxx
Minimum current selector or MCS number xxxx


Buffer or Buffer Driver or
Used interchangeably for buffer driver or BUF number


BUFxxxx
xxxx


Min (MMxxxx, IDMyyyy)
Selecting minimum of 2 FET currents (e.g. when



IDM412D < IDM414D then Min (IDM412D, IDM414D) = IDM412D


LTA or LTAxxxx
Loser take all or LTA number xxxx


LTA (IDxxxxx, IDyyyyy)
Similar to MCS, the loser of 2 FET currents takes all (e.g.



when IDM412D < IDM414D then LTA (IDM412D, IDM414D) =



IDM412D


WTA or WTAxxxx
Winner take all


LTAA or LTAAxxxx
Loser take all amplifier or LTAA number xxxx


ICM or ICMxxxx
Inverting current mirror or ICM number xxxx


NICM or NICMxxxx
Non-Inverting current mirror or NICM number xxxx


ICMA or ICMAxxxx
Inverting current mirror amplifier or ICMA number xxxx


NICMA or NICMAxxxx
Non-Inverting current mirror amplifier or NICMA number



xxxx


FCTA
Folded cascode transconductance amplifier


FCTA, or folded cascode
Used interchangeably


amplifier, or cascode



amplifier, or



transconductance



amplifier



FCS or FCSxxxx
Floating current source or FCS number xxxx


CSA or CSAxxxx
Common source amplifier or CSA number xxx


CGA or CSAxxxx
Common gate amplifier or CSA number xxxx


KCL
Kirchhoff's Current Law


KVL
Kirchhoff's Voltage Law









Numerous embodiments are described in the present application and are presented for illustrative purposes only and is not intended to be exhaustive. The embodiments were chosen and described to explain principles of operation and their practical applications. The present disclosure is not a literal description of all embodiments of the disclosure(s). The described embodiments also are not, and are not intended to be, limiting in any sense. One of ordinary skill in the art will recognize that the disclosed embodiment(s) may be practiced with various modifications and alterations, such as structural, logical, and electrical modifications. For example, the present disclosure is not a listing of features which must necessarily be present in all embodiments. On the contrary, a variety of components are described to illustrate the wide variety of possible embodiments of the present disclosure(s). Although particular features of the disclosed embodiments may be described with reference to one or more particular embodiments and/or drawings, it should be understood that such features are not limited to usage in the one or more particular embodiments or drawings with reference to which they are described, unless expressly specified otherwise. The scope of the disclosure is to be defined by the claims.


Although process (or method) steps may be described or claimed in a particular sequential order, such processes may be configured to work in different orders. In other words, any sequence or order of steps that may be explicitly described or claimed does not necessarily indicate a requirement that the steps be performed in that order. The steps of processes described herein may be performed in any order possible. Further, some steps may be performed simultaneously despite being described or implied as occurring non-simultaneously (e.g., because one step is described after the other step). Moreover, the illustration of a process by its depiction in a drawing does not imply that the illustrated process is exclusive of other variations and modifications thereto, does not imply that the illustrated process or any of its steps are necessary to the embodiment(s). In addition, although a process may be described as including a plurality of steps, that does not imply that all or any of the steps are essential or required. Various other embodiments within the scope of the described disclosure(s) include other processes that omit some or all of the described steps. In addition, although a circuit may be described as including a plurality of components, aspects, steps, qualities, characteristics and/or features, that does not indicate that any or all of the plurality are essential or required. Various other embodiments may include other circuit elements or limitations that omit some or all of the described plurality.


Throughout this disclosure, the body terminal of PMOSFETs can be either connected to their respective PMOSFET source terminals or to the positive power supply, VDD. Similarly here, the body of NMOSFETs can be either connected to their respective NMOSFET source terminal or to the negative power supply, VSS. Moreover, the negative supply voltage, VSS, can be alternatively connected to the ground (GND) potential. Given that one of these teaching's target application are for ultra low power and portable electronics, the transistors utilized in circuits operate under the subthreshold region, but it is also possible to operate transistors, throughout this disclosures and illustrations, in the normal regions. Throughout this disclosure and its illustrations, current sources or current mirrors may be constructed with single FETs or cascodes one, depending on cost-performance considerations such as die size, output impedance, gain, speed, head room, amongst others. Illustrations in some embodiments utilizing NMOSFET (e.g., in a current mirror or RGC-CM), can be modified to utilize their complementary FET counterparts (i.e., PMOSFETs). Moreover, embodiments utilizing amplifier's with PMOSFET input stages can also be modified to utilize NMOSFET input stages, or complementary input stage (i.e., both PMOSFET and NMOSFET operating in parallel) for rail to rail input dynamic range. Moreover, embodiments can utilize amplifiers, or their variations obvious to one skilled in the art, that use double PMOSFETs coupled with NMOSFET level shifters or double NMOSFETs coupled with PMOSFET level shifters to attain rail-to-rail input operations. The U.S. Provisional Patent Application Ser. No. 62/304,373 utilizes an amplifier's input stage based in complementary input pairs (i.e., both PMOSFET and NMOSFET operating in parallel) for rail to rail input dynamic range. The U.S. Provisional Patent Application Ser. No. 62/320,512 and the U.S. Provisional Patent Application Ser. No. 62/415,496 utilize an amplifier's input stage that are based in double PMOSFETs coupled with NMOSFET level shifters to attain input voltage rail-to-rail operations. For demonstrative clarity and simplicity, the assumption is made that VA, β, η, gm, gds, rds, rO, and ID for PMOSFETs and NMOSFETs are substantially equal unless otherwise specified. In order to demonstrate achievable typical specifications, simulations are performed on some of the circuits illustrating the embodiments. These simulations are not intended to guarantee the embodiment performance to a particular range of specifications. Circuit simulations use the Top-Spice simulator, and are based on approximate device models for a typical standard 0.18 μm CMOS process fabrication. To simulate for sensitivity of design to device model variations (i.e., simulate for design margin), this disclosure provides some worst case simulations (WC) that subject the circuit embodiments to variations in device parameters (e.g., manufacturing process fluctuations in VTH and COX), which can indicate the circuit's design sensitivity to normal manufacturing (wafer lot-to-lot) variations. For example, WC circuit simulations illustrate variations on performance specifications such as slew rate of settling time as a function of varying VTH by ±10% from center value, and independently from varying COX by ±5% from center value (i.e., 16 process corner combinations).


Note that the teachings in this disclosure are applicable to high voltage or high current or high speed applications, and combinations thereof, as well. The teachings in this disclosure may be applied to other manufacturing fabrication processes that contain transistors, such as fabrication processes with deep sub-micron CMOS, BiCMOS, BJT, and JFET, amongst other device and manufacturing platforms that make devices that can function as transistors.


Current mirrors, current sources, amplifier, and output buffers are building blocks for any integrated circuits (IC), including for higher order functions such as analog to digital converters (ADC), digital to analog converters (DAC), regulators, references, fitters, data acquisition systems (DAS), and other building blocks in any analog and mixed mode ICs, and system on a chip (SOC). It is advantageous for IC building blocks such as current sources, current mirrors, amplifiers, and buffer drivers to: (1) consume low currents, (2) operate with low power supply voltages, especially for portable applications, (3) have near rail-to-rail input-output spans since there is little voltage headroom to waste, especially under low power supply voltages, (4) have high output impedance or high output gain, especially under ultra low operating currents when gain is lower, (5) have high-speeds, especially under low operating currents when speed gets slower, (6) have low noise, especially under low operating currents when noise increases, (7) have their other performance specifications, such as power supply rejection and common mode rejection, amongst others, unimpeded when utilizing circuit arrangements to improve gain, speed, and noise operating under low operating currents, and low power supply voltage conditions, (8) perform to specifications over fabrication process variations, and under different power supply and temperature conditions, (9) be small so they cost less, and be able to integrate multiple building block channels on the same die for better channel-to-channel matching, (10) use standard CMOS fabrication for lower cost, availability, and proven quality.


As the demand for portable, green electronics, and energy harvesting (e.g., self-powered ICs) continues to grow, so does the requirement for circuits with lower power supplies and ultra low current consumptions. Low power electronics also require performance to specifications at low VDD and VSS where there is less available voltage headroom to process signals. Meeting low power consumption and signal-to-noise specifications at low VDD and VSS require ICs with the widest input-output dynamic range possible. Although special fabrication processes can provide special transistors that enable a circuit to operate with lower power supplies and lower current consumption, but such special manufacturing are generally expensive, and may create a high cost barrier for the full market potentials to be realized in a timely manner.


Also, low power (ultra low current and low power supply voltage combined) may be required in some medical or defense applications where it is not safe to frequently replace an implantable or embedded battery operated IC. This factor would require ultra low power so that the ICs draw extremely low current so that the battery life is significantly extended.


Another example is next generation energy harvesting electronics that is wireless and battery less. They can function perpetually without ever needing to be connected to power source and with no need of being recharged. There are sub-categories of energy harvesting ICs that can be designated as self-powered ICs. Energy harvesting is part of green electronics that rely on harvesting or scavenging energy from the environment such as solar, mechanical, thermal, or magnetic, to name a few. These kinds of energies help generate voltage potential that can for example be stored on a super capacitor, which can power ultra low power electronics for signal processing.


Additionally, some biometrics system on a chip may require small die size for applications requiring multi-channel current mirrors or amplifiers for conditioning multiple sensor's outputs simultaneously. For example, multiple sensors implanted in tooth dentures that simultaneously read levels of sugar, salt, acidity, temperature, and other non-vitals, which may require and at a minimum will benefit from small circuit size for better matching between each circuit channels in one die.


Furthermore, to target high volume markets and reduce the risk of low yields for long term production, manufacturing organizations generally have an unfavorable view of circuits that require special processes. Optimal yield and quality generally avoid circuits that require variations to a standard process, or complex circuits whose specifications may depend on multiple device or (manufacturing) process parameters. Generally circuits requiring non-standard fabrication, or complex circuits are harder to optimize for maximal production yields, or they may compromise rugged end-product manufacturing (quality) goals. It is also of note that complex circuits may hinder transient response, in start up and turn off phases, for example. This may be a risky trait, particularly in energy harvesting applications that may subject the IC to less predictable or disorderly power supply on and off patterns in the field.


Besides ultra low power, emerging applications such as energy harvesting that was noted above and bio-metrics, require small circuits to keep the costs down in order to realize their full and highest volume market potentials.


As stated earlier, current mirrors are fundamental building blocks in electronics. Some of performance specifications for a current mirrors are to have high ROUT and wide input-output voltage spans. Also, making current mirrors with simple circuits that are low power and low cost is beneficial to cost-performance-quality tradeoffs, including for high order other building blocks such as amplifiers, where current mirrors are utilized.


Utilizing Regulated Cascode (RGC) is a way to increase the Rout of a current mirror. FIG. 1G is an illustration of a prior art RGC current mirror (RGC-CM), which is not complex and yet it is low cost and rugged, but its' output-output voltage span is restrictive. As depicted in FIG. 1G, the drain terminal of M119G is the output of the RGC-CM, where V101G=VGSM119G+VGSM111G. As such, the minimum V103G≈VonM119G+VGSM111G which restricts the voltage span of the current mirror's output.


Operating at low or ultra low operating currents, generally causes an amplifier's gain to be lower. Utilizing RGC-CM is a way to increase the gain of amplifiers, in part, by way of increasing the amplifier's ROUT at the amplifier's high-gain node.


Increasing an amplifier's gain, utilizing RGC-CM, without restricting its input and output voltage span, are disclosed in the literature and in the references provided, including in [S. Yan and E. Sanchez-Sinencio, “Low voltage analog circuit design: A tutorial”, IEICE Trans. Analog Integrated Circuits and Systems, vol. E83A , no. 2, pp. 179-196, 2000], and [P. E. Allen & D. R. Holberg, CMOS Analog Circuit Design, 2nd Ed, Oxford University press, 2002]. Generally, the available RGC-CMs with a wide input-output voltage span have higher transistor count (i.e., larger die size), consume high power, have unfavorable transient response considering current consumption, or are complex (e.g., performance depends on different types of FETs, or performance depends on multiple device parameters) which are generally unfavorable for rugged manufacturing in terms of cost-performance-quality traits.


As noted earlier, low voltage electronics require rail-to-rail input-output operations. A folded cascode transconductance amplifier (FCTA), is a suitable amplifier topology for rail to rail operations. From a high level perspective, FCTA contains a common source amplifier (CSA) at its inputs. Then the differential current outputs of this CSA feed the differential current input of a common gate amplifier (CGA) whose differential output currents are fed onto a differential input to single output current mirror, that sum at the FCTA high impedance output node to make an output voltage (VOUT). The regulated cascode current mirror (RGC-CM) is generally used in the CGA and current mirror sections of an amplifier to improve their ROUT, which increase's the amplifier's gain. Moreover, its beneficial for the RGC-CM to have wide input-output voltage span in order not to restrict the amplifier's input-output voltage span, in which it is utilized. Operating the inputs of a FCTA rail to rail is generally accomplished by running pairs PMOSFET and NMOSFET (complementary) CSAs in parallel as inputs. Near the rails, either PMOS CSA runs out of headroom while the NMOS CSA takes control and keeps feeding the next CGA gain stage, or vice versa.


Given the wide common mode range in an amplifier is especially beneficial for near rail-to-rail input-output voltage spans, utilizing a floating current source (FCS) or emulating its equivalent function, is generally a way to reduce the errors generated in upper and lower current source in the CGA and current mirror (summing node) gain stage. However, in order to operate the amplifier with lowest VDD (besides operating the amplifier inputs-outputs rail-to-rail), all of the amplifier's elements, including the FCS need to operate with lowest VDD. FIG. 2D is an illustration of a prior art FCS, which is low power and fast, but its minimum VDD≥2VGS+VIDS would limit the amplifier for low power supply voltage applications.


Generally an amplifier's high gain stage is coupled with a buffer driver, which makes a buffer amplifier that would be able to drive external loads. In order to operate the buffer amplifier with ultra low currents and in low power supply environment, the buffer driver contained in it, must also consume ultra low currents and be capable of operating to specification with low power supplies. As noted earlier, operating in low or ultra low currents, slows down the speed. Hence, it would be advantageous to have a buffer driver that is fast, inherently, at low currents and can operate with low power supply voltage and have near rail-to-rail input-output voltage spans. Running fast and operating with low currents at low power supplies, still requires a buffer driver with high sink-source current drive capability to handle current requirements of different external loads. For example, some emerging portable applications use resistive sensors to measure environmental toxicity. The resistivity of such toxicity sensors can drop significantly when they are activated to make a measurement. Hence, it would be advantageous to have a buffer driver that can handle external low resistive loads when the sensor is activated, but return to low currents consumptions seamlessly, when the resistive sensor is no longer activated.


As mentioned earlier, next generation energy harvesting, and wireless and batteryless electronics are emerging applications, that require ultra low power ICs. All else equal, operating analog ICs at ultra low currents present additional challenges such as high noise besides low gain, and slow dynamic response (e.g., as noted earlier and generally, the lower the current, the higher the noise, the lower the gain, and the lower the speed). Hence, it would be advantageous to have a low noise amplifier noise that consumers low operating currents, and one that can be fast and operate with low power supply voltage.


Therefore, in summary here are a list of advantages of these teachings. One, is to make current mirrors that are simple, small, low cost, low power, have high ROUT, and wider input-output voltage span, whose embodiments are illustrated in FIG. 1B, FIG. 1D, and FIG. 1F. Two, as just stated, low current reduce an amplifier's gain, generally. Hence, another benefit of this teaching is to improve amplifiers gain and input-output voltage span, while keeping its power consumption low, by utilizing the disclosed RGC-CMs in amplifiers. The amplifier of FIG. 1A utilizes the RGC-CM coupled with DCSC (illustrated in FIG. 1B). The amplifier of FIG. 1C utilizes the RGC-CM coupled with ICMA (illustrated in FIG. 1D). The amplifier of FIG. 1E utilizes the RGC-CM coupled with composite amplifier, CSGA (illustrated in FIG. 1F). As a result of utilizing the disclosed RGC-CMs in the amplifiers, the amplifier's performance is improved, some aspects of which are illustrated in simulations FIG. 11 to FIG. 1M. Three, it would be advantageous to improve floating current sources (FCS), or emulate the function of FCS, that can operate at low power supply voltage. By utilizing the disclosed FCSs in amplifiers, the said amplifier can operate with low VDD while maintaining its accuracy. The embodiments of amplifiers utilizing first (FCS200A), second (FCS200B), and third (FCS200C) FCS are illustrated in FIG. 2A, FIG. 2B, and FIG. 2C, respectively. Four, another benefit is to lower the output noise of an amplifier by narrow banding the amplifier. To reinvigorate the speed of the narrow banded amplifier, a boost-on signal is initiated which dynamically and rapidly injects a substantial current into the amplifier's bias current to speed up its slew rate, when the amplifier's inputs get off balance due to a large transient differential input signal. Subsequently, after the amplifier has had some time to respond and regulate itself and as the amplifier's inputs approach balance, a boost-off signal dynamically injects a slow and decaying (to zero) current into amplifier's bias circuitry, instead of turning the boost current rapidly, in order to improve the amplifier's settling time. Additionally, the boost circuitry and the amplifier's time constants, that determine the dynamic response of the boost and that of the amplifier, are approximately matched in order to improve smooth transitions in and out of boost and to improve consistency of the amplifier's dynamic response across process, temperature and operation variations. The embodiments of noise reduction and speed boost circuits are illustrated in FIG. 3A, and FIG. 3B. Simulations depicted in FIG. 3C and FIG. 3D illustrate some aspects of the improvements in noise, and the dynamic response of an amplifier with and without the disclosed noise reduction and speed boost circuits. Five, another feature of this disclosure is to make a low voltage, low current, and high-speed buffer driver that is coupled with the high gain stage of an amplifier (FIG. 4A). The buffer driver utilizes CNICM to monitor and rectify the sink-source currents of FET that drive external loads, before the sink-source FET current signals are fed to MCS or LTA circuits. The output of MCS or LTA circuits is then fed onto a NICMA or ICMA or CMA which helps regulate the current in the inactive sink-source FET. The buffer driver can operate with low power supply voltage and consume low currents. Because the elements utilized in the buffer driver run chiefly in current mode, it is inherently fast. Also, because the elements utilized in the buffer driver run chiefly in current mode, voltage swings are inherently small, which helps the buffer driver operate with lower power supplies. The embodiments of buffer driver circuits are illustrated in FIG. 4B, FIG. 4C, and FIG. 4D. Simulations depicted in FIG. 4E and FIG. 4F and FIG. 4G illustrate some aspects of the improvements in current drive capability of amplifiers (that utilize the disclosed buffer drivers) while running at very low operating currents.


Note that the following papers providing additional analysis of relevance to low power and low cost amplifier designs are also hereby incorporated by reference in their entirety: (1) A. Far, “Small size class AB amplifier for energy harvesting with ultra low power, high gain, and high CMRR,” 2016 IEEE International Autumn Meeting on Power, Electronics and Computing (ROPEC), Ixtapa, Zihuatanejo, Mexico, 2016, pp. 1-5; (2) A. Far, “Amplifier for energy harvesting: Low voltage, ultra low current, rail-to-rail input-output, high speed,” 2016 IEEE International Autumn Meeting on Power, Electronics and Computing (ROPEC), Ixtapa, Zihuatanejo, Mexico, 2016, pp. 1-6; (3) A. Far, “Low Noise Rail-To-Rail Amplifier for Energy Harvesting Runs Fast at Ultra Low Currents,” 2017 IEEE Canadian Conference on Electrical and Computer Engineering (CCECE), Windsor, ON, 2017; and (4) A. Far, “Ultra Low Current and Low Voltage Class AB Buffer Amplifier,” 2017 IEEE Canadian Conference on Electrical and Computer Engineering (CCECE), Windsor, ON, 2017.


SECTION (I): Detailed Description of Regulated Cascode Current Mirror (RGC-CM) Coupled with Diode Connected Self Cascode (DCSC), as Illustrated in FIG. 1B



FIG. 1B is a circuit schematic showing a RGC-CM100B coupled with DCSC100B according to an embodiment. One differentiation of this teaching, compared to the conventional teachings, is in arranging a RGC-CM100B coupled with DCSC100B to generate a DC voltage shift thereby allowing the input-output voltage of the current mirror to get closer to the rail.


The basic idea of RGC-CM100B coupled with DCSC100B in FIG. 1B, is to use negative feedback of an auxiliary amplifier (AAUX) that has a gain of ‘G’ (e.g., formed by M111B and I105B), which regulates the VDS of M117B as constant as possible, irrespective of the variations in the output node 102B of the current mirror, which is VD of M119B. Here, the DCSC shifts down the VDS of M117B down by VDSM113B which helps V102B (output of current mirror) get closer to VSS.


Throughout the description of FIG. 1B, ‘X’ is the W/L of a MOSFET, and ‘r’, ‘s’, ‘t’, ‘u’, ‘v’, and ‘w’ are scale factors for MOSTEL W/Ls. These MOSFET scale factors can be set approximately in the ranges of 0.01≤r≤100, 0.01≤s≤100, 0.01≤t≤100, 0.01≤u≤100, 0.01≤v≤100, and 0.01≤w≤100 depending on considerations such as current consumption, voltage span, and die size, amongst others. Also, current source (e.g., I111B, I105B, and I107B) scale factors can be set approximately in the ranges of 0.01≤o≤100, 0.01≤p≤100, 0.01≤q≤100. For example, for a embodiment of FIG. 1A that utilizes plurality of RGC-CM100B coupled with DCSC100B of FIG. 1B, the MOSFET's W=4 μm, and X=1 μm computes to W/L=4. In this example, the MOSFET W/L scale factors are set to r=10, v=3, and u=t=s=t=w=1. Also, here the current sources, i=5 nA for IM111B=IM105B=IM107B=5 nA, where o=p=q=1.


The connections of the elements in FIG. 1B are described as follows. The body terminal of all NMOSFETs in FIG. 1B are connected to node 2 which is VSS, and the body terminals of PMOSFETs are connected to node 1 which is VDD. Bias current sources I111B, I105B, and I107B have their upper terminals connected to node 1, which is VDD. The source terminal of NMOSFETs M126B, M111B, and M117B are connected to node 2 that is VSS. The lower terminal of I111B is connected to node 100B, which is connected to the gate terminal of M126B, and the drain terminal of M126B, and the gate terminal of M117B. Node 101B is connected to the lower terminal of I105B, the drain terminal of M111B, and the gate terminal of M119B. Node 102B is connected to the drain terminal of M119B, which is the output of the current mirror (or terminal Io100B). Node 103B is connected to the gate terminal of M111B, the source terminal of M115B, and the drain terminal of M113B. Node 104B is connected to the lower terminal of I107B, the gate terminal of M115B, the drain terminal of M115B, and the gate terminal of M113B. Node 105B is connected to the source terminal of M113B, the drain terminal of M117B, and the source terminal of M119B. The IM111B terminal is the input current to the RGC-CM100B coupled with DCSC100B, and IM119B terminal is its output.


Describing the details of the circuit in FIG. 1B is as follows. The I111B flows through M126B, whose VGS is the same as that of M117B, which mirrors and scales v×IM126B=u×IM117B. For descriptive clarity of FIG. 1B, let's assume current source scale factors of o=q=1, and the MOSFET scale factors of r=10, s=t=u=w=1, and v=2. As such, I107B≈1i (which flows through M115B and M113B) is fed into node 105B where IDM117B≈2i, and hence the current mirror's output which is IDM119B≈IDM117B−I107B≈1i.


The current mirror of FIG. 1B utilizes DCSC which is composed of two FETs, M115B and M113B, whose function is briefly discussed. The basic function of DCSC here is to provide a DC voltage shift, which is a function of the difference (ΔVGS)=VGSM113B−VGSM115B≈VDSM113B≈VT×ln(r/s). In the subthreshold region, this DC voltage shift is chiefly dependent on VT which is well controlled and it is approximately independent of CMOSFET's VTH (note that the value of VTH has normal but wide fluctuations in fabrication manufacturing).


In FIG. 1B, the AAUX which is configured as a simple common source amplifier (CSA) which can be composed of only one FET and one current source, M111B-I105B, where I105B can be supplied via a one FET of a PMOS current mirror. This AAUX, has an approximate gain of G∝VA/VT, and it regulates the gate voltage of M119A in order to increase the ROUT at node 102B, which is the output of the RGC-CM. The ROUT is approximately ∝gm2×rds3∝(VA/VT)2×(VA/ID). Note that ROUT of RGC-CM on actual silicon will likely be lower due to second order effects such as η (subthreshold slope factor) and substrate leakage, amongst other factors. Moreover, it is of note that for additional headroom at RGC-CM's input-output, multiple DCSCs can be cascaded to developed a larger DC voltage shift, for example.


In summary, the prior art illustrated in FIG. 1G, the minimum V103G=VonM119G+VGSM111G. Comparatively, the embodiment of FIG. 1B utilizes a simple circuit that improves the minimum V102B≈VonM119B+VGSM111B−VT×ln(r/s). Also note that VDSM117B=VGSM103B−VDSM113B, and VGSM115B−VGSM113B+VDSM117B≈VDSM111B≈VT×ln(r/s).


In conclusion, the benefits of the FIG. 1B, RGC-CM100B coupled with DCSC100B include the following. First, the same type FETs may be used in each of the AAUX (e.g., a M111B which is NMOSFETs), the DCSC100B (e.g., M115B-M113B which are also NMOSFETs), and the cascode FETs in the current mirror (e.g., M117B-M119B) whose device parameters match and track, and hence benefiting the amplifier's DC, AC, and transient specifications stability over fabrication process, temperature, and power supply variations. Second, the bias voltages in the RGC-CM100B coupled with DCSC100B are self biased (i.e., no need for separate voltages to set its internal bias nodes), which lower complexity and lowers current consumption, reduces die size and cost, and improved quality. Third, the RGC-CM100B coupled with DCSC100B has low transistor count (i.e., it is small) and low current because its' auxiliary amplifier's gain is achieved by only two FETs and the widening of output voltage headroom is achieved by only two FETs (e.g., M115B and I113B) in DCSC. Fourth, the RGC-CM100B coupled with DCSC100B's headroom is widened by VT×ln(r/s) with 2 FETs, which can practically be programmed or pre-determined to have a range of, for example, 50 mV to 150 mV. For sub-1V power supply environments, every 100 mV extra headroom in the current mirror translates to 10% extra operating room at the current mirror output. Fifth, as stated earlier, electronic functions used in energy harvesting applications may require more smooth and stable power up and down transient responses (e.g., for power save reasons or for frequent switching from a magnetic power harvester source to a kinetic power harvester source). The simple and small circuit arrangement, in RGC-CM100B coupled with DCSC100B that utilizes few FETs (i.e., 2 FETs to make a single stage AAUX to increase Rout), is beneficial for AC and transient response of the current mirror. This is because the single stage amplifier neither require multiple AC loops nor complicated AC and transient compensation. For energy harvesting applications, that may subject ICs to less predictable or disorderly power supply on and off patterns, this simple arrangement in RGC-CM coupled with DCSC can provide additional guard band for smoother dynamic performance.


Section (II): Detailed Description of Regulated Cascode Current Mirror (RGC-CM) Coupled with Inverting Current Mirror Amplifier (ICMA), as Illustrated in FIG. 1D



FIG. 1D is a circuit schematic showing a RGC-CM100D coupled with ICMA100D, according to another embodiment. A differentiation of this teaching, compared to conventional teachings, is in arranging a RGC-CM100D coupled with ICMA100D, where ICMA100D has a current input terminal whose operating voltage can be programmed with flexibility. As a result, this aspect of ICMA100D can widen the input-output voltage range of the current mirror. Here, the ICMA100D serves the function of auxiliary amplifier, AAUX, that increase its ROUT. The benefits of this arrangement include the widening of voltage output span, and increasing the ROUT of the RGC-CM100D coupled with ICMA100D can provide low power and high speed with low cost. Throughout the description of FIG. 1D, ‘X’ is the W/L of a MOSFET, and ‘n’, ‘r’, ‘s’, ‘t’, ‘u’, ‘v’, and ‘w’ are scale factors for MOSTEL W/Ls. These MOSFET scale factors can be set approximately in the ranges of 0.01≤n≤100, 0.01≤r≤100, 0.01≤s≤100, 0.01≤t≤100, 0.01≤u≤100, 0.01≤v≤100, 0.01≤w≤100 depending on considerations such as current consumption, voltage span, and die size, amongst others. Also, for current source (e.g., I113D, I105D, I107D, and I109D) scale factors can be set approximately in the ranges of 0.01≤o≤100, 0.01≤p≤100, 0.01≤q≤100, and 0.01≤m≤100. For example, for an embodiment of FIG. 1C that utilizes RGC-CM100D coupled with ICMA100D of FIG. 1D, for MOSFET's W=4 μm, and X=1 μm, which computes to W/L=4. In this example, the MOSFET W/L scale factors are set to n=0.05, v=3, and r=s=t=u=w=1. Also, current source scale factors can be set to q=2, and o=p=m=1 for i=5 nA such that IM111D≈IM113D≈IM115D≈IM121D≈IM119D≈½×IM117D≈IM128D≈5 nA.


The connections of the elements in FIG. 1D are described as follows. The body terminal of all NMOSFETs in FIG. 1D are connected to VSS, and the body terminals of PMOSFETs are connected to node 1 that is VDD. Bias current sources I113D, I105D, I107D, and IM109D have their upper terminals connected to node 1 that is VDD. The body terminal of all NMOSFETs in FIG. 1B are connected to node 2 that is VSS, and the body terminals of PMOSFETs are connected to node 1 that is VDD. The source terminal of NMOSFETs M128D, M111D, M113D, M117D, and M121D are connected to node 1 that is VSS. The lower terminal of I113D is connected to node 100D, which is connected to the gate terminal of M128D, and the drain terminal of M128D, and the gate terminal of M117D. Node 101D is connected to the lower terminal of I105D, the drain terminal of M111D, and the gate terminal of M119D. Node 102D is connected to the drain terminal of M115D, the drain terminal of M113D, the gate terminal of M113D, and the lower terminal of I107D. Node 103D is connected to the drain terminal of M119D, which is the output of the current mirror (or terminal Io100D). Node 104D is connected to the source terminal of M115D, the drain terminal of M117D, and the source terminal of M119D. Node 105D are connected to the gate terminal of M121D, the drain terminal of M121D, and the lower terminal of I109D. The IM113D is the input current to the RGC-CM100D coupled with ICMA100D, and IDM119D is its output current that is its mirror.


Note that the ICMA100D function is performed by M115D, M113D, I107D, M111D, and I105D. For clarity in describing the RGC-CM100D coupled with ICMA100D of FIG. 1D, it is assumed that n=0.05, v=2, and r=s=t=u=w=1. Also, it is assumed that the current source scale factors to q=2, and o=p=m=1 for i=5 nA.


During steady state conditions, I113D flows through M128D, whose VGS is the same as that of M117D, which causes v×IM128D=u×IM117D or 2IM128D=IM117D. As such, IDM111D≈IDM113D≈IDM115D≈IDM121D≈IDM119D≈½×IDM117D≈IDM128D≈i≈5 nA.


As noted earlier, the regulated cascode (RGC) is used in a current mirror (CM), chiefly, to increase its ROUT. When, there are voltage changes (ΔvX) at the output (node 103D), the goal is for the output current (IDM119D=i) variations (ΔiX) to be minimized. The ΔvX here can cause a ΔiX on M119D, because M119D has a finite impeadance between its source and drain terminals. Since IM117D=2I128D=2i is fixed, then at node 104A, the ΔiX would flow through M115D, which is the current input terminal of ICMA100D. Given that I107D=2i is also constant, as explained earlier, this change in current, ΔiX, flowing through M115D gets subtracted from the steady state current of IDM113D. The auxiliary amplification, AAUX, function inside the ICMA100D is performed by M111D and I105D whose ROUT∝VA/ID (when CMOS is operating in the subthreshold region). Consequently, the ΔiX in IDM113D is mirrored onto M111D whose drain (at node 101D is coupled to I105D=i that is a constant current source) responds to this ΔiX, with negative gained voltage change of about ΔiX×ROUT. At node 101D, this negative gained voltage change is fed-back to the gate terminal of M119D. This negative gain voltage feedback, is the mechanism that regulates the VGS of M119D so that ΔiX→near zero.


In summary, the ICMA100D regulates the gate voltage of M119D in order to increase the ROUT of at node 103D. The ROUT of RGC-CM100D coupled with ICMA100D is approximately ∝gm2×rds3∝(VA/VT)2×(VA/ID), assuming subthreshold operations and assuming that the gain from node 104D to 102D is about unity with equal currents flowing through M115D, M113D, and M111D.


The RGC-CM100D coupled with ICMA100D output voltage span is also improved substantially, mainly because of the flexibility in setting V105D fairly independently (e.g., V105D∝I109D, and W/L of M121D) which bias the VGM115D. The minimum V103D=VonM119D+VonM117D, and VonM117D=VGSM121D−VGSM115D=ΔVGS≈VT×ln(r/n). Hence, minimum V103D=VonM119E+VT×ln(r/n) above VSS, and approximately independent of CMOSFET's VTH.


In conclusion, the benefits of the proposed RGC-CM100D coupled with ICMA100D illustrated in of FIG. 1D include the following. First it contains the function of the auxiliary amplifier that is composed of the same type FETs (e.g., a M115D, M113D, and M111D that are all NMOS) as that of the FETs in the cascoded current mirror (e.g., M117D, M119D that also NMOS). Note that constant current sources ID105D and ID107D can be made of single PMOSFETs to serve only the function of constant ‘bias current’ source. Therefore, the DC, AC, and transient specifications stability and manufacturability of RGC-CM can be improved, because utilizing the device parameters of the same type of FETs (e.g., NMOSFETs) better match and track each other over normal variations in fabrication process, temperature, and power supply. Second, RGC-CM100D coupled with ICMA100D has low transistor count (i.e., it is small) and low current because the expanded output span as well as the auxiliary amplification function to increase ROUT can be accomplished by 4 FETs (M111D, M113D, M115D, M121D) and 3 current sources (I105D, I107D, and I109D which can be made with 3 FETs as well). Third, it can operate fast in part because ICMA operate in current mode. Moreover, ICMA is fast because small geometry FETs can be selected in the ICMA signal path (M111D, M113D, and M115D) with a scale factor of 1 (i.e., they are not composed of plurality of FETs arranged in parallel). Fourth, the headroom is widened substantially allowing the RGC-CM100D coupled with ICMA100D's output voltage to swing within VonM119E+VT×ln(r/n) above the power supply rail. As noted earlier, for sub-1V power supply environments, every 100 mV extra headroom in the current mirror translates to 10% extra operating room at the current mirror output. Fifth, as stated earlier, electronic functions used in energy harvesting applications may subject ICs to less predictable or disorderly power supply on and off patterns, thereby requiring more smooth transient responses. This ICMA100D operates chiefly in current mode which is inherently fast, and its AAUX has a single pole (simplifying its compensation) that can provides for additional transient performance guard band.


Section (III): Detailed Description of Regulated Cascode Current Mirror (RGC-CM) Coupled with Composite Amplifier (CSGA), Illustrated in FIG. 1F



FIG. 1F is a circuit schematic showing a RGC-CM100F coupled with CSGA100F according to a embodiment. As explained in previous sections, increasing the ROUT of a current mirror is one of the motivation for utilizing an auxiliary amplifier (AAUX). One of the differentiation of the RGC-CM100F coupled with CSGA100F in FIG. 1F is in the arrangement of CSGA100F that performs the function of AAUX to increase RO as well as input-output voltage span of the current mirror by utilizing a few FETs. Here, the CSGA100F utilizes a first differential common source amplifier (CSA), with a built-in offset, that feeds its output current into first a common gate amplifier (CGA) that is inherently fast and whose gate is biased from the common source summing junction of the first CSA. The designation of ‘composite amplifier’ is because this CSGA100F intertwines the composite of PMOSFETs and NMOSFETs to generate both the gain and widen the input-out voltage span of the current mirror. The gain ‘G” (attained chiefly by M113F and I109F) of CSGA100F keeps the VDS of M117F as constant as possible, ideally speaking, irrespective of the variations in VD of M119F. Here, the input-output voltage span of the current mirror is widened by lowering the VDSM117F. This voltage is set by the built-in input offset voltage (ΔVGS≈VDSM117F≈VT×ln(r/s)) generated by scaling M115F-M111F that are the differential inputs the first CSA within the CSGA100F. Alternatively, (to provide the CSA with a DC voltage shift), the M111F input of CSA can be biased at an alternative voltage instead of ground or VSS, which avoids scaling M115F-M111F and saves die area.


Throughout the description of FIG. 1F, ‘X’ is the W/L of a MOSFET, and ‘r’, ‘s’, ‘t’, ‘u’, ‘v’, and ‘w’ are scale factors for MOSTEL W/Ls. These MOSFET scale factors can be set approximately in the ranges of 0.01≤r≤100, 0.01≤s≤100, 0.01≤t≤100, 0.01≤u≤100, 0.01≤v≤100, and 0.01≤w≤100 depending on considerations such as current consumption, voltage span, and die size, amongst others. Also, current sources (e.g., I111F, I107F, I109F, and I119F) scale factors can be set approximately in the ranges of 0.01≤n≤100, 0.01≤o≤100, 0.01≤p≤100, 0.01≤q≤100. For example, for an embodiment of FIG. 1E that utilizes RGC-CM coupled with composite amplifier of FIG. 1F, the MOSFET's W=4 μm, and X=1 μm computes to W/L=4. In this example, the MOSFET W/L scale factors are set to s=0.0625, v=2, and u=t=r=t=w=1. Also, here the current sources, i=5 nA for 2×IM119F=2×IM109F=IM111F=IM107F=10 nA, where q=o=1, n=p=2.


The connections of the elements in FIG. 1F are described as follows. The body terminal of all NMOSFETs in FIG. 1F are connected to node 2 which is VSS, and the body terminals of PMOSFETs are connected to node 1 which is VDD. Bias current sources I119F, I107F, and I109F have their upper terminals connected to node 1, which is VDD. The lower terminal of I111F is connected to node 2, which is VSS. The drain terminal of M111F, and the gate terminal of M111F are connected to node 2, which is VSS. The source terminal of NMOSFETs M131F and M117F are connected to node 2 that is VSS. Node 100F is connected to drain terminal of M131F, the gate terminal of M131F, and the gate terminal of M117F. Node 101F is connected to source terminal of M111F, the source terminal of M115F, the gate terminal of M113F, and the lower terminal of current source I107F. Node 101F is the summing junction of CSA containing the CSA differential input FETs (M111F, M115F) and the current source I107F that provides the tail current to the CSA differential input FETs (M111F, M115F). Node 102F is connected to drain terminal of M113F, and the gate terminal of M119F, and the lower terminal of I109F. Node 103F is connected to source terminal of M113F, the drain terminal of M115F, and the upper terminal of I111F. Node 104F is connected to source terminal of M119F, the drain terminal of M117F, and the gate terminal of M115F. Node 105F is connected to the drain terminal of M119F and it is the output terminal, IO100F, of the RGC-CM.


The details of various functions of the circuit in FIG. 1F, which is RGC-CM100F coupled with CSGA100F is now described. The I119F flows through M131F, whose VGS is the same as that of M117F, which mirrors and scales v×IDM131F=u×IDM117F. For descriptive clarity, it is assumed that current source scale factors of q=o=1, n=p=2, and the MOSFET scale factors of r=0.0625, s=t=u=w=1, and v=1. As such, IDM109F≈1i flows through MM113F, which subtracts from I111F≈2i thus leaving IDM115F≈1i. Consequently, IDM115F≈1i subtracts from I107F≈2i which leaves IDM111F≈1i.


The CSGA100F of FIG. 1F serves at least two functions, which is first to provide a build-in offset (ΔVGS) that widens the input-output voltage span of the current mirror. The ΔVGS≈VGSM115F−VGSM111F=VDSM117F≈VT×ln(r/s). Note that the build-in offset is chiefly dependent on VT (generally, constant value at ˜26 mV by laws of device physics) and approximately independent of CMOSFET's VTH (note that the value of VTH has normal but wide fluctuations in fabrication manufacturing). The second function of the CSGA100F is to serve as the AAUX as follows. A change in voltage at node 105F which is the output of the current mirror, ΔvOUT, generate a differential current, ΔiOUT, at node 104F. Since IDM117F is constant, then ΔiOUT causes a voltage change at node 104F, Δvx. This Δvx is effectively applied (to a CSA with its inputs as M115F-M111F) between the nodes 104F (VDSM117F) and VSS, setting aside the built in offset voltage which is also a constant voltage≈VT×ln(r/s). With ΔvX applied to CSA, an inverse current output change, −ΔiX is generated that is fed into a CGA (containing M113F and IM109F) that is inherently fast. The −ΔiX applied to the output of this CGA (at its high impedance node) generates a negative gained voltage change at node 102F that regulates the VGM119F. It can be noted in FIG. 1F that while ΔVGS can be generated by scaling PMOSFETs, but ΔVGS is independent of PMOSFET device parameters and it is approximate a function of VT (thermal voltage).


As indicated earlier, it is possible to connect the gate terminal of M111F (the other input of the CSA) to a secondary bias voltage instead of VSS or GND that established the VGM111F (e.g., VGM111F+VGSM111F−VGSM115F≈VDS117F VDS), which would then allow to make r=s=1. This arrangement would save area since M115F or M111F needs not be scaled, and the secondary bias voltage would facilitate the headroom needed at node 104F


Node 101F which is the summing junction of CSA (source terminals of M111F and M115F) establishes a DC voltage that biases VGM113F which is the gate terminal of the CGA. The CGAs are inherently fast and CSA output (drain terminal of M115F) is fed into the current input of CGA (source terminal of M113F). Therefore, the manner of arranging the RGC-CM100F coupled with CSGA100F in FIG. 1F serves the function of auxiliary amplifier and can speed up its dynamic response.


In summary, voltage movements, ΔvOUT, at node 105F (output terminal of the current mirror) cause a voltage, ΔvX, at node 104F. The Δvx would cause a negative change in current, −ΔiX, in node 103F, which in turn cause a gained voltage change at the output of the CSGA100F, ≈−ΔiX×−ROUT aux≈−G×ΔvX at node 102F, which is the gate of M119F. As a result, the source voltage of M119F follows its gate voltage, thereby regulating the voltage at node 104F until ΔvX→near zero again and as such the ROUT of the RGC-CM can be increased in this arrangement.


The gain, ‘G’, of the RGC-CM amplifier is approximately G∝VA/VT, since FETs operate in the subthreshold region. The Rout of RGC-CM is approximately ∝gm2×rds3∝(VA/VT)2×(VA/ID). As stated earlier, the ROUT of RGC-CM on actual silicon will be lower due to second order effects such as η (subthreshold slope factor) and substrate leakage, amongst other factors.


In summary, the benefits of the FIG. 1F RGC-CM100F coupled with CSGA100F include the following. First, same channel type FETs are utilized in the current mirrors (M117F, M119F) and CGA (containing M113F), whose device parameters match and track, and hence benefiting the amplifier's DC, AC, and transient specifications stability over fabrication process, temperature, and power supply variations. The generated built-in offset is made using PMOSFETs where the offset voltage is chiefly a function VT, and it is independent of PMOSFET device parameters (because they operate in the subthreshold region). Second, the bias voltages in the RGC-CM100F coupled with CSGA100F are self biased (i.e., no need for separate voltages to set its internal bias nodes), which lowers complexity and current consumption. Third, it has low transistor count (i.e., it is small) and low current because its' auxiliary amplifier's gain function and the widening of input-output voltage span is achieved by 3 FETS (M111F, M113F, and M115F) and 3 current sources (I103F, I107F, and I109F) where each current source can be made of a single FET. Fourth, headroom to the rails can be set (e.g., VT×ln(r/s) above VSS) independent of the auxiliary amplifier, and practically be programmed or pre-determined to have a range of, for example, 50 mV to 150 mV above VSS. Therefore, the input-output voltage span of the current mirror is widened. As noted earlier, for sub-1V power supply environments, every 100 mV extra headroom in the current mirror translates to 10% extra operating room at the current mirror output.


Section (IV): Detailed Description of Amplifier (AMP100A) is Illustrated in FIG. 1A, Utilizing Plurality of RGC-CM100B Coupled with DCSC100B as Illustrated in FIG. 1B



FIG. 1A is a circuit schematic showing a embodiment of a folded cascode transconductance amplifier (FCTA) that utilizes the plurality of RGC-CM100B coupled with DCSC100B illustrated in FIG. 1B. When the RGC-CM100B coupled with DCSC100B is utilized in the FCTA, it expands the amplifier's input-output voltage span and increases its output impedance (ROUT) which in turn increases the amplifier's gain (AV). Beside higher gain and wider input-output voltage span, the other improvements associated with RGC-CM100B coupled with DCSC100B, are carried over to the amplifier which improve he amplifier's gain, enables its input and output to operate rail-to-rail, lower its cost, lower its power, and increase its speed.


Note that alternative amplifier embodiments are possible such as an amplifier with NMOS input stage or complementary (PMOS and NMOS) rail-to-rail input stages, other amplifier topologies that are not FCTA, amongst others. The connections of the elements of AMP100A of FIG. 1A are described as follows. The body terminal of all NMOSFETs in FIG. 1A are connected to node 2 that is the VSS, and the body terminals of PMOSFETs are connected to node 1 that is the VDD. The upper terminals of the bias current sources I108A, I109A, I110A, I111A, I101A, I103A, I105A, and IM107A are connected to node 1 that is VDD. Bias current sources I100A, I102A, I104A, and I106A have their lower terminals connected to node 2 that is VSS. The source terminal of PMOSFETs M100A, M106A, M110A, and M116A are connected to node 1 which is the positive supply voltage, VDD. The source terminal of NMOSFETs M122A, M126A, M101A, M107A, M111A, and M117A are connected to the negative supply voltage, VSS. Node 100A is connected to the gate terminal of M120A. Node 101A is connected to the gate terminal of M121A. Node 100A is the VIN+ terminal of the amplifier and node 101A is the VIN− terminal of the amplifier. Node 102 is connected to the source terminal of M120A, the source terminal of M121A, and the lower terminal of I108A. Node 103A is connected to the drain terminal of M122A, the gate terminal of M125A, and the lower terminal of I109A. Node 104A is connected to the drain terminal of M123A, the gate terminal of M123A, the gate terminal of M124A, and the lower terminal of I110A. Node 105A is connected the drain terminal of M124A, the source terminal of M123A, and the gate terminal of M122A. Node 106A is connected to the drain terminal of M126A, the source terminal of M125A, and the source terminal of M124A. Node 107A is connected to the drain terminal of M125A, the lower terminal of I111A, the gate terminal of M126A, the gate terminal of M107A, and the gate terminal of M117A. Node 108A is connected to the drain terminal of M107A, the source terminal of M109A, and the source terminal of M103A as well as the drain terminal of M120A. Node 109A is connected to the drain terminal of M117A, the source terminal of M119A, and the source terminal of M113A as well as the drain terminal of M121A. Node 110A is connected to the drain terminal of M101A, the gate terminal of M109A, and the lower terminal of I101A. Node 111A is connected to the drain terminal of M105A, the gate terminal of M105A, the gate terminal of M103A, and the lower terminal of I103A. Node 112A is connected the drain terminal of M103A, the source terminal of M105A, and the gate terminal of M101A. Node 113A is connected to the drain terminal of M108A, the gate terminal of M106A, the gate terminal of M116A, and the drain terminal of M109A. Node 114A is connected to the drain terminal of M111A, the gate terminal of M119A, and the lower terminal of I105A. Node 115A is connected to the drain terminal of M115A, the gate terminal of M115A, the gate terminal of M113A, and the lower terminal of I107A. Node 116A is connected the drain terminal of M113A, the source terminal of M115A, and the gate terminal of M111A. Node 117A is connected the drain terminal of M102A, the source terminal of M104A, and the gate terminal of M100A. Node 118A is connected to the drain terminal of M100A, the gate terminal of M108A, and the upper terminal of I100A. Node 119A is connected to the drain terminal of M106A, the source terminal of M108A, and the source terminal of M102A. Node 120A is connected to the drain terminal of M104A, the gate terminal of M104A, the gate terminal of M102A, and the upper terminal of I102A. Node 121A is connected to the drain terminal of M116A, the source terminal of M118A, and the source terminal of M112A. Node 122A is connected to the drain terminal of M110A, the gate terminal of M118A, and the upper terminal of I104A. Node 123A is connected to the drain terminal of M114A, the gate terminal of M114A, the gate terminal of M114A, and the upper terminal of I106A. Node 124A is connected the drain terminal of M112A, the source terminal of M114A, and the gate terminal of M110A. The high impedance (high gain) output of AMP100A, is VOUT, which is node 125A that is connected to the drain terminal of M118A, and the drain terminal of M119A.


Note that five RGC-CM100B coupled with DCSC100B are utilized in the amplifier, AMP100A, embodiment of FIG. 1A. The NMOS ones are RGC101A containing M122A-M123A-M124A-M125A-M126A-I109A-I110A; RGC102A containing M101A-M103A-M105A-M107A-M109A-I101A-I103A; and RGC103A containing M111A-M113A-M115A-M117A-M119A-I105A-I107A. The PMOS ones are RGC104A containing M100A-M102A-M104A-M106A-M108A-I100A-I102A; and RGC105A containing M110A-M112A-M114A-M116A-M118A-I104A-I106A. It would be possible that AMP100A could function properly without RGC101A and RGC104A, which can be substituted with diode connected cascodes for current mirrors. The RGC101A and RGC104A are added in the embodiment of FIG. 1A, in part, for better systematic matching considerations and improved AC performance. Note that, in order to avoid repeating how the amplifier is benefited by each of RGC101A to RGC105A, the explanation of one RGC-CM (e.g., RGC103A) is deemed sufficient, regarding the function of RGC-CM (coupled with DCSC) in the main amplifier, FCTA.


For explanation regarding RGC-CM100B coupled with DCSC100B refer to its detailed description, but below is a brief description of how they operate and benefit a FCTA. The FCTA has generally 3 parts, ‘common source amplifier’ (CSA), ‘common gate amplifier’ (CGA), and a current mirror (CM). The VIN is applied to a differential CSA, containing M120A and M121A, whose output feed the CGA, containing M109A and M119A. The differential outputs of this CGA feed the CM, made up of M106A and M116A, to make a single ended output, VOUT which is also the high impedance node (125A) of FCTA. The current I111A=i controls VGSM126A that establishes IM126A, which is mirrored and scaled onto IM107A=IM117A=3i. The CSA's input currents IM120A≈M121A≈i are fed into the source terminals of M109A and M119A, respectively, which are the differential inputs of the CGAs. The I103A=i and I105A=i flow through the DCSCs of RGC102A and RGC103A, respectively, which are passed onto M107A and M117A, and in that order. Therefore, the operating currents of CGA's (containing M109A and M119A) are IM109A≈IM107A−I103A−IM120A≈3i−2i=i and IM119A≈IM117A−I105A−IM121A≈3i−2i=i. As explained in the RGC-CM coupled with DCSC in previous section regarding RGC103A, the auxiliary amplifier (composed of M111A-I105A) regulates the VGM119A which increase the output impedance of the RGC-CM. This same discussion is applicable to RGC102A, RGC104A, and RGC105A. As a result, effectively, the output impedance and gain of CGAs and current mirrors in FCTA is increased. Note that main amplifier's ROUT∝ro×(ro×gm)2 and Av is ∝ROUT×gmi. Hence, Av∝(ro×gm)3, in the subthreshold region of operations for CMOS. To save current consumption, note that Av can be increased by raising only I108A and I111A, instead of increasing the current consumption of the whole amplifier, given that Av is ∝ROUT×gmi, and gmi is the input stage transconductance that is about ∝I108A/VT


Moreover, the DC common mode range of the main amplifier inputs (M120A and M121A) is expanded because, with the DC voltage shift generated by the DCSCs in RGC102A and RGC103A, the V108A and V109A can get closer to the rails. The DC voltage shift generated in DCSC of RGC103A is approximately, VGSM113A−VGSM115A=ΔVGS≈VDSM113A≈VT×ln(r/s). Hence, V109A≈VGSM111A−VT×ln(r) from the negative rail.


Additionally, the same type of FETs (e.g., NMOSFETs) are utilized in the CGAs function, the DC voltage shift function that expands the VOUT span, and the auxiliary amplifier function that increases ROUT of the RGC-CM. Utilizing the same type of FETs here, improves the consistency of DC, AC, and transient performance over temperature, power supply, and process variations. FIG. 1H illustrates a worst case (WC) AC simulation of FIG. 1A that Av˜105 dB, PM˜45° is achievable. FIG. 1I upper graph (i) is the WC transient simulation of FIG. 1A indicating that SR˜2V/μS and ts˜1 μS is achievable. Moreover, FIG. 1I lower graph (ii) indicates current consumption of ˜180 nano Ampere is achievable for FIG. 1A circuit.


In summary, besides increasing an amplifier's gain, the benefits of utilizing plurality of RGC-CM100B coupled with DCSC100B in an amplifier are as follows. First, the same type (e.g., NMOSFET) FETs are used in each of the main amplifier's CGA, and RGC-CM's auxiliary amplifiers plus DCSCs. Given that same type (e.g., NMOSFET) FET's device parameters match and track each other, therefore the FCTA's consistency of DC and AC specifications and stability is improved over fabrication process, temperature, and power supply variations. Second, each of the amplifier's RGC101A to RGC105A are self biased (i.e., no need for separate voltages to set its internal bias nodes) saving current and die space. Third, the listed benefits of RGC-CM100B coupled with DCSC100B (see previous section) carry over to the amplifier, including small size, low current, and faster dynamic response. Fourth, the DC voltage shift provided in the RGC-CM100B coupled with DCSC100B widen the amplifier's input-output span closer to the rails. As stated before, for example, a 75 mV head-room expansion at the input and output of each of the upper PMOS and the lower NMOS based RGC-CM100B coupled with DCSC100B, can expand the voltage span at the input of the amplifier as well as at the high gain (impedance) node of FCTA by 150 mV or 15%, which is beneficial especially in the sub-1V power supply environment.


Section (V): Detailed Description of Amplifier (AMP100C) in FIG. 1C Utilizing Plurality of RGC-CM100D Coupled with ICMA100D, as illustrated in FIG. 1D



FIG. 1C is a circuit schematic showing a embodiment of an folded cascode transconductance amplifier (FCTA) that utilizes plurality of RGC-CM100D coupled with ICMA100D, illustrated in FIG. 1D. When RGC-CM100D coupled with ICMA100D is utilized in an amplifier, it can expand its input and output voltage span and increase its output impedance, ROUT, which in turn increases the amplifier's gain, Av. Also, the benefits of RGC-CM100D coupled with ICMA100D are carried over to improve the amplifier with low power and higher speed. Note that alternative amplifier embodiments are possible such as an amplifier with NMOS input stage or complementary rail-to-rail input stages, other amplifier topologies that are not FCTA, amongst others.


The connections of the elements in FIG. 1C are described as follows: The body terminal of all NMOSFETs in FIG. 1C are connected to node 2 which is VSS, and the body terminals of PMOSFETs are connected to node 1 which is VDD. The upper terminals of the bias current sources I101C, I103C, I105C, I107C, I109C, I110C, I111C, IM112C, and IM113C are connected to node 1, which is VDD. Bias current sources I100C, I102C, I104C, I106C, and I108C have their lower terminals connected to node 2, which is VSS. The source terminal of PMOSFETs M100C, M102C, M106C, M110C, M112C, M116C, and M120C are connected to node 1 which is VDD. The source terminal of NMOSFETs M124C, M125C, M128C, M101C, M103C, M107C, M111C, M113C, M117C, and M121C are connected to node 2 which is VSS. Node 100C is the VIN+ terminal of the amplifier and connected to the gate terminal of M122C. Node 101C is the VIN− terminal of the amplifier and connected to the gate terminal of M123C. Node 102C is connected to the source terminals of M122C, and M123C. Node 103C is connected to the drain terminal of M122C, the drain terminal of M107C, the source terminal of M105C, and the source terminal of M109C. Node 104C is connected to the drain terminal of M123C, the drain terminal of M117C, the source terminal of M115C, and the source terminal of M119C. Node 105C is connected to the drain terminal of M100C, the gate terminal of M108C, and the upper terminal of I100C. Node 106C is connected to the drain terminal of M102C, the gate terminal of M102C, the drain terminal of M104C, and the upper terminal of I102C. Node 107C is connected to the drain terminal of M108C, the drain terminal of M109C, the gate terminal of M106C, and the gate terminal of M116C. Node 108C is connected to the drain terminal of M106C, and the source terminal of M104C, the source terminal of M108C. Node 109C is connected to the drain terminal of M110C, the gate terminal of M118C, and the upper terminal of I104C. Node 110C is connected to the drain terminal of M112C, the gate terminal of M112C, the drain terminal of M114C, and the upper terminal of I106C. Node 111C is connected to the drain terminal of M116C, the source terminal of M114C, and the source terminal of M118C. Node 112C is connected to the drain terminal of M120C, the gate terminal of M120C, the gate terminal of M114C, the gate terminal of M104C, and the upper terminal of I108C. Node 113C is connected to the drain terminal of M124C, the gate terminal of M127C, and the lower terminal of I111C. Node 114C is connected to the drain terminal of M125C, the gate terminal of M125C, the drain terminal of M126C, and the lower terminal of I112C. Node 115C is connected to the drain terminal of M127C, the lower terminal of I113C, the gate terminal of M128C, the gate terminal of M107C, and the gate terminal of M117C. Node 116C is connected to the drain terminal of M128C, and the source terminal of M127C, the source terminal of M126C. Node 117C is connected to the drain terminal of M101C, the gate terminal of M109C, and the lower terminal of I101C. Node 118C is connected to the drain terminal of M103C, the gate terminal of M103C, the drain terminal of M105C, and the lower terminal of I103C. Node 119C is connected to the drain terminal of M111C, the gate terminal of M119C, and the lower terminal of I105C. Node 120C is connected to the drain terminal of M113C, the gate terminal of M113C, the drain terminal of M115C, and the lower terminal of I107C. Node 121C is the high impedance high gain node or the output, VOUT, of the amplifier and connected to the drain terminal of M118C, and the drain terminal of M119C. Node 122C is connected to the drain terminal of M121C, the gate terminal of M115C, the gate terminal of M105C, the gate terminal of M126C, the gate terminal of M121C, and the lower terminal of I109C.


Note that five RGC-CM100D coupled with ICMA100D are utilized in the amplifier of FIG. 1C. The NMOS ones are RGC101C made up of M124C-M125C-M126C-M127C-M128C-I111C-I112C; RGC102C made up of M101C-M103C-M105C-M107C-M109C-I101C-I103C; and RGC103C made up of M111C-M113C-M115C-M117C-M119C-I105C-I107C. The PMOS ones are RGC104C made up of M100C-M102C-M104C-M106C-M108C-I100C-I102C; and RGC105C made up of M110C-M112C-M114C-M116C-M118C-I104C-I106A. It would be possible that the amplifier of FIG. 1A would function properly without RGC101C and RGC104C, which can be substituted with diode connected cascodes. The RGC101C and RGC104C are added in the embodiment of FIG. 1C, in part, for better matching considerations and improved AC performance. Note that, in order to avoid repeating how the amplifier is benefited by each of RGC101C to RGC105C, the explanation of one RGC-CM (e.g., RGC103C) is deemed sufficient, regarding the function of RGC-CM100D coupled with ICMA100D in the FCTA.


For explanation regarding RGC-CM100D coupled with ICMA100D refer to its detailed description. A general description is provided for more context in how utilizing RGC-CM100D coupled with ICMA100D can improve the amplifier's performance. In the amplifier embodiment of FIG. 1C, the VIN is applied to a differential CSA, containing M122C and M123C, whose current outputs feed a differential CGA, containing M109C and M119C. Then, the differential current outputs of this CGA feed a ‘current mirror’, made up of M106C and M116C, to make a single ended output, VOUT. The RGC-CM100D coupled with ICMA100D (of FIG. 1D) is chiefly utilized in the CGAs and ‘current mirrors’ of the FCTA of FIG. 1C, in order to increase the amplifier's output impedance, expand its input-output voltage span, which helps the FCTA's dynamic response at low power.


Via M127C and M128C, the current I113C≈i is mirrored and scaled onto IDM107C≈IDM117C≈3i. The FCTA input currents IM122C≈IM123C≈i (part of CSA) are fed into the source terminals of M109C and M119C, respectively, which are the differential inputs of the CGAs. The I103C=2i and I107C=2i flow through ICMAs of RGC102C and RGC103C, respectively, and passed onto M107C and M117C, and in that order. Therefore, the operating currents of CGA's are IDM109C≈IDM107C−I103C−IDM103C−IDM122C≈i and IM119C≈IDM117C−I107C−IDM113C−IDM123C≈i. As explained in the RGC-CM100D coupled with ICMA100D section, the auxiliary amplifier function (performed by ICMA containing M115C, M113C, M111C, I107C, and I107C) regulates the VGM119C which increase the output impedance and gain of FCTA's CGA (containing M111C), and thereby increases the ROUT and gain (Av) of the FCTA at node 121C. Note that main amplifier's ROUT∝ro×(ro×gm)2 and Av is ∝ROUT×gmi. Hence, Av∝(ro×gm)3, with the amplifier FETs operating in the subthreshold region. To save current consumption, note that amplifier's gain can be increased by raising only I110C and I113C, instead of increasing the current consumption of the whole amplifier, given that Av is ∝ROUT×gmi, and gmi is the input stage transconductance that is roughly ∝I110C/VT.


Moreover, the DC common mode range of the main amplifier inputs (M122C and M123C) is expanded. The M105C and M115C source terminals sense the 103C and 104C signals in current mode. The DC input and output range of the amplifier is limited by V103C and V104C, which can be predetermined by the gate voltages of M105C and M115C, which are set by the scale and current of M121C. As such, VGS121C−VGS115C=VGS121C−VGS105C=ΔVGS≈VDSM117C≈VDSM107C≈VT×ln(r/n), which enable the input and output voltages of the amplifier to can get much closer to the rails. Additionally, note that the same type of FETs (e.g., NMOSFETs) are utilized in the CGAs function, the DC voltage shift function that expands the VIN and VOUT span, plus the auxiliary amplifier function (performed by ICMA) that increases ROUT of RGC103C and RGC105C (and hence raises the Av of the amplifier) which improves consistency of DC, AC, and transient performance over temperature, power supply, and process variations


In summary, besides providing the extra gain, the benefits of utilizing plurality of RGC-CM100D coupled with ICMA100D in an amplifier are as follows. First, the same type (e.g., NMOSFET) FETs are used in each of the main amplifier's common gate amplifiers (CGA), and RGC-CM's auxiliary amplifiers (AAUX) whose function is accomplished by the ICMAs. Given that for example the NMOSFET device parameters match and track each other better, therefore the consistency and stability of the main amplifier's DC, AC, and transient specifications are improved over fabrication process, temperature, and power supply variations. If for example CGA was based on NMOFETS, and AAUX was based on mix of PMOSFETs and NMSOFETs, then that would increase the risk of inconsistencies in the amplifier's performance in the long run in manufacturing. FIG. 1J illustrates a worst case (WC) AC simulation of FIG. 1C that Av˜110 dB (i.e., gain of over 315,000), and PM˜40° is achievable with the amplifier utilizing the plurality of RGC-CM coupled with ICMA. FIG. 1K upper graph (i) is the WC transient simulation of FIG. 1C indicating that SR˜1.7 V/μS and ts˜1.3 μS is achievable with the amplifier of FIG. 1C utilizing the plurality of RGC-CM coupled with ICMA. Moreover, FIG. 1K lower graph (ii) indicates current consumption of ˜180 nano Ampere (i.e., one over one billion of an Ampere) is achievable with the amplifier of FIG. 1C utilizing the plurality of RGC-CM coupled with ICMA. Second, the aforementioned benefits (see prior section) of the same RGC-CM100D coupled with ICMA100D is utilized in the amplifier in repeated (plurality of) instances, which carries over its benefits to the amplifier including small size, low current, and higher speed. Third, the DC voltage shift provided in the RGC-CM100D coupled with ICMA100D widens the amplifier's rail to rail span, which is beneficial especially in the sub-1V power supply environment where every 10 mV, of extra voltage swing closer to the rails, counts.


Section (VI): Detailed Description of Amplifier (AMP100E) Utilizing Plurality of RGC-CM100F Coupled with CSGA100F, as Illustrated in FIG. 1F



FIG. 1E is a circuit schematic showing a embodiment of a folded cascode transconductance amplifier (FCTA) that utilizes plurality of RGC-CM100F coupled with CSGA100F of FIG. 1F that is described in the previous sections. When RGC-CM100F coupled with CSGA100F is utilized in a FCTA, it expands the input-output voltage span and increases the amplifier's output impedance which in turn increases its gain (Av). Also, the other benefits of RGC-CM100F coupled with CSGA100F, are carried over to improve the FCTA performance including small size, low cost, low power, and higher speed.


Note that alternative amplifier embodiments may be possible such as an amplifier with NMOS input stage or complementary (PMOS and NMOS) rail-to-rail input stages, other amplifier topologies that are not folded cascode transconductance, amongst others.


The connections of the elements in FIG. 1E are described as follows. The body terminal of all NMOSFETs in FIG. 1E are connected to node 2 which is VSS, and the body terminals of PMOSFETs are connected to node 1 which is VDD. Bias current sources I101E, I103E, I104E, I107E, I109E, I110E, I113E, I115E, and I119E have their upper terminals connected to node 1, which is VDD. The lower terminal of I100E, I102E, I106E, I108E, I117E, I105E, and I111E are connected to node 2, which is VSS. Node 1 is also connected to the drain terminal of M100E, the gate terminal of M100E, the drain terminal of MM110E, and the gate terminal of M110E. Also, node 2 is connected to the drain terminal of M123E, the gate terminal of M123E, the drain terminal of M101E, the gate terminal of M101E, the drain terminal of M111E, and the gate terminal of M111E. The source terminal of M131E, M107E, and M117E are connected to node 2 that is VSS. The source terminal of M106E and M116E are connected to node 1 that is VDD. Node 101E is connected to gate terminal of M120E, and it is the positive side of the amplifier's input voltage terminal, or VIN+. Node 102E is connected to gate terminal of M121E, and it is the negative side of the amplifier's input voltage terminal, or VIN−. Node 103E is connected to source terminal of M120E, the source terminal of M121E, and the lower terminal of current source I112E. Node 104E is connected to the drain terminal of M120E, the source terminal of M109E, the drain terminal of M107E, and the gate terminal of M105E. Node 105E is connected to the drain terminal of M121E, the source terminal of M119E, the drain terminal of M117E, and the gate terminal of M115E. Node 106E is connected to source terminal of M100E, the source terminal of M104E, the gate terminal of M102E, and the upper terminal of current source I100E. Node 108E is connected to drain terminal of M102E, the gate terminal of M108E, and the upper terminal of I102E. Node 107E is connected to source terminal of MM102E, the drain terminal of M104E, and the lower terminal of I104E. Node 109E is connected to the source terminal of M108E, the drain terminal of M106E, and the gate terminal of M104E. Node 110E is connected to the drain terminal of M109E, the drain terminal of M108E, the gate terminal of M106E, the gate terminal of M116E. Node 111E is connected to source terminal of M110E, the source terminal of M114E, the gate terminal of M112E, and the upper terminal of current source I106E. Node 112E is connected to the source terminal of M112E, the drain terminal of M114E, and the lower terminal of I110E. Node 113E is connected to drain terminal of M112E, the gate terminal of M118E, the upper terminal of I108E. Node 114E is connected to the source terminal of M118E, the drain terminal of M116E, and the gate terminal of M114E. Node 115E is connected to the drain terminal of M119E, the drain terminal of M118E, and it is the output of the amplifier, VOUT. Node 116E is connected to source terminal of M123E, the source terminal of M127E, the gate terminal of M125E, and the lower terminal of current source I113E. Node 117E is connected to drain terminal of M125E, the gate terminal of M129E, and the lower terminal of I115E. Node 118E is connected to source terminal of M125E, the drain terminal of M127E, and the upper terminal of I117E. Node 119E is connected to the drain terminal of M129E, the gate terminal of M131E, the gate terminal of M107E, and the gate terminal of M117E. Node 120E is connected to the source terminal of M129E, the drain terminal of M131E, and the gate terminal of M127E. Node 121E is connected to source terminal of M101E, the source terminal of M105E, the gate terminal of M103E, and the lower terminal of current source I101E. Node 122E is connected to drain terminal of M103E, the gate terminal of M109E, and the lower terminal of I103E. Node 123E is connected to source terminal of M103E, the drain terminal of M105E, and the upper terminal of I105E. Node 124E is connected to source terminal of M111E, the source terminal of M115E, the gate terminal of M113E, and the lower terminal of current source I107E. Node 125E is connected to drain terminal of MM113E, the gate terminal of M119E, and the lower terminal of I109E. Node 126E is connected to source terminal of M113E, the drain terminal of M115E, and the upper terminal of I111E.


Note that there are five of RGC-CM100F coupled with CSGA100F that are utilized in the amplifier embodiment of FIG. 1E. The lower ones on the NMOS side of FCTA current mirrors are RGC101E made up of M123E-M125E-M127E-M129E-M131E-I113E-I115E-I117E; RGC102E made up of of M101E-M103E-M105E-M107E-M109E-I101E-I103E-I105E; and RGC103E made up of M111E-M113E-M115E-M117E-M119E-I107E-I109E-I111E. On the PMOS side of FCTA current mirrors there are RGC104E made up of M100E-M102E-M104E-M106E-M108E-I100E-I102E-I104E; and RGC105E made up of M110E-M112E-M114E-M116E-M118E-I106E-I108E-I110E. It would be possible that the amplifier of FIG. 1E would function properly without RGC101E and RGC104E, which can be substituted with diode connected cascodes. The RGC101E and RGC104E are utilized in the amplifier of FIG. 1E, in part, for systematic matching considerations and improved AC performance. Note that, in order to avoid repeating how the amplifier is benefited by each of RGC101E to RGC105E, the explanation of one RGC-CM100F coupled with CSGA100F (e.g., RGC103E) is deemed sufficient, regarding its function in the FCTA.


For explanation regarding RGC-CM100F coupled with CSGA100F, refer to its detailed description. A general description is provided for more context in how utilizing RGC-CM100F coupled with CSGA100F can improve the main amplifier's performance. The VIN is applied to a differential CSA, containing M120E and M121E, whose current outputs feed a differential CGA, containing M109A and M119A. The differential current outputs of this CGA feed a current mirror function containing M106E and M116E, to make a single ended output, VOUT. For the FCTA during steady state, the current I119E≈i controls VGSM131E that establishes IDM131E, which is mirrored and scaled onto IDM107E≈IDM117E≈2i. The FCTA's CSA input currents IM120E≈M121E≈i are fed into the source terminals of M109E and M119E, respectively, which are the differential inputs of the CGAs. At node 105E, the IDM120E≈1i subtracts from IDM117E≈2i which provides for IDM119E≈1i. At node 104E, the IDM121E≈i subtracts from IDM107E≈2i which provides for IDM109E≈1i. As explained in the RGC-CM coupled with composite amplifier in previous sections, the AAUX (containing M113E-I109E) regulates the VGM119E which increases ROUT of RGC-CM in RGC103E. Similar discussion in applicable to role of RGC102E, RGC104E, and RGC105E in FCTA here. As a result, effectively, the output impedance and gain of CGAs and current mirrors in FCTA are increased, and thereby the ROUT and gain (Av) of the FTCA are increased at node 115E. Note that main amplifier's ROUT is about ∝ro×(ro×gm)2 and Av is about ∝ROUT×gmi∝(ro×gm)3, in the subthreshold region of operations. To save current consumption, note that Av can be increased by raising only I112E and I119E, instead of increasing the current consumption of the FCTA, given that Av is ∝ROUT×gmi, and gmi is the input stage transconductance that is roughly ∝I112E/VT. Moreover, the input DC common mode range of the FCTA (M120E and M121E) is expanded because, with the built-in offsets (generated by the scaled M101E-M105E and M111E-M115E) in RGC102E and RGC103E, the V104E and V105E can get closer to VSS. As noted earlier, this built-in offsets in for example RGC103E is approximately, VGSM111E−VGSM115E=ΔVGS≈VDSM117E≈VT×ln(r/s). Hence, by proper scaling and operating current in M120E-M121E , the inputs of FCTA can span to the negative rail, VSS.


The voltage span of the output of FCTA is also improved. The built in offset in PMOSFETs in RGC101E to RGC103E is chiefly a function of VT and mostly independent of PMOSFET device parameters, such as PMOSFET VTH. Similarly, the built in offset in NMOSFETs in RGC104E and RGC105E is chiefly a function of VT and mostly independent of NMOSFET device parameters, such as NMOSFET VTH. Therefore, the voltages at nodes 104E, 105E, 109E, and 114E are mostly independent of MOSFET device parameters, and track each other given that they are mostly a function of VT. Therefore, maximum VOUT is approximately ≤VDD−VT×ln(r/s)−VDSM118E-sat. Also, minimum VOUT is approximately ≤VSS+VT×ln(r/s)+VDSM119E-sat.


The amplifier is improved in consistency of DC, AC, and transient performance over temperature, power supply, and process variations, in part because the auxiliary amplifier (AAUX) function in each of RGC-CM100F coupled with CSGA100F is made of the same channel FET as the FCTA's CGA and current mirror. The fact that the built-in offset for all RGC-CM (coupled with composite amplifier) is mostly a function of VT also helps reduce systematic mismatches in the FCTA signal path and helps improve FCTA performance over temperature, power supply, and process variations.



FIG. 1L illustrates a worst case (WC) AC simulation of FIG. 1E that Av˜110 dB, PM˜30° is achievable with the FCTA after utilizing the plurality of RGC-CMs coupled with composite amplifiers. FIG. 1M upper graph (i) is the WC transient simulation of FIG. 1E indicating that SR˜3V/μS and ts˜1 μS is achievable with the FCTA after utilizing the plurality of RGC-CMs coupled with composite amplifiers. Moreover, FIG. 1M lower graph (ii) indicates current consumption of ˜180 nano Ampere is achievable with the FCTA after utilizing the plurality of RGC-CMs coupled with composite amplifiers.


In summary, besides providing the extra gain, the benefits of utilizing RGC with composite amplifier in the main amplifier include the following. First, the same type of FETs are used in FCTA's CGA and current mirrors and those used in RGC-CM's AAUX. The build-in offset in RGC-CM is generated by PMOSFET and NMOSFET. However, the build-in offset itself is roughly independent of either PMOSFET and NMOSFET device parameters such as CMOSFET VTH. The build-in offset is mostly a function of VT which is highly predictable and insensitive to process variations. Therefore the FCTA's DC and AC specifications and stability is improved over fabrication process, temperature, and power supply variations. Second, each of the amplifier's RGC101E to RGC105E are self biased (i.e., no need for separate voltages to set its internal bias nodes) saving current and die space. Third, the aforementioned benefits of the same RGC-CM100F coupled with CSGA100F that is utilized in the FCTA in repeated instances, carries over to the FTCA the improvements in small size, low current, higher speed, and wider output voltage span as well as consistency of specification performance over fabrication process, temperature, and power supply variations. Fourth, the DC voltage shift provided in the RGC-CM100F coupled with CSGA100F expands the amplifier's DC common mode input and output ranges closer to the rails. As noted earlier, for example, a 75 mV head-room expansion at the input-output of each of the upper PMOS and the lower NMOS of RGC-CM100F coupled with CSGA100F, can expand the voltage span at the input-output of the amplifier as well as at the high gain (impedance) node of the FCTA by 150 mV or 15%, which is beneficial especially in the sub-1V power supply environment.


Section (VII): Detailed Description of First Embodiment of an Amplifier of FIG. 2A Utilizing the First Floating Current Source (FCS200A)



FIG. 2A is a schematic circuit diagram of the embodiment illustrating an amplifier utilizing the first FCS (FCS200A) shown in BLOCK: FCS200A at the left bottom side of FIG. 2A. The embodiment discloses FCS200A, which emulates the benefits of a FCS in the amplifier current mirror network, enables the amplifier to operate at low VDD≥˜VGS+2VDS, improves matching of upper and lower currents independent of the common mode voltage swings.


The amplifier of FIG. 2A is a conventional folded cascode transconductance amplifier (FCTA) that utilizes FCS200A. As indicated in prior sections, broadly speaking, a FCTA is composed of a CSA that converts the input differential voltage to differential currents which are then fed onto a differential input and differential output CGA that is fast. In a FCTA, the differential current outputs of CGA are then steered onto a CM to generate a single output voltage gain at VOUT. The amplifier of FIG. 2A has two complementary inputs though, a PMOSFET input and a NMOSFET input, operating in parallel in order to provide rail-to-rail input operations which is consistent with the benefits of this disclosure. This arrangement is chosen so that either the PMOSFET inputs keep on operating when NMOSFETs run out of headroom near the negative rail (VSS or GND), or the NMOSFET inputs keep on operating when PMOSFETs run out of headroom near the positive rail (VDD). On the NMOSFET side of FCTA, the VIN is applied to a differential CSAN200A containing M209A and M211A. The differential current outputs of CSAN200A then feed into a CGAP200A containing M202A-M214A in conjunction with a current mirror (CMP200A) containing M200A-M212A, the combination of which generates a single ended output at node 212A. On the PMOSFET side of FCTA, the VIN is applied to a differential CSAP200A containing M208A and M210A. The differential current outputs of CSAP200A then feed into a CGAN200A containing M203A-M215A in conjunction with a current mirror (CMN200A) containing M201A-M213A, combination of which generates a single ended output at node 212A, which is the amplifier's high impedance, high gain output (VOUT).


There are other amplifier configurations that can utilize this FCS. One such example would be a FCTA with a gm control circuit to keep the amplifier's input transconductance (gm) constant across input voltage common mode range (VINCM). Another example would be FCTA that utilizes regulated cascode current mirrors (RGCs) to improve the amplifier's performance, including increasing the gain of the amplifier. Note also that in FIG. 2A that the current source scale factors (e.g., 2×i=I205A or 1×i=I201A, etc) and MOSFET scale factors (e.g., W/L of M201A versus W/L of M213A, or W/L of M204A versus W/L of M202A) can be altered for different cost-performance goals, such as speed versus power versus accuracy versus larger die which would cost more.


The connections of the elements in FIG. 2A are described as follows. The body terminal of all NMOSFETs in FIG. 2A are connected to node 2 which is VSS, and the body terminals of PMOSFETs are connected to node 1 which is VDD. The lower terminals of current sources I200A, I202A, and I205A are connected to node 2. The upper terminals of current sources I201A, I203A, and I204A are connected to node 1. The source terminals of M200A, M206C, and M212A are connected to node 1. The source terminals of M201A, M207A, and M213A are connected to node 2. Node 200A is connected to gate terminal of M206A, the drain terminal of M206A, the gate terminal of M204A, the gate terminal of M202A, and the gate terminal of M214A. Node 201A is connected to the gate terminal of M207A, the drain terminal of M207A, the gate terminal of M205A, the gate terminal of M203A, and the gate terminal of M215A. Node 202A is connected to the gate terminal of M200A, the drain terminal of M204A, the gate terminal of M212A, and the upper terminal of current source I200A. Node 203A is connected to the gate terminal of M201A, the drain terminal of M205A, the gate terminal of M213A, and the lower terminal of current source I201A. Node 204A is connected to the drain terminal of M209A, the drain terminal of M200A, the source terminal of M204A, the source terminal of M202A, and the drain terminal of M203A. Node 205A is connected to the drain terminal of M208A, the drain terminal of M201A, the source terminal of M205A, the source terminal of M203A, and the drain terminal of M202A. Node 206A is connected to the source terminal of M208A, the source terminal of M210A, and the lower terminal of current source I204A. Node 207A is connected to the source terminal of M209A, the source terminal of M211A, and the upper terminal of current source I205A. Node 208A is connected to the drain terminal of M211A, the drain terminal of M212A, and the source terminal of M214A. Node 209A is connected to the drain terminal of M210A, the source terminal of M215A, and the drain terminal of MM213A. Node 210A is connected to the gate terminal of M208A, and the gate terminal of M209A, and it is the negative input terminal (VIN−) of the amplifier. Node 211A is connected to the gate terminal of M210A, and the gate terminal of M211A, and it is the positive input terminal (VIN+) of the amplifier. Node 212A is the high impedance output (high gain node) of the amplifier, VOUT, which is connected to the drain terminal of M215A, and the drain terminal of M214A. To be clear, the BLOCK FCS200A contains the following FETs: M200A, M202A, M204A, M206A, I200A, and I202A plus M201A, M203A, M205A, M207A, I201A, and I203A.


One of the reasons for utilizing a FCS (or its equivalent function in an amplifier current mirror network) such as that of a FCTA, is to make the biasing at the summing junction of the FCTA output more insensitive to the common mode voltage swings in order to improve the amplifier's accuracy. The prior art FCS circuit is illustrated in FIG. 2D, IM209D+IM208D flow into drains of M202D and M203D, and then onto M200D and M201D, respectively, thus equalizing the operating currents in M200D and M201D for the FCS. While this prior art FCS circuit in FIG. 2D is simple, but its minimum VDD is restrictive. The minimum VDD of FIG. 2D is the greater of VGSM204D30 VGSM206D+VDSI202D or ˜VGSM205E+VGSM207E+VDSI203E, which is restrictive. The BLOCK: FCS200A at the left bottom side of FIG. 2A, improves the FCS performance including lowering its minimum power supply voltage, and described as follows. For clarity, the operations of the FCS200A independent of the amplifier is described first. Thus, IDM208A, and IDM209A are set to zero for now and non-idealities such as device mismatches are set aside. The FCS200A utilizes the first regulating circuit which is made up of M204A, and M206A, and current sources I200A, and I202A. Given that VGSM202A=VGSM204A⇒IDM202A≈IDM204A≈IN=I200A. The second regulating circuit utilized in FCS200A is made up of M205A, and M207A, and current sources I201A, and I203A. Also, given that VGSM205A=VGSM203A⇒IDM205A≈IDM203A≈IP=I201A. The KCL at node 205A computes to IDM205A+IDM203A+IDM202A=IDM201A≈2×IP+IDM202A≈2×IP+IN. Similarly, the KCL at node 204A computes to IDM204A+IDM202A+IDM203A=IDM200A≈2×IN+IDM203A≈2×IN+IP. Setting IN=IP≈1×i, results in IDM200A≈IDM201A≈3×i, which will mirror onto M212A and M213A, respectively. If there is a mismatch of 10% and for example IP0.9×i and IN≈1×i⇒IDM200A2.8×i and IDM201A2.9×i⇒IDM200A/IDM201A≈97%. As such an error due to mismatch between IP and IN is roughly attenuated by a factor of 3 in this FCS200A embodiment. Now that the FCS200A is explained, the operations of the amplifier including the FCS200A is described next.


The steady state operations of the FIG. 2A amplifier that utilizes FCS200A is described first. With the amplifier operating under steady state conditions, on the upper side of the amplifier, IDM209A≈IDM211A≈1×i. The first regulating circuit, containing M204A and current source IM200A, regulate the gate voltage of M200A, and the KCL operates on node 204A where IDM200A≈IDM204A+IDM202A+IDM203A+IDM209A≈4×i . Given that currents through M200A and M212A (≈2×i) are mirrored and scaled, the sum of currents at node 208A would result in IDM214A≈IDM212A−IDM211A≈i. Similarly, in steady state conditions, the amplifier's inputs are balanced and IDM208A≈IDM210A≈1×i. The second regulating circuit, containing M205A and current source IM201A, regulate the gate voltage of M201A, and the KCL operates on node 205A where IDM201A≈IDM205A+IDM203A+IDM202A+IDM208A4×i . Given that currents through M201A and M213A (≈2×i) are mirrored and scaled, the sum of currents at node 209A would result in IDM215A≈IDM213A−IDM210A≈i. Therefore, at the summing gain node 212A of FCTA, operating current are in balance with IDM214A≈IDM215A≈i.


With regards to node 204A, operating currents IDM202A=IDM204A≈i and IDM203A≈i are held constant, setting aside non-idealities. As the KCL operates on node 204A, an input voltage change (ΔvIN) at the FCTA amplifier applied across M209A-M211A generates a current change (ΔiN) in IDM209A. This in turn would cause the first regulating circuit (containing M204A and current source IM200A) to regulate the gate voltage of M200A which results in the ΔiN to flow into M200A, while IDM200A is ‘current mirrored’ with IDM212A (and scaled). Note that the dynamic response of this current mirror (M200A-M212A) is also improved. This is because the second regulating circuit containing M204A and current source IM200A is configured as a common gate amplifier (CGA), which is very fast, and whose output drives the gate terminals of M200A-M212A.


Similarly, with regards to node 205A, as explained earlier, the operating currents IDM203A=IDM205A≈i and IDM202A≈i are held constant, setting aside non-idealities. As the KCL operates on node 205A, an input voltage change (ΔvIN) at FCTA, applied across M208A-M210A, generates a current change (ΔiP) in IDM208A. This in turn would cause the second regulating circuit (containing M205A and current source IM201A) to regulate the gate voltage of MM201A which would result in the ΔiP to flow into MM201A, while IDM201A is ‘current mirrored’ with IDM213A (and scaled). Note also that the dynamic response of this current mirror (M201A-M213A) is improved. Similarly, this is because the second regulating circuit containing M205A and current source IM201A is configured as a common gate amplifier (CGA), which is inherently fast, and whose output drives the gates terminals of M201A-M213A.


The FCS200A minimum VDD≥VGS2VDS is improved using fewer transistors and less current (versus prior art of FIG. 2D where minimum VDD≥2VGS+VIDS). Note that, in utilizing FCS200A, the input common mode span of the amplifier remains wide, given that node 203A, 209A, and 204A, 208a are set a VDS above and below rails.


In summary, the FCS200A block that is utilized in the amplifier of FIG. 2A, operates as follows: the NMOSFET cascode current source (M201A and M203A cascoded on node 205A) and the PMOSFET cascode current source (M200A and M202A cascoded on node 204A) are arranged such that the drain current of M202A is fed into node 205A while the drain current of M203A is fed into node 204A. Concurrently, a first regulating circuit regulates the gate terminal M200A, and a second regulating circuit regulates the gate terminal M201A such that the operating currents of M200A and M201A are substantially equalized, while the FCS200A operates with low power supply voltages.



FIG. 2E upper graph (i) is the WC transient simulation of FIG. 2A indicating that SR˜1V/μS and ts˜1.5 μS is achievable with the FCTA utilizing the first FCS. Moreover, FIG. 2D middle graph (ii) indicates current consumption of ˜60 nano Ampere is achievable, and FIG. 2D bottom graph (iii) indicates that with FCTA in unity gain configuration subjected to a 1 volt (V) change in common mode input voltage followed by 1V change in output voltage would cause a ˜40 Pico ampere (pA) change in the operating current of FCS, which is about 30 nA.


In conclusion, some of the benefits of utilizing FCS200A in an amplifier may include one or more of the following. First, the FCS can operate at lower power supply, using fewer transistors with less current, which is beneficial for the amplifier that utilizes this FCS. Second, the FCS can provides matching between upper and low cascoded current sources which improves amplifier's performance, including lowering its offset and noise. Third, given the regulating circuit of FCS200A is based on common gate amplifier (CGA) configuration, the dynamic response of the FCS is improved which improves the dynamic response of the amplifier that utilizes it.


Section (VIII): Detailed Description of Second Embodiment of an Amplifier of FIG. 2B Utilizing the Second Floating Current Source (FCS200B)



FIG. 2B is a schematic circuit diagram of the embodiment illustrating an amplifier utilizing the second FCS, which is depicted in BLOCK: FCS200B at the left bottom side of FIG. 2B. Similar to the first FCS discussed in the previous sections, the disclosed FCS200B, emulating the function of a floating current mirror in the amplifier's current mirror network, here that enables the amplifier to operate at low VDD≥˜VGS+2VDS, improves speed, and improve accuracy by making the current matching more independent of common mode voltage swings. Refer to the description provided regarding the FCTA of FIG. 2A in the previous section, which is similar to the amplifier herein FIG. 2B that utilizes FCS200B. The FCS200B illustrated in BLOCK: FCS200B (at the left bottom side of FIG. 2B) which also utilizes two cascode current sources, PMOSFETs (M200B and M202B) and NMSOFETs (M201B and M203B). The upper and lower FETs (M202B and M203B) in the two cascodes are criss crossed by feeding the middle PMOSFET (M202B) drain current into the middle NMOSFET (M203B) source terminal and conversely feeding the middle NMOSFET (M203B) drain current into the middle PMOSFET (M202B) source terminal. Concurrently, regulating circuits regulate the gate voltages of the upper and lower FETs (M200B and M201B) in the two cascodes such that their operating currents (IDM200B and IDM201B) are substantially equalized.


As noted in the prior sections, it would be possible that there are other amplifier configurations besides FCTA that can utilize this FCS200B. Moreover, note that in FIG. 2B the current source scale factors (e.g., 2×i=I205B or 1×i=I201B, etc) and MOSFET scale factors (e.g., W/L of M201B versus W/L of M213B, or W/L of M204B and W/L of M202B versus W/L of M216B and W/L of M220B) can be altered in aiming for different performance goals, such as speed versus power versus accuracy.


The connections of the elements in FIG. 2B are described as follows. The body terminal of all NMOSFETs in FIG. 2B are connected to node 2 which is VSS, and the body terminals of PMOSFETs are connected to node 1 which is VDD. The lower terminals of current sources I200B, I202B, I205B, I206B, and I208B are connected to node 2. The upper terminals of current sources I201B, I203B, I204B, I207B, and I209B are connected to node 1. Source terminals of M200B, M206B, M212B, and M218B are connected to node 1. Source terminals of M201B, M207B, M215B, and M219B are connected to node 2. Node 200B is connected to the gate terminal of M206B, the drain terminal of M206B, the gate terminal of M202B, the gate terminal of M214B, and the gate terminal of M220B. Node 201B is connected to the gate terminal of M207B, the drain terminal of M207B, the gate terminal of M203B, the gate terminal of M215B, and the gate terminal of M221B. Node 202B is connected to the drain terminal of M218B, the source terminal of M220B, and the source terminal of M216B. Node 203B is connected to the drain terminal of M219B, the source terminal of M221B, and the source terminal of M217B. Node 204B is connected to the drain terminal of M220B, the gate terminal of M218B, and the upper terminal of I208B. Node 205B is connected to the drain terminal of M221B, the gate terminal of M219B, and the lower terminal of I209B . Node 206B is connected to the drain terminal of M216B, the gate terminal of M216B, the gate terminal of M204B, and the upper terminal of I206B. Node 207B is connected to the drain terminal of M217B, the gate terminal of M217B, the gate terminal of M205B, and the lower terminal of I207B. Node 208B is connected to the gate terminal of M200B, the drain terminal of M204B, the gate terminal of M212B, and the upper terminal of current source I200B. Node 209B is connected to the gate terminal of M201B, the drain terminal of M205B, the gate terminal of M213B, and the lower terminal of current source I201B. Node 210B is connected to the drain terminal of M200B, the source terminal of M204B, the source terminal of M202B, the drain terminal of M203B, and the drain terminal of M209B. Node 211B is connected to the drain terminal of M201B, the source terminal of M203B, the source terminal of M205B, the drain terminal of M202B, and the drain terminal of M208B. Node 212B is connected to the source terminal of M208B, the source terminal of M210B, and the lower terminal of current source I204B. Node 213B is connected to the source terminal of M209B, the source terminal of M211B, and the upper terminal of current source I205B. Node 214B is connected to the drain terminal of M211B, the drain terminal of M212B, and the source terminal of M214B. Node 215B is connected to the drain terminal of M210B, the source terminal of M214B, and the drain terminal of M215B. Node 216B is connected to the gate terminal of M208B, and the gate terminal of M209B, and it is the negative input terminal (VIN−) of the amplifier. Node 217B is connected to the gate terminal of M210B, and the gate terminal of M211B, and it is the positive input terminal (VIN+) of the amplifier. Node 218B is the high impedance output (high gain node) of the amplifier, VOUT, which is connected to the drain terminal of M214B, and the drain terminal of M215B. To be clear, the BLOCK FCS200B containing the following: M200B, M202B, M204B, M206B, M216B, M218B, M220B, I200B, I202B, I206B, on the top side and I208B plus M201B, M203B, M205B, M207B, M217B, M219B, M221B, I201B, I203B, I207B, and I209B on the bottom side.


For clarity and consistency with the prior sections, the operations of the FCS200B is described first, independent of that of the amplifier. Thus, IDM208A, and IDM209A are assumed to be zero for the purpose of this segment's description, and also non-idealities such as device mismatches are set aside.


The FCS200B top side utilizes the first regulating circuit made up of M204B, M206B, M216B, M218B, and M220B, and current sources I200B, I202B, I206B, and I208B. In this embodiment, the I200B≈I206B≈I208B≈1i are set as equals constant current sources that bias VGSM204B≈VGSM220B≈VGSM216B. Applying the KVL to the voltage loop containing VGSM202B=VGSM220B−VGSM216B+VGSM204B≈VGSM220B. Therefore, IDM202B≈IDM220≈1i. The first regulating circuit is a current input amplifier, where VGSM206B−VGSM220B+VGSM216B establish the VGM204B. As such, M204B and I200B function like a common gate amplifier (CGAP200B), which is fast. The output of this CGAP200B regulates VGSM200B at the node 208B until the KCL at node 210B is satisfied, which is when IDM200B≈IDM202B+IDM203B+IDM204B+IDM209B.


The second regulating circuit utilized in the bottom side of FCS200B is made up of M205B, M207B, M217B, M219B, and M221B, and current sources I201B, I203B, I207B, and I209B. Similar to the top side, the I201B≈I207B≈I209B≈1i are set as equal constant current sources that bias VGSM205B≈VGSM217B≈VGSM221. Applying the KVL to the voltage loop containing VGSM203B=VGSM221B−VGSM217B+VGSM205B≈VGSM221B. Therefore, IDM203B≈IDM221≈1i. The second regulating circuit is a current input amplifier, where VGSM207B−VGSM221B+VGSM217B establish the VGM205B. As such, M205B and I201B function like a common gate amplifier (CGAN200B), which is fast. The output of this CGAN200B regulates VGSM201B at the node 209B until the KCL at node 211B is satisfied, which is when IDM201B≈IDM203B+IDM202B+IDM205B+IDM208B.


The steady state operations of the FIG. 2B amplifier that utilizes FCS200B is described as follows. In steady state conditions, the amplifier's inputs are balanced and IDM208B≈IDM210B≈1×i . The first regulating circuit, containing M204B and IM200B, regulates the gate voltage of M200B until KCL is satisfied by operating on node 210B where IDM200B≈IDM204B+IDM203B+IDM202B+IDM209B≈4×i. Given that currents through M200B and M212B (≈2×i) are mirrored and scaled, the sum of currents at node 214B would result in IDM214B≈IDM212B−IDM211B≈i. Similarly, with the amplifier operating under steady state conditions, on the upper complementary side of the amplifier, IDM209A≈IDM211A≈1×i. The second regulating circuit, containing M205B and IM201B, regulate the gate voltage of M201B until KCL is satisfied by operating on node 211B where IDM201B≈IDM205B+IDM203B+IDM202B+IDM208B≈4×i. Given that currents through M201B and M213B (≈2×i) are mirrored and scaled, the sum of currents at node 215B would result in IDM215B≈IDM213B−IDM210B≈i.


With regards to node 210B, operating currents IDM202B=IDM204B≈i and IDM203B≈i are held constant, setting aside non-idealities. As the KCL operates on node 210B, an input voltage change (ΔvIN) across M209B-M211B generates a current change (ΔiN) in IDM209B that would cause the first regulating circuit (containing M204B and current source IM200B) to regulate the gate voltage of M200B. As a result, ΔiN would flow into M200B, while IDM200B is ‘current mirrored’ with IDM212B (and scaled). Note that the dynamic response of this current mirror (M200B-M212B) is also improved. This is because this first regulating circuit containing M204B and current source IM200B is configured as a common gate amplifier (CGA), which is inherently fast, and whose output drives the gate terminals of the current mirror containing M200B-M212B.


With regards to node 211B, the operating currents IDM203A=IDM205A≈i and IDM202A≈i are held constant, setting aside non-idealities. While the KCL operates on node 211B, the input voltage change (ΔvIN) across M208B-M210B generates a current change (ΔiP) in IDM208B which would cause the second regulating circuit (containing M205B and current source IM201B) to regulate the gate voltage of MM201B. As a result, this ΔiP would flow into MM201B, while IDM201B is ‘current mirrored’ with IDM213B (and scaled). Note also that the dynamic response of this current mirror (M201B-M213B) is improved. This is because this second regulating circuit containing M205B and current source IM201B is configured as a common gate amplifier (CGA), which is very fast, and whose output drives the gates terminals of the current mirror containing M201B-M213B. The FCS200B minimum VDD≥VGS+2VDS is improved compared to the prior art of FIG. 2D where minimum VDD≥2VGS+VIDS. Note that, in utilizing FCS200B, the input common mode span of the amplifier remains wide, given that nodes 203B, 211B, 215B and 202B, 210B, 214B can be set as VDS sat above and below rails.


In summary, the FCS200B block that is utilized in the amplifier of FIG. 2B, operates as follows. The NMOSFET cascode current source (M201B and M203B cascoded on node 211B) and the PMOSFET cascode current source (M200B and M202B cascoded on node 210B) are arranged such that the drain current of M202B is fed into node 211B while the drain current of M203B is fed into node 210B. Concurrently, a first regulating circuit regulates the gate terminal M200B, and a second regulating circuit regulates the gate terminal M201B such that the operating currents of M200B and M201B are substantially equalized, while the FCS200B operates with low power supply voltages.



FIG. 2F upper graph (i) is the WC transient simulation of FIG. 2B indicating that SR˜1V/μS and ts˜1 μS is achievable with the FCTA utilizing the second FCS. Moreover, FIG. 2E middle graph (ii) indicates current consumption of ˜80 nano Ampere is achievable, and FIG. 2E bottom graph (iii) indicates that with FCTA in unity gain configuration subjected to a 1 volt (V) change in common mode input voltage followed by 1V change in output voltage would cause a ˜5 Pico Ampere (pA) change in the operating current of FCS, which is about 40 nA.


In conclusion, some of the benefits of utilizing FCS200B in an amplifier are the following. First, the FCS can operate at lower power supply which frees the amplifier to operate with lower power supply as well. Second, performance of FCS is improved, including its dynamic response, by separating nodes 207B from 201B and separating nodes 206B from 200B, which helps roughly shield the CGAs used in FCS from transients on nodes 200B and 200B. Hence, the transient response of the amplifier, in which the FCS is utilized, can be improved. Third, the FCS can provides some matching between upper and low cascoded current sources which improves amplifier's performance, including lowering its offset and noise of the amplifier in which the FCS is utilized. Fourth, given the regulating circuit of FCS200B is based on common gate amplifier, the dynamic response of the FCS is improved which improves the dynamic response of the amplifier in which it is utilized.


Section (IX): Detailed Description of Third Embodiment of an Amplifier of FIG. 2C Utilizing the Third Floating Current Source (FCS200C)



FIG. 2C is a schematic circuit diagram of the embodiment illustrating an amplifier utilizing the third FCS, which is depicted in BLOCK: FCS200C at the left bottom side of FIG. 2C. Similar to the FCSs discussed in previous sections, the disclosed FCS200B emulates the function of floating current source in the amplifier's current mirror network, here and enables the amplifier to operate at low VDD≥˜VGS2VDS, and improve accuracy by making current matching more independent of common mode swings.


As just noted, from a high level functional perspective, the embodiment of FCS200C here is similar to that of FCS200A and FCS200C disclosed in the previous sections. The embodiment of the FCS200C is illustrated in BLOCK: FCS200B (at the left bottom side of FIG. 2C) which also utilizes two cascode current sources, PMOSFETs (M200C and M202C) and NMSOFETs (M201C and M203C). Here, the middle cascode FETs (M202C and M203C) are arranged with their drain and source terminals crisscrossed. As such, the middle PMOSFET (M202C) drain current is fed into the middle NMOSFET (M203C) source terminal and conversely the middle NMOSFET (M203C) drain current is fed into the middle PMOSFET (M202C) source terminal. Concurrently, regulating circuits regulate the gate voltages of the upper and lower FETs (M200C and M201C) in the two cascodes such that their operating currents (IDM200C and IDM201C) are substantially equalized.


As noted in the prior sections, it is possible that there are other amplifier configurations, besides a FCTA amplifier topology, that can utilize this FCS200C. Moreover, note that in FIG. 2C, it would be possible that the current source scale factors (e.g., 2×i=I205C or 1×i=I209C, etc) and MOSFET scale factors (e.g., W/L of M201C versus W/L of M213C, or W/L of M220C versus W/L of M202C or W/L of M205C and W/L of M217C) can be altered in aiming for different performance goals, such as speed versus power versus accuracy.


The connections of the elements in FIG. 2C are described as follows. The body terminal of all NMOSFETs in FIG. 2C are connected to node 2 which is VSS, and the body terminals of PMOSFETs are connected to node 1 which is VDD. The lower terminals of current sources I200C, I202C, I205C, and I208C are connected to node 2. The upper terminals of current sources I201C, I203C, I204C, and I209C are connected to node 1. The source terminals of M200C, M206C, M212C, M218C, M222C, and M224C are connected to node 1. Source terminals of M201C, M207C, M213C, M219C, M223C, and M225C are connected to node 2. Node 200C is connected to the gate terminal of M206C, the drain terminal of M206C, the gate terminal of M202C, the gate terminal of M214C, the gate terminal of M220C, and the upper terminal of I202C. Node 201C is connected to the gate terminal of M207C, the drain terminal of M207C, the gate terminal of M203C, the gate terminal of M215C, the gate terminal of M221C, and the lower terminal of I203C. Node 202C is connected to the gate terminal of M216C, the drain terminal of M218C, and the source terminal of M220C. Node 203C is connected to the gate terminal of M217C, the drain terminal of M219C, and the source terminal of M221C. Node 204C is connected to the gate terminal of M218C, the drain terminal of M220C, and the upper terminal of I208C. Node 205C is connected to the gate terminal of M219C, the drain terminal of M221C, and the upper terminal of I209C. Node 206C is connected to the gate terminal of M222C, the gate terminal of M224C, the drain terminal of M224C, and the drain terminal of M204C. Node 207C is connected to the gate terminal of M223C, the gate terminal of M225C, the drain terminal of M225C, and the drain terminal of M205C. Node 208C is connected to the gate terminal of M200C, the gate terminal of M212C, the drain terminal of M216C, and the drain terminal of M222C. Node 209C is connected to the gate terminal of M201C, the gate terminal of M213C, the drain terminal of M217C, and the drain terminal of M223C. Node 210C is connected to the drain terminal of M200C, the source terminal of M202C, the drain terminal of M203C, the gate terminal of M204C, and the drain terminal of M209C. Node 211C is connected to the drain terminal of M201C, the source terminal of M203C, the drain terminal of M202C, the gate terminal of M205C, and the drain terminal of M208C. Node 212C is connected to the source terminal of M204C, the source terminal of M216C, and the upper terminal of I200 C. Node 213C is connected to the source terminal of M205C, the source terminal of M217C, and the lower terminal of I201 C. Node 214C is connected to the source terminal of M208C, the source terminal of M210C, and the lower terminal of I204 C. Node 215C is connected to the source terminal of M209C, the source terminal of M210C, and the upper terminal of I205 C. Node 216C is connected to the drain terminal of M212C, the source terminal of M214C, and the drain terminal of M211 C. Node 217C is connected to the drain terminal of M213C, the source terminal of M215C, and the drain terminal of M210 C. Node 218C is connected to the gate terminal of M208C, the gate terminal of M209C, and it is the positive input voltage terminal of the amplifier VIN+. Node 219C is connected to the gate terminal of M210C, the gate terminal of M211C, and it is the negative input voltage terminal of the amplifier VIN−. Node 220C is connected to the drain terminal of M214C, the drain terminal of M215C, and it is the (high gain, high impedance) output voltage terminal of the amplifier VOUT. To be clear, the BLOCK FCS200C is made up of the following: top side M200C, M202C, M204C, M206C, M216C, M218C, M220C, M222C, M224C, I200C, I202C, and the bottom side I208C plus M201C, M203C, M205C, M207C, M217C, M219C, M221C, M223C, M225C, I201C, I203C, and I209C


For clarity and consistency with the prior sections, the operations of the FCS200C is described first, independent of that of the amplifier. Thus, IDM208C, and IDM209C are set to zero and non-idealities such as device mismatches are set aside. The FCS200C utilizes the first regulating circuit, on the top side, made up of M206C, M218C, and M220C, and current sources, I202C, and I208C plus the amplifier AN200C (containing M204C, M266C, M226C, M222C, and current source I200C). Here, AN200C's output regulates VGSM200C until its inputs are substantially equalized, which is when VGSM202C≈VGSM220C that occurs when IDM202C≈IDM220C. Therefore, IDM202C≈IDM220≈1i, and the KCL operating on node 210C would result in IDM200C≈IDM202C+IDM203C+IDM209C.


On the complementary or bottom side, the FCS200C utilizes the second regulating circuit made up of M207C, M219C, and M221C, and current sources, I203C, and I209C plus the amplifier AP200C (containing M205C, M217C, M225C, M223C, and current source I201C). Here, AP200C'S output regulates VGSM201C until its inputs are substantially equalized, which is when VGSM203C≈VGSM221C that occurs when IDM203C≈IDM221C. Therefore, IDM203C≈IDM221≈1i, and the KCL operating on node 211C would result in IDM201C≈IDM203C+IDM202C+IDM208C.


The steady state operations of the FIG. 2C amplifier that utilizes FCS200C is described as follows. In steady state conditions, the amplifier's inputs are balanced and IDM208C≈IDM210C≈1×i . As stated earlier, the first regulating circuit containing AN200C regulates the gate voltage of M200C until KCL is satisfied by operating on node 210C where IDM200C≈IDM203C+IDM202C+IDM209C≈3×i . Given that currents through M200C and M212C (≈2×i) are mirrored and scaled, the sum of currents at node 216C would result in IDM214C≈IDM212C−IDM211C≈i. Similarly, with the amplifier operating under steady state conditions, on the top or complementary side of the amplifier, IDM209A≈IDM211A≈1×i. Also as stated earlier, the second regulating circuit containing AP200C regulates the gate voltage of M201C until KCL is satisfied by operating on node 211C where IDIDM201C≈IDM203C+IDM202C+IDM208C3×i. Given that currents through M201C and M213C (≈2×i) are mirrored and scaled, the sum of currents at node 217C would result in IDM215C≈IDM213C−IDM210C≈i.


Again while the KCL operates on node 210C, an input voltage change (ΔvIN) across M209C-M211C generates a current change (ΔiN) in IDM209C that would cause the first regulating circuit, containing AN200C, to regulate the gate voltage of M200C. As a result, ΔiN would flow into M200C, while IDM200C is ‘current mirrored’ with IDM212C (and scaled). Similarly, with the KCL operating on node 211C, the input voltage change (ΔvIN) across M208C-M210C generates a current change (ΔiP) in IDM208C that would cause the second regulating circuit containing AP200C, to regulate the gate voltage of MM201C. As a result, this ΔiP would flow into MM201C, while IDM201C is ‘current mirrored’ with IDM213C (and scaled).


The FCS200C minimum VDD≥VGS+2VDS is improved compared to the prior art of FIG. 2D where minimum VDD≥2VGS+VIDS. Note that, in utilizing FCS200C, the input common mode span of the amplifier remains wide, given that nodes 203C, 211C, 217C and 202C, 210C, 216C can be set as VDS above and below rails.



FIG. 2G upper graph (i) is the WC transient simulation of FIG. 2C indicating that SR˜1.5V/μS and ts˜1 μS is achievable with the FCTA utilizing the third FCS. Moreover, FIG. 2F middle graph (ii) indicates current consumption of ˜60 nano Ampere is achievable, and FIG. 2F bottom graph (iii) indicates that with FCTA in unity gain configuration subjected to a 1 volt change in common mode input voltage followed by 1V change in output voltage would cause a ˜70 Pico Ampere (pA) change in the operating current of FCS, which is about 20 nA.


In summary, the FCS200C block that is utilized in the amplifier of FIG. 2C, operates as follows. The NMOSFET cascode current source (M201C and M203C cascoded on node 211C) and the PMOSFET cascode current source (M200C and M202C cascoded on node 210C) are arranged such that the drain current of M202C is fed into node 211C while the drain current of M203C is fed into node 210C. Concurrently, a first regulating circuit regulates the gate terminal M200C, and a second regulating circuit regulates the gate terminal M201C such that the operating currents of M200C and M201C are substantially equalized, while the FCS200C operates with low power supply voltages. Some of the benefits of utilizing FCS200C in an amplifier are the following. First, the FCS can operate at lower power supply which frees the amplifier to operate with lower power supply as well. Second, the FCS can provides better matching between upper and low cascoded current sources which improves amplifier's performance, including lowering its offset and noise.


Section (X): Detailed Description of Amplifier Illustrated in FIG. 3A, Utilizing a First Noise Reduction Plus Speed Boost Circuit



FIG. 3A is a circuit schematic showing an amplifier (BLOCK 301A) utilizing the noise reduction plus speed boost circuit (BLOCK 300A). One of the contribution of this disclosure is a method of lowering the output noise of an amplifier by narrow-banding it, and then reinvigorating the slower dynamic range of narrow-banded amplifier, by introducing a bias current boost (that feeds the amplifier operating current) which is dynamically enabled when the input of the amplifier become imbalanced after receiving a large differential input transient signal. The dynamic current boost aims to speed the amplifier's dynamic response by boosting not just the amplifier's slew rate but also its settling time.


To improve noise the amplifier is narrow banded. To maintain ultra low power consumption, the steady state quiescent current consumption of the amplifier is kept at ultra low levels. To re-invigorate the dynamic response of the narrow banded amplifier back-up, the intermittent (dynamic) current consumption of the amplifier is increased. One way to narrow band the amplifier is to connect a capacitor (Ce) to the high impedance (high gain) node of the amplifier. Given the low current consumption of the amplifier, the added capacitor to narrow band the amplifier, also makes the dynamic response of the amplifier slow. The disclosed method to reinvigorate the narrow banded (for lowering the noise of) amplifier, provides a ‘fast boost-on’ current and ‘slow boost-off’ current helps optimize for faster dynamic response.


The embodiment of FIG. 3A utilizes a folded cascode transconductance amplifier (FCTA). This teaching arranges utilization of the same devices with similar device parameters (e.g., PMOSFETs) as inputs, compensation capacitors (e.g., FET capacitors, or normal capacitors), and bias resistors in both FCTA and BLOCK300A, which helps with smoother dynamic (transition) response going into and coming out of boost mode, and more consistent yield to performance specifications over process and operating condition variations.


The scale factors for FETs and current sources (e.g., M314A, M316A, I300A, I301A, I302A, and I304A, M306A-M3108A, M304A-M310A,) can be altered depending on factors such as speed, and power consumption goal, amongst others. For example, with 0.01≥b≥1000000, 0.01≥t≥1000000, for the embodiment of FIG. 3A, the current sources, i=10 nA where b=10 and t=20. Also, in this segment of description the terms: amplifier, FCTA, or BLOCK301A are used interchangeably. The terms: noise reduction plus current boost circuit, current boost circuit, speed boost circuit, or BLOCK300A are also used interchangeably. This disclosure utilizes the current boost method with an amplifier configured as FCTA. However, there are alternative amplifier configurations that can utilize this disclosure's noise reduction plus speed boost method that would be possible. Some examples would be to apply the method of lowering noise plus boosting dynamic response of an amplifier by modifying BLOCK300A to match in accordance with an amplifier with NMOS input stage, or a FCTA with cascoded or regulated cascoded current mirrors, or complementary (PMOS and NMOS) rail-to-rail input stages, or amplifier topologies that are not folded cascode transconductance, amongst others.


The connections of the elements in FIG. 3A are described as follows. The body terminal of all NMOSFETs in FIG. 3A are connected to node 2 that is the VSS, and the body terminals of PMOSFETs are connected to node 1 that is the VDD. The upper terminals of the bias current sources I300A, I302A, and I301A, are connected to node 1 that is VDD. The lower terminal of the bias current source I304A is connected to node 2 that is VSS. The lower terminal of the voltage source V301A is connected to node 2 that is VSS. The ground node (GND) or node 0 is connected to the first terminals of effective capacitances Ce300A and Ce301A. The source terminal of PMOSFETs M314A, M316A, M305A, M307A, M309A, and M311A are connected to node 1 which is the positive supply voltage, VDD. The source terminal of NMOSFETs M304A, M306A, M308A, M317A, M319A, M321A, and M323A are connected to the negative supply voltage, VSS. Node 300A is connected to the gate terminal of M300A and the gate terminal of M301A. Also, node 300A is the VIN+ terminal of the FCTA and BLOCK300A. Node 301A is connected to the gate terminal of M302A and the gate terminal of M303A. Also, node 301A is the VIN− terminal of the FCTA and BLOCK300A. Node 302 is connected to the source terminal of M300A, the source terminal of M302A, and the lower terminal of I300A. Node 303A is connected to the source terminal of M301A, the source terminal of M303A, and the drain terminal of M305A. Node 304A is connected to the drain terminal of M300A, the gate terminal of M304A, the drain terminal of M304A, and the gate terminal of M310A. Node 305A is connected to the drain terminal of M301A, the drain terminal of M319A, and the source terminal of M313A. Node 306A is connected to the drain terminal of M302A, the gate terminal of M306A, the drain terminal of M306A, and the gate terminal of M308A. Node 307A is connected to the drain terminal of M303A, the drain terminal of M321A, and the source terminal of M315A. Node 308A is connected to the drain terminal of M308A, and the source terminal of M310A. Node 309A is connected to the drain terminal of M317A, the gate terminal of M307A, the drain terminal of M307A, and the gate terminal of M305A. Node 310A is connected to the drain terminal of M310A, the gate terminal of M312A, and the lower terminal of current source I302A. Node 311A is connected to the drain terminal of M313A, the gate terminal of M309A, the drain terminal of M309A, and the gate terminal of M311A. Node 312A is connected to the source terminal of M312A, and the upper terminal of current source I304A. Node 313A is the output, VOUT, of the FCTA and is connected to the drain terminal of M311A, the drain terminal of M315A, and the second terminal of effective capacitance Ce301A. Node 314A is connected to the drain terminal of M312A, the gate terminal of M314A, the drain terminal of M314A, the gate terminal of M316A, and the second terminal of effective capacitance Ce300A. Node 315A is connected to the gate terminal of M315A, the gate terminal of M313A, and the positive terminal of voltage source VV301A. Node 316A is connected to the drain terminal of M316A, the drain terminal of M323A, the gate terminal of M323A, the gate terminal of M321A, the gate terminal of M319A, the gate terminal of M317A, and the lower terminal of current source I301A.


The BLOCK300A contains M300A, M302A (configured in CSA), M304A, M306A, M308A, M310A (configured in minimum current selector, MCS300A), I302A, M312A, I304A, M314A, M316A, Ce300A, and Ce301A (configured in providing the ‘boost on’ and ‘boost off’ signal as well as shaping the ‘fast boost-on’ {e.g., slew} current and ‘slow boost-off’ {e.g., slow decay with one-time constant} current that is fed into the FCTA. The FCTA contains M301A, M303A (input stage configured in CSA similar to that of the BLOCK300A), M313A, M315A, (configured in CGA) M309A, M311A (configured in current mirror), M305A, M307A, M317A, M319A, M321A, M323A, V301A, and I301A (configured in the operating current and bias circuitry network for the FCTA).


Describing the details of the circuit in FIG. 3A is as follows to give a general description for more context in how utilizing the noise reduction plus speed boost circuit can improve the amplifier's performance. The VIN is applied to a differential CSA, containing M319A and M321A, has its current outputs feed the next CGA, containing M313A and M315A. The differential outputs of this CGA feed the CM, containing M309A and M311A, to make a single-ended output, VOUT. The bandwidth of the amplifier is approximately











gm

M





301

A


/

Ce

301





A






(

i

V
T


)

/

Ce

M





301





A




)

.





All else equal, increasing the operating current ‘i’, speeds it up and vice versa, and increasing the effective capacitance at the high impedance node of the FCTA at node 313A narrows bands and makes its dynamic response slower and vice versa.


An amplifier's noise generally increases when the amplifier operates with low currents. This disclosure reduces the output noise of the amplifier by narrow banding the amplifier at its high impedance node 313A by increasing Ce301A, while keeping the steady state current consumption at ultra low levels. Because narrow banding the amplifier, slows its dynamic response (speed), then the operating current ‘i’ is boosted dynamically, which is triggered when the amplifier input (VIN) stop tracking each other (and go off balance) after (VIN) receive a large transient signal. In this teaching, the dynamic response of the amplifier, generally speaking, goes through two phases: the ‘slewing time’ or ‘slew rate’ phase and the ‘settling time’ phase. The slew rate (SR∝2i/Ce301A) of FCTA is largely determined by the effective capacitance (Ce301A) at the amplifier's high impedance output node 313, and the operating current of the amplifier, ‘i’, that is scaled as in IDM315A and IDM311A. The settling time of the FCTA, that is distinguished from SR for the purpose of this disclosure, is largely dominated by input stage gmM301A, Ce301A, and the amplifier's output impedance that is largely a function of rds∝VA/i. Because both the gm, ‘i’, and Ce of the noise reduction plus speed boost circuit, BLOCK300A, and that of the amplifier, FCTA, are arranged to be a function of similar device parameters on silicon, therefore the dynamic response of FCTA and BLOCK 300A track each other more consistently over process and operating condition variations.


Although the gmM301A of FCTA and gmM300A of FCTA and BLOCK300A at their input stages are similar, but an aspect of this disclosure is that FCTA and BLOCK300A are arranged to respond differently to their input voltage entering and exiting balance (i.e., in and out of steady state conditions). This aspect of the disclosure, that will be described shortly, helps a smoother and more consistent dynamic response in and out of the speed boost phase, over process and operating condition variations.


First, the steady state phase for FIG. 3A is described. When the amplifier and BLOCK300A inputs are in steady-state, the boost signal remains off. The input stage of BLOCK300A and that of the FCTA are similar, made up of the PMOSFET primary pair (M300A-M302A and M301A-M303A, respectively). The input stage differential pair currents IDM300A≈IM302A≈i are fed onto the minimum current selector, MCS300A.


First, at a high level, lets describe the operations of the minimum current selector (MCS300A) composed of M304A mirrored and scaled with M310A coupled with M306A mirrored and scaled with M308A. When IDM304A≈0→VGSM304A≈0→VGM310A≈0, and thus the output current of MCS300A which is outputted via IDM310A≈0, thus the min(IDM304A, IDM306A)≈IDM304A≈0 is selected. Conversely, when IDM306A≈0→VGSM306A≈0→VGM308A≈0, and thus the output current of MCS300A which is outputted via IDM310A≈0, thus the min(IDM304A, IDM306A)≈IDM306A≈0 is selected. In the steady state phase there is current equilibrium, when IDM304A≈i→VGSM304A≈VGSi→VGM310A≈VGSi. Similarly in equilibrium, when IDM306A≈i→VGSM306A≈VGSi→VGM308A≈VGSi. Given that M310A is scaled at 2× the size of M304A and M308A is scaled at 2× the size of M306A, and that M310A is in series with M310A, and that VGM308A≈VGM310A≈VGSi, therefore the output current of MCS300A which is outputted via IDM310A≈i. Thus, in the steady state phase when IDM304A≈IDM306A≈i, the FETs in MCS300A, including M304A, M306A, M308A, and M310A are in balance and the MCS300A operates accordingly and min(IDM304A, IDM306A)≈IDM304A≈IDM306A≈i.


Therefore, in steady state given that the output of MCS300A generates i≈IDM310A>I302A≈0.5i, then V310A that controls the boost on-off signal, pulls down (V310A≈VSS) and there is the boost-off phase. As a result, VGM312A≈0→IDM312A≈0→IDM314A≈IDM316A≈0. Thus, in steady state phase, the amplifier's bias current is supplied only by I301A, and with IDM316A≈0 amplifier remains in the ‘boost off’ phase.


In this segment, the ‘boost on’ phase is described. When the boost is inactive, IDM323A≈I301A establishes the ultra low quiescent operating current of the amplifier. While the boost is active, BLOCK300A dynamically generates a boost current that is added to IDM323A to boost the FCAT dynamic response. Here is how this happens. When large differential signal transients are applied at VIN, then all or majority of the I300A current 2i would flow through, for example, M302A. Thus, IDM302A≈IDM306A≈2i and IDM300A≈IDM304A≈0<<2i. Therefore, as noted earlier, then MCS300A operates accordingly and min(IDM304, IDM306)≈IDM304A≈0<<2i. In this phase, MCS300A output (flowing through M310A) conduct near or at zero current, and hence V310A is pulled up by I302A (when V310≈VDD) and there is the boost-on phase. In the boost-on phase, all of I304A≈b×i passes through M312A (that functions like a current switch) onto M314A that is mirrored and scaled with M316A. Hence, IDM316A≈b×t×i, which is the boost-on dynamic current that is added onto the amplifier bias network via M316A to M323A. Note that the operating currents in the input stage of FCTA (M301A-M303A) is boosted via the amplifier's bias network that contains M323A-M317A-M307A-M305A. As noted earlier, boosting the operating current in the amplifier's input stage, increases its ‘i’ and gm and thus speeds up the amplifier's dynamic response. Similarly, the operating currents in the high-impedance high-gain stage of FCTA (M313A-M315A-M319A-M311A) is boosted via the amplifier's bias network that contains M323A-M319A-M321A. As noted earlier, boosting the operating current in the amplifier's high impedance stage, increases its ‘i’ and also speeds up the amplifier's dynamic response, including its slew rate.


Before the dynamics of getting in and out of the current boost phase is described, a clarification with respect to Ce301A and Ce301A is in order. From a high level point of view, this disclosure is flexible in how Ce301A and Ce301A are made. For example, Ce301A and Ce301A can be made of similar devices such as standard capacitors, FET capacitors, or relying on Ce associated with the intrinsic capacitances coupled to the 313A node of amplifier and the 314A node of BLOCK300A. The dynamic response of the FCTA and BLOCK300A are designed to match by using the same type of devices that impact their dynamic response, including using similar Ce and running the amplifier and the noise reduction (plus speed boost) circuits with similar bias currents that track each other over process and operating variations. Matching the dynamic response of FCTA and BLOCK300A helps with smooth transient response in and out of boost plus consistency of dynamic response across process and operating condition variations. The Ce301A is the effective capacitance at the high gain high impedance VOUT node 313A of the amplifier that low pass filter filters the noise. This Ce301A can be made with FETs or regular capacitors available in standard CMOS fabrication processes. Part of the Ce301A can also be contributed by the intrinsic input capacitance associated with, for example, a buffer transistor drivers that could be coupled to the main FCTA's VOUT node 313A. Similarly, the Ce300A is the effective capacitance at node 314A, in BLOCK300A, that contributes to shaping the transient profile of the ‘fast boost-on’ current and ‘slow boost-off’ current that is injected into the amplifier. This Ce300A capacitance can be made with FETs (e.g., PMOSFET capacitors), or regular capacitors available in standard CMOS fabrication processes. Part or all of the effective capacitance, Ce300A, can also be contributed by the intrinsic input capacitance associated with, for example, CeM314A and CeM316A.


Next the dynamics of getting in and out of current boost phase is described. When the boost-on signal is triggered, I304A≈b×i slews the node 314A. The ‘boost-on’ signal at node 310A turns on M312A, which acts like a current switch, which causes I304A to provide a ‘fast boost-on’ current onto M314A that is mirrored and scaled up onto M316A. Note that the speed at which node 314A slews (i.e., the dynamic profile of ‘fast boost-on’ current) is also a function of Ce314A, beside I304A.


For a better perspective about the dynamic response, note that while the ‘fast boost-on’ current is still feeding the FCTA's current bias network, the FCTA's SR remains boosted. At some point when the amplifier's inputs approach balance, BLOCK300A triggers the boost off signal (bringing down V310A to VSS) which shuts off the current switch MM312A. Here, the flow of I304A is closed off, and the amplifier enters the ‘boost-off’ phase when a decaying current continues feeding the FCTA bias network. When the boost off signal is triggered and I304A is cut off from M314A, the profile of ‘slow boost-off’ current is a function of the Ze314A and Ce300A, where Ze314A∝1/gmM314A, which will be described further shortly.


An aspect of this disclosure is that the boost-off signals is generated responding to different levels of inputs equilibrium, although both FCTA and BLOCK300A are receiving and mentoring the same input voltage. The BLOCK300A in this disclosure is arranged such that when its inputs can be coarsely equalized, it triggers the boost-off signal. BLOCK300A can be arranged such that its inputs can be coarsely equalized in different ways. Some example are for BLOCK300A to have lower gain than the main amplifier, or having the MSC300A trigger node 310A when its currents (e.g., M304A and M306A) are coarsely equalized instead of accurately, or inserting a hysteresis in the inputs stage of BLOCK300A, or combination thereof, amongst other means. Another aspect of this disclosure is that after the boost off signal is triggered in BLOCK300A, subsequently there continues to be a slow decay current (thus the term ‘slow boost-off’ current) that feeds the amplifier's operating bias current network, until the slow boost-off current dies off. This arrangement improves the FCTA settling time. The aforementioned attributes (boost off signal triggered when inputs are equalized coarsely, plus the slow-boost off current) improve the amplifier's settling time. As noted earlier, the ‘slow boost-off’ current decay rate is roughly determined by node 314A's time constant set that is set approximately by the equivalent capacitance, Ce314A, and impedance, Ze314A, at node 314A. Note that in this embodiment Ce314A is dominated by effective capacitance CeM314A and CeM316A, and Ze314A is dominated by approximately 1/gmM314A computed in the saturation region, in light of the boosted transition currents from subthreshold to saturation region during the intermittent boosted phase. As stated earlier, this embodiment enables the SR and settling time of both the amplifier and BLOCK300A (which provides and shapes the dynamic response profile of the ‘fast boost on’ and ‘slow boost-off’ currents for the FCTA) to approximately track and match each other. As such, this trait benefits the amplifier with smooth dynamics response (in and out of boost phases) and its consistency over process and operating condition variations.


Comparative simulations in FIG. 3C. depicts improvement in FIG. 3A amplifier's output noise. Based on approximate device models, by utilizing the noise reduction and current boost circuit, the VOnoise of the amplifier at 10 Hz can be roughly 4 μV/√{square root over (Hz)} compared to roughly 8 mV/√{square root over (Hz)} for the comparable amplifier that does not utilize the noise reduction and current boost circuit. Moreover, FIG. 300D depicts improvement in the FIG. 3A amplifiers settling time (τS) with the amplifier utilizing the noise reduction and current boost circuit can be roughly 8 μs compared to μs 110 for the comparable amplifier that does not utilize the noise reduction and current boost circuit.


In conclusion, here is a summary of some of the improvements to an amplifiers performance that can be attained by this disclosure. First the amplifier's output noise is reduced by narrow banding the amplifier. Second, the static operating current consumption of the amplifier is kept ultra low. Third, the slowed dynamic response of the narrow-banded amplifier (both slew rate and settling time) are reinvigorated and boosted. Fourth, lower noise, and most other attributes of the amplifier including its gain, bandwidth, static power consumption, common mode range, PSRR, and CMRR are generally not affected by the boost stage, since boost is only engaged dynamically when inputs experience large differential signals. Fifth, amplifier's input structure and that of the boost stage are substantially similar, and hence the boosting function accommodates the full common mode range and power supply span. Sixth, as noted earlier, the AC, slew rate, and transient profiles of the amplifier and that of the boost circuit of BLOCK300A should approximately track each other over temperature and process variations. This is because generally, the amplifier and the boost stage's operating currents, gain (e.g., ∝VA/VT2), and input's 1/gm (e.g., ∝VT/i) roughly track each other, as do their poles in BLOCK300A and FCTA are roughly a function of similar ∝1/gm, and ∝CeFET (or standard capacitors), and ∝ro≈VA/i given that they are made of similar device parameters for both BLOCK300A and FCTA. Seventh, the switching threshold of the amplifier's input stage from small signal to large signal needs to be large enough (e.g., offset mismatch between the boost and the amplifier input stages, ΔVOFS) so that a false or premature turning on of the boost function is avoided. This is the case considering that in steady state (before boost-on phase), when the VIN imbalance is first detected, the amplifier and boost stage input's 1/gmPMOS where i≈10 s of nano-amperes. Note that the BLOCK300A operating currents remain ultra-low, while the main amplifier currents increase substantially in the boost-on phase to the micro-ampere range, which take the amplifier FETs out of the subthreshold region. This is a transitory change in the gain as well as the dynamic response of the amplifier while in the boost-on phase, compared to BLOCK300A (that stays in subthreshold region) while monitoring and receiving the same input voltage as the amplifier. Note also that, it is possible to provide some hysteresis at the input of BLOCK300A as guard-band against unwanted boost signal toggles. Eighth, to arrange for the boost-off signal to be triggered (setting in motion the boost current decay) when the BLOCK300A inputs are coarsely equalized (as compared to the amplifier's inputs whose inputs continue to converge towards finer balance and finer equalization), the amplifier's settling time is improved. Ninth, the maximum boost current, b×t×i, is fixed and is proportional to amplifier's static bias current, ‘i’, which helps control peak dynamic current consumption. This trait also facilitates the boost stage's peak speed to tracks that of the main amplifier, over process, temperature, and operating variations. Note that ‘i’ can be made independent of VTH and mostly a function of VT and μPMOS which are more tightly controlled in manufacturing, and helps with consistency of performance specification across manufacturing variations. Tenth, care is taken to minimize dependence of amplifier's specifications on multiple device parameters such as NMOSFET, VTH NMOS/PMOS, and N+/P+ resistors. Instead, amplifier's specifications mostly rely on PMOSFETs which dominate pertinent signal paths, and this can help optimize yield and help lower noise (e.g., PMOSFET 1/f noise<<NMOSFET 1/f noise) further.


Section XI: Detailed Description of Amplifier Illustrated in FIG. 3B, Utilizing a Second Noise Reduction Plus Speed Boost Circuit



FIG. 3B is a circuit schematic showing a amplifier utilizing the noise reduction plus speed boost circuit (BLOCK300B). The embodiment illustrated in circuit schematic section in BLOCK300B (of FIG. 3B) utilizes a Loser Take All LTA300B circuit (compared to the MCS300A that was utilized in BLOCK300A of FIG. 3A). All else is substantially similar and the description of the disclosure is interchangeably applicable between FIG. 3B and that of FIG. 3A. Beyond what is already described in the previous section, the description here provides the description of elements, and detailed explanation of the LTA300B, with a summary in its conclusion.


Similar to the teaching in previous section, the disclosure here is a method of lowering the output noise of an amplifier by band passing it while keeping its static current consumption ultra low, and concurrently speeding up the amplifier's dynamic response, by boosting the amplifier's operating current in the face of the amplifier's inputs receiving a large transient signal. The noise reduction and speed boost circuit of BLOCK300B utilizes a LTA300B to detect when inputs are imbalanced. Similar to FIG. 3A, the circuit of FIG. 3B speeds up the amplifier's dynamic response by boosting both its SR as well as its' settling time. The scale factors for FETs and current sources (e.g., M304B, M310B, M314B, M316B, I300B, I301B, I302B, and I304B) can be altered depending on factors such as speed, power consumption goal, amongst others. For example, with 0.01≤b≤1000000, 0.01≤t≤1000000, for an embodiment of FIG. 3B, the current sources, i=10 nA where b=10, and t=20. This disclosure utilizes the current boost method with an amplifier configured as FCTA. However, there are alternative amplifier configurations that can utilize this disclosure's noise reduction plus speed boost method that would be possible. Some examples would be to apply the method of lowering noise plus boosting dynamic response of an amplifier by modifying BLOCK300B to match in accordance with an amplifier with NMOS input stage, or a FCTA with cascoded or regulated cascoded current mirrors, or complementary (PMOS and NMOS) rail-to-rail input stages, or amplifier topologies that are not FCTA, amongst others. It would be also be possible to utilize variations of MCS300A or LTA300B or combination thereof to detect an imbalance at the input of an amplifier, including for example utilizing Winner Takes All (WTA) function, amongst others.


The connections of the elements in FIG. 3B are described as follows. The body terminal of all NMOSFETs in FIG. 3B are connected to node 2 that is the VSS, and the body terminals of PMOSFETs are connected to node 1 that is the VDD. The upper terminals of the bias current sources I300B, I302B, and I301B, are connected to node 1 that is VDD. The lower terminal of the bias current source I304B is connected to node 2 that is VSS. The lower terminal of the voltage source V301B is connected to node 2 that is VSS. The ground node (GND) or node 0 is connected to the first terminals of effective capacitances Ce300B and Ce301B. The source terminal of PMOSFETs M314B, M316B, M305B, M307B, M309B, and M311B are connected to node 1 which is the positive supply voltage, VDD. The source terminal of NMOSFETs M304B, M304B′, M306B, M306B′, M308B, M310B, M317B, M319B, M321B, and M323B are connected to the negative supply voltage, VSS, which is node 2. Node 300B is connected to the gate terminal of M300B and the gate terminal of M301B. Also, node 300B is the VIN+ terminal of the FCTA and BLOCK300B. Node 301B is connected to the gate terminal of M302B, the gate terminal of M302B′, and the gate terminal of M303B. Also, node 301B is the VIN− terminal of the FCTA and BLOCK300B. Node 302 is connected to the source terminal of M300B, the source terminal of M302B, the source terminal of MM302B′, and the lower terminal of I300B. Node 303B is connected to the source terminal of MM301B, the source terminal of M303B, and the drain terminal of M305B. Node 304B is connected to the drain terminal of M300B, the gate terminal of M304B, the drain terminal of M304B, and the gate terminal of M304B′. Node 305B is connected to the drain terminal of M301B, the drain terminal of M319B, and the source terminal of M313B. Node 306B is connected to the drain terminal of M302B, the drain terminal of M304B′, the gate terminal of M306B, the drain terminal of M306B, and the gate terminal of M306B′. Node 307B is connected to the drain terminal of M303B, the drain terminal of M321B, and the source terminal of M315B. Node 308B is connected to the drain terminal of M306B′, the drain terminal of M308B, the gate terminal of M308B, and the gate terminal of M310B. Node 309B is connected to the drain terminal of M317B, the gate terminal of M307B, the drain terminal of M307B, and the gate terminal of M305B. Node 310B is connected to the drain terminal of MM310B, the gate terminal of M312B, and the lower terminal of current source I302B. Node 311B is connected to the drain terminal of M313B, the gate terminal of M309B, the drain terminal of M309B, and the gate terminal of M311B. Node 312B is connected to the source terminal of M312B, and the upper terminal of current source 1304B. Node 313B is the output, VOUT, of the FCTA and is connected to the drain terminal of M311B, the drain terminal of M315B, and the second terminal of effective capacitance Ce301B. Node 314B is connected to the drain terminal of M312B, the gate terminal of MM314B, the drain terminal of M314B, the gate terminal of M316B, and the second terminal of effective capacitance Ce300B. Node 315B is connected to the gate terminal of M315B, the gate terminal of M313B, and the positive terminal of voltage source V301B. Node 316B is connected to the drain terminal of M316B, the drain terminal of M323B, the gate terminal of M323B, the gate terminal of M321B, the gate terminal of M319B, the gate terminal of M317B, and the lower terminal of current source I301B.


The BLOCK300B contains M300B, M302B, and M302B (configured in CSA), M304B, M304B′, M306B, M306B′, M308B, M310B (configured in Looser Take All, LTA300B), I302B, M312B, I304B, M314B, M316B, Ce300B, and Ce301B (configured in providing the ‘boost on’ and ‘boost off’ signal as well as shaping the ‘fast boost-on’ (e.g., slew) current and ‘slow boost-off’ {e.g., slow decay with one-time constant} current that is fed into the FCTA. The FCTA contains M301B, M303B (input stage configured in CSA similar to that of the BLOCK300B), M313B, M315B, (configured in CGA) M309B, M311B (configured in current mirror), M305B, M307B, M317B, M319B, M321B, M323B, V301B, and I301B (configured in the operating current and bias network circuitry for the FCTA).


Here, description of the operations Loser Take All, LTA300B (which is similar in function to that of prior section regarding MCS300A) is provided, ignoring non-idealities such as mismatches. The boundary conditions are described first. When large signals are applied to VIN and input FETs M300B-M302B-M302B′ become imbalanced. First, the case is described when all of I300B=4×i flows through M300B, and IDM302B≈IDM302B′≈0<<4×i. Therefore, IDM304B≈2×IDM304B′≈4×i. Given that IDM302B≈0 and IDM304B′≈2×i→V306B≈0→VGSM306B≈0→IDM306B≈IDM306B′≈0. Given that IDM302B′≈0 and IDM306B′≈0→IDM308B≈0≈IDM310B. Thus the output current of LTA300B is outputted via IDM310B≈0, thus the loser(IDM300B, IDM302B&B′)≈IDM302B&B′≈0 takes all. Note that IDM302B&B′ denotes IDM302B, IDM302B′ that are equal (having the same VGS) whose sum is 2i during steady state.


Next, considering the case where the large signal VIN causes an imbalance such that all of I300B=4×i flows through M302B and M302B′. Here, IDM300B≈0<<4×i and IDM302B≈IDM302B′≈2×i. Therefore, IDM304B≈2×ID′M304B≈0<<4×i. With IDM302B≈2×i and IDM304B′≈0→IDM306B≈2×i≈IDM306B′. With IDM302B′≈2×i and IDM306B′≈2×i, therefore no current is left for M308B. Thus, ID308B≈0≈IDM310B. Therefore the output current of LTA300B is outputted via IDM310B≈0, thus the loser(IDM300B, IDM302B&B′)≈IDM300B≈0 takes all.


In summary, when there is an imbalance (in either direction) as a result of a large signal VIN, the LTA300B generates a zero current through IDM310B≈0. As such I302B=1×i pulls up on node 310B (V310B≈VDD)→M312B turns on and passes I304B=b×i onto M314B whose current is mirrored and scaled up onto M316B with IDM314≈b×t×i . This IDM314B is the boost-on current that is added to the FCTA quiescent current, I304B, to feed the FCTA current bias network through M323B.


Next considering the steady state, or static, conditions when the amplifier is regulated its input FETs M300B-M302B-M302B′ are in balance. In steady state, IDM300B≈2×IDM302B≈2×IDM302B′≈2×i. Therefore, ID304B≈2×IDM304B′≈2×i and IDM302B≈i. Therefore, there is no current left for M306B→IDM306B≈IDM306B′≈0. With IDM306B′≈0, then all of IDM302B′≈i flows through M308B which is mirrored and scaled up 2× into M310B and thus 2×IDM308B≈IDM310B≈2×i. Thus the output current of LTA300B is outputted via IDM310B≈2×i since there are no loser, but equals per se here, the loser(IDM300B, IDM302B&B′)≈IDM300B≈ID′M302B&B≈2×i takes all. Therefore, when the amplifier's inputs arrive at balance in steady state conditions, the output current of LTA300B generates a IDM310B≈2×i (½ the tail current of the input stage). As such, IDM310B pulls down on I302B=1×i which takes node 310B down (V310B≈VSS)→M312B turns off and blocks I304B=b×i from flowing into M314B. Thus, IDM314B≈0≈IDM316B. As a result, the ultra low static FCTA quiescent current, I304B, feeds M323B in steady state conditions with no boost current.


In summary, FIG. 3B is a circuit schematic based on a method of reducing an amplifier's output noise by narrow-banding the amplifier, which slows its dynamic response, while independently reinvigorating the amplifier's narrow-banded dynamic response via boosting both its' slew rate and settling time. This is accomplished as follows. PMOSFET or standard capacitors can be used at the high impedance node to low pass filter the amplifier's output noise. A loser take all (LTA) circuit is utilized to detect and trigger a boost on and boost off signal, when the amplifier's inputs approximately go in and out of balance. A boost function is enabled, in response to large differential transient input signals, by injecting a dynamic current into the amplifiers pre-existing static bias current network, which at first rapidly speeds up the amplifier's slew rate. When inputs approximately approach balance, the boost signal is triggered off. But here, a slow and decaying current continues feeding the amplifier's bias current network, until the decaying current fades off. This decaying current (which is initially sizable enough when compared to the otherwise statically ultra low current operant in the un-boosted mode) roughly follows a single pole trajectory, which helps speed up the amplifier's settling time.


As explained earlier, while reinvigorating the SR and settling time, the price to pay for lowering the amplifier's noise, is the increased dynamic current consumption (intermittent power consumption). After the amplifier's inputs receive a large signal transient, and the boost-on signal is triggered, the ‘fast boost-on’ current kick starts speeding up the amplifier's dynamic response by boosting up its slew rate, which moves the amplifier back on the path towards balancing its inputs ahead of boost-off signal trigger. To save on the intermittent power consumption, the boost off signal can be triggered sooner which cuts off the slewing current boost sooner. However, a ‘slow boost-off’ arrangement continues injecting a dynamic decaying current into the amplifier bias network (which eventually levels off to zero) to speed up the amplifier's settling time. This is accomplished, in part, as explained previously above by arranging BLOCK300B to initiate the boost-off signal when its inputs are not equalized in the same size and fashion as that of the inputs of FCTA (e.g., the amplifier has approached but not fully regulated).


The whole amplifier operates in the subthreshold region when it is un-boosted, but its' input, transconductance gain, and (would be) output buffer stages would become faster by receiving the boosted dynamic current and transition in and out of saturation during the intermittent boosted phases. It is also of note that PMOSFETs can be used at input stages and also as active resistors to set the operating bias currents for the whole amplifier (including, input stages, gain stage, the boost circuit, and output stage), which can establish both their input stage's transconductance (gmPMOS). Moreover, PMOSFET or normal capacitors can be utilized to set the dominant poles of each of the amplifier gain stage and that of the boost stage. In other words, dynamic response of both the FCTA and BLOCK300B can be largely proportional to the same (operating current) ‘i’ and (dominant effective capacitance) ‘Ce’, and same kind of device parameters. Hence, the dynamic response in and out of boost track each other, and follow a reasonably smooth and stable passage, in and out of boost phases. Moreover, the profiles of the amplifier and the boost stage dynamic response can be more consistent from lot-to-lot in manufacturing, across operating variations.


Section (XII): Describing a Prior Art Amplifier Gain Stage Illustrated in FIG. 4A, Coupled with (Inverting) Buffer Driver


Generally a buffer amplifier, as depicted in FIG. 5, contains an amplifier section coupled with a buffer driver section. The amplifier section amplifies the signals applied to its inputs, and has high input impedance and high output impedance and is sometimes referred to gain stage. Buffer driver section has high input impedance and low output impedance with its outputs capable of sinking and sourcing currents to drive external loads, such as resistors, capacitors, inductors or a combination thereof. Combination of amplifier and buffer driver makes a buffer amplifier that can amplify signals applied to its inputs. A buffer driver also has high input impedance, low output impedance, and its output is capable of sinking and sourcing currents to drive external loads. Therefore, there may be formed a buffer amplifier (AMP400) comprised of a amplifier gain stage of FIG. 4A, coupled with a buffer driver (e.g., inverting buffer driver). The (inverting) buffer drivers can be anyone of the disclosed BUF400B, BUF400C, and BUF400D embodiments illustrated in FIG. 4B, FIG. 4C, or FIG. 4D (respectively), which will be discussed in the next sections. Note that AMP400B is formed by amplifier gain stage of FIG. 4A coupled with BUF400B. The AMP400C is formed by amplifier gain stage of FIG. 4A coupled with BUF400C. The AMP400D is formed by amplifier gain stage of FIG. 4A coupled with BUF400D. The connection of the elements in FIG. 5 are described as follows. Node 500A is the negative input terminal of the Buffer Amplifier that is connected to the positive input terminal of the Amplifier (that provides the gain stage). Node 501A is the positive input terminal of the Buffer Amplifier that is connected to the negative input terminal of the Amplifier (that provides the gain stage). Node 502A is the output terminal of the Amplifier (that provides the gain stage) that is connected to the input terminal of the (inverting) buffer driver (that provides current drive capability to an external load). Node 503A is the output of the Buffer Driver that is also the output of the Buffer Amplifier (that provides the Buffer Amplifier current drive capability to an external load). Note that it would be obvious for one in the art to modify the disclosed teaching wherein the output of the Amplifier can be composed of 2 outputs (instead of one output at node 502A) that would feed 2 inputs of a Buffer Driver (instead of one input to the Buffer Driver). Moreover, note that the designation ‘−G’ denotes the inverting aspect of the disclosed preferred embodiment of the Buffer Driver. It would be obvious to one skilled in the art to utilize a variation of the disclosed Buffer Driver, that is non-inverting, coupled with an Amplifier (that provides the gain stage) whose inputs are re-arranged in order to keep the signal signs in proper order in making the Buffer Amplifier


As such, the amplifier gain stage's output node 414A (vo1400A) of FIG. 4A can connect to the buffer drivers vin1's, which can be the input node 414B of BUF400B, or input node 414C of BUF400C, or input nodes 414D of BUF400D. Accordingly, gain output node 415A (vo1400A) of FIG. 4A can connect to buffer driver vin2's, which can be the input node 415B of BUF400B, or input node 415C of BUF400C, or input nodes 415D of BUF400D, respectively. Also, note that either one or both of the inputs of (inverting) buffer drivers, can be current input terminals.


In the next sections of this disclosure, AMP400B, AMP400C, or AMP400D are configured in unity gain. In unity gain configuration, the output of BUF400B which is Vo400B, or output of BUF400C which is Vo400C, or output of BUF400D which is Vo400D connect to the negative input terminal node 412A or vin− of the amplifier gain stage of FIG. 4A. To explain how an amplifier's gain stage interacts with an (inverting) buffer driver, first assume an AMP400 which is comprised of amplifier gain stage of FIG. 4A coupled with an ideal (inverting) buffer driver, and describe both the steady state and non-steady state of amplifier gain stage of FIG. 4A. In steady state conditions, and assuming no non-idealities or mismatch, a AMP400 configured in unity gain (node 412A or vin− is connected to the output of an ideal inverting buffer driver) would have vin−=vin+ (i.e., Δvin=0). In the input stage of FIG. 4A, there is current equilibrium with IDM409A≈IDM411A≈0.5×IDM407A≈i and IDM408A≈IDM410A≈0.5×IDM406A. In the upper and lower current mirrors, there is also current equilibrium, with IDM403A≈IDM413A≈IDM415A≈2i and IDM402A≈IDM412A≈IDM414A≈2i. Accordingly, (IDM416A≈IDM418≈i)≈(IDM417A≈IDM419≈i). Second, assume AMP400 is also in unity gain configuration, but it is initially in non-steady-state condition, for example vin−<vin+ (albeit a very small finite −Δvin). Here, (IDM410A≈IDM408A)<(IDM409A≈IDM411A) and IDM406A>IDM407A. At node 405A, the increased IDM406A is mirrored via IDM403A to increase (IDM413A≈IDM415A). At node 404A, the decreased IDM406A is mirrored via IDM402A to decrease (IDM412A≈IDM414A). With less of (IDM410A≈IDM408A) flowing into nodes 411A and 409A, there will be more of the increased (IDM413A≈IDM415A) available to flow through (IDM417A≈IDM419A). Conversely, with more of (IDM409A≈IDM411A) flowing into nodes 410A and 408A, there will be less of the decreased (IDM412A≈IDM414A) available to flow through (IDM416A≈IDM418). Hence, initially, in non-steady state condition when vin−<vin+→(IDM416A≈IDM418A)<(IDM417A≈IDM419A)→initially both node 414A (Vo1400A) and node 415A (Vo2400A) are pulled down together. Conversely, when vin−>vin+, the opposite dynamics occur and initially both node 414A (Vo1400A) and node 415A (Vo2400A) are pulled up together. Nodes 414A and 415A (are a pair of high-gain high-impedance outputs of the amplifier's gain stage that connect to inputs to the inverting buffer drivers, vin1 and vin2 that) have the same polarity, and move up and down together.


Next, considering the AMP400B, or AMP400c, or AMP400D. Note that steady state conditions may be disturbed indirectly as well. For example, a transitory current imbalance, Δi, at one of the current input node of the (inverting) buffer drivers that may be initiated by sink-source current in output FETs. In the presence of such transitory Δi, the closed loop containing the amplifier gain stage (FIG. 4A) in concert with BUF400B, or BUF400C, or BUF400D could aim to regulate out the Δi until current balance is returned in steady state (e.g., by adjusting the voltages at node 414A and 415A via IDM416A≈IDM418 and IDM417A≈IDM419A). This aspect of the description of AMP400 will be relevant to the disclosures in the next sections.


It would be possible that the buffer driver methodology described in this disclosure would operate with other alternative amplifiers gain stages, beside that of FIG. 4A. Alternative amplifier gain stages to that of FIG. 4A could be NMOS input FCTA, or PMOS input FCTA, FCTA with RGC, FCTA with input stage that contains double PMOS with an NMOS level shifter, FCTA with input stage that contains double PMOS with an NMOS level shifter, or other non-FCTA configurations, amongst others. In each of the next sections, the AMP400B, or AMP400C, or AMP400D are configured in unity gain, and BUF400B, or BUF400C, or BUF400D are described at steady state while sinking and sourcing current for an external load (i.e., sink and source mode). Also, for each of (inverting) buffer drivers, the steady state mode is described with no load.


Section (XIII): Detailed Description of Buffer Driver Comprised of Complementary Non-Inverting Current Mirrors (CNICM), Minimum Current Selector (MCS) and Inverting Current Mirror Amplifier (ICMA), Illustrated in FIG. 4B (4B)



FIG. 4B is a circuit schematic showing a buffer driver (BUF400B) (e.g., inverting buffer driver) comprising of CNICM400B and MCS400B coupled with ICMA400B. This section build on what was already described in prior section. Note that the difference between the BUF400D embodiment illustrated in FIG. 4D (which is described in its respective section), compared with the BUF400B embodiment illustrated here in FIG. 4B is the inclusion of CNICM400B. As such, the detailed description, including benefits are equally applicable between to BUF400D of FIG. 4D and BUF400B of FIG. 4B.


A buffer driver may be required to sink or source large amounts of current when interfacing with an external load. It would be advantageous to reduce the current consumption, in a buffer driver, that is attributed to the function of monitoring and sensing of output FET driver sink-source currents (i.e. M438B, M435B) that varies depending on the size an external load. One of the benefits of CNICM400B here is to help reduce the buffer driver's load-dependent current consumption. Also, CNICM400B helps the buffer driver with having more consistent performance to specifications that are less dependent on the external load variations.


Throughout the description of FIG. 4B, MOSFET scale factors (e.g., ‘s’, ‘k’, or W/L rations for M421B-M435B, M428B-M429B-M431B-M433B, M434B-M436B, etc.) and current source scale factors (e.g., for I405B-I406B-I407B-I408B) can be modified depending on cost-performance goals such as die size, current consumption, speed, and cross over distortion, and others. For the FIG. 4B disclosure, s=k=20 but ‘s’, ‘k’ can have wide ranges 0.01≤s≤1000000, 0.01≤k≤1000000. The connections of the elements in FIG. 4B are described as follows. The body terminal of all NMOSFETs in FIG. 4B are connected to node 2 which is VSS, and the body terminals of PMOSFETs are connected to node 1 which is VDD. For clarification, note that having one current source (e.g., I405B and I406B) with neither the upper nor the lower terminal connected to a power supply (i.e., VDD, VSS), is the equivalent of having two current sources where the first one has one of its terminals connected to VDD and the second one has one of its terminals connected to VSS. Bias current sources I407B, and I408B have their lower terminals connected to node 2, which is VSS. The source terminal of NMOSFETs M421B M423B, M425B, M427B, M429B, M431B, and M435B are connected to node 2 that is VSS. The source terminal of PMOSFETs M420B, M422B, M424B, M426B, M428B, M430B, M432B, M434B, M436B, and M438B are connected to node 1 that is VDD. Node 414B is the vin1400B terminal of the buffer driver, and is connected to the gate terminal of M420B and the gate terminal of M438B. Node 415B is the vin2400B terminal of the buffer driver, and is connected to the gate terminal of M421B, the gate terminal of M435B, the drain terminal of M436B, and the upper terminal of current source I408B. Node 416B is connected to the drain terminal of M420B, the drain terminal of M422B, the gate terminal of M422B, the gate terminal of M424B, and the upper terminal of current source I405B. Node 417B is connected to the drain terminal of M421B, the drain terminal of M423B, the gate terminal of M423B, the gate terminal of M425B, and the lower terminal of current source I405B. Node 418B is connected to the drain terminal of M424B, the drain terminal of M426B, the gate terminal of M426B, the gate terminal of M428B,, and the upper terminal of current source I406B. Node 419B is connected to the drain terminal of M425B, the drain terminal of M427B, the gate terminal of M427B, the gate terminal of M433B, and the lower terminal of current source I406B. Node 420B is connected to the gate terminal of M429B, the drain terminal of M429B, the drain terminal of M428B, and the gate terminal of M431B. Node 421B is connected to the drain terminal of M431B and the source terminal of M433B. Node 422B is connected to the drain terminal of M433B, the gate terminal of M430B, the drain terminal of M430B, and the gate terminal of M432B. Node 423B is connected to the drain terminal of M432B, the gate terminal of M434B, the drain terminal of M434B, and the gate terminal of M436B, and the upper terminal of current source I407B. Node 424B is the Vo400B terminal of the (inverting) buffer driver, and is connected to the drain terminal of M438B and the drain terminal of M435B.


The (inverting) buffer driver or BUF400B is comprised of BLOCK400B which is the output FETs, or FET400B (containing source PMOSFET M438B and sink NMOSFET M435B), plus BLOCK401B which is the inverting current mirror amplifier or ICMA400B (containing PMOSFETs M430B, M432B, M434B, M436B, and current sources I407B, I408B), plus BLOCK402B which is the minimum current selector or MCS400B (containing PMOSFET MM428B, and NMOSFETs M429B, M431B, and M433B), plus BLOCK403B which is the complementary non-inverting current mirror or CNICM400B (containing on PMOSFETs side M420B, M422B, M424B, M426B, and on the NMOSFET side M421B, M423B, M425B, M427B, and current sources I405B and I406B). As noted earlier I405B can be is the functional equivalent of 2 current sources, and is shown as one current source for demonstrative simplicity. There can be I405B1 and I405B2 where upper terminal of I405B1 is connected to VDD and lower terminal of I405B2 is connected to VSS. As such, lower terminal of I405B1 and upper terminal of I405B2 can be the functional equivalence of the lower and upper terminals of I405B. As stated earlier, AMP400B is comprised of the amplifier gain stage of FIG. 4A coupled with BUF400B, where node 414A is connected to node 414B, node 415A is connected to node 415B. AMP400B is configured in unity gain with node 412A (the amplifier gain stage negative input terminal) connected to node 420B (or Vo400B).


Note that it would be possible to utilize a ICMA that is a complementary version of ICMA400B (i.e., PMOSFET based or NMOSFET based). It would also be possible to utilize an MCS that is a complementary version of MCS400B (i.e., PMOSFET base or NMOSFETs based). It would also be possible that a plurality of MCS coupled with a plurality of MCSs can be utilized, which are complementary to one another (e.g., NMSOFET based ICMA coupled with primarily PMOSFET based MCS, utilized along with PMSOFET based ICMA coupled with primarily NMOSFET based MCS).


Moreover, it would be possible to design a buffer driver that utilize other equivalent functional implementations that are complementary non-inverting current mirrors (CNICM), minimum current selector (MCS) or inverting current mirror amplifier (ICMA). For example, a loser take all (LTA) can perform the equivalent function of MCS. Another example would be to utilize, a current mirror amplifier (CMA) instead of an non-inverting current mirror amplifier (ICMA) by arranging the proper circuit signal signs. Another example would be to utilize a current amplifier to perform the equivalent function of a ICMA.


In describing the function of CNICM400B, the non-idealities (e.g., offsets and mismatches) are set aside and the boundary conditions are discussed (when M438B is maximally on and M435B is off, and when M438B is off on and M435B is maximally on). After that, CNICM400B's steady state condition is described (when the quiescent currents in M438B and M435B is equal to i×s, where MOSFET scale factors s=k). The M438B current is scaled down (by a factor of ‘s’) and mirrored by M420B. For clarity of description here, the IDM420B is the input, and IDM426B is the output for PMOSFET side of CNICM400B. If IDM438B≈0→IDM420B≈0→all of I405B≈2i flows onto IDM422B≈2i→IDM424B≈2i. Thus, all of I406B≈2i is taken by M424B→IDM426B≈0, which in effect is the (zero scale) non-inverting current mirror of IDM438B≈0. If IDM438B≈imax→IDM420B≈imax×1/s. Assuming that imax×1/s>2i. Note here that the maximum current consumed in IDM420B is ‘2i’ that substantially lowers VDS of M420B, which substantially curbs the load-dependent source current consumption of BUF400B. Thus, all of I405B≈2i is taken by M420B→IDM422B≈0→IDM424B≈0. Thus, all of I406B≈2i flows into M426B→IDM426B≈2i, which in effect is the rectified non-inverting current mirror of IDM438B≈imax. Note that the maximum (full scale) output current possible for M426B is 2i which corresponds to imax of M438B (and its scaled down and mirrored current imax×1/s of M420B). Also note that IDM426B that is monitored by MCS400B (via M428B, M429B, M431B) has its full scale current caped at 2i enabling it to respond more consistently to imax swings imposed on FET400B for sourcing current, for example, for heavy external loads (e.g., imax>>2i×s).


For steady state, let's assume that there is no external load, and that IDM438B≈s×i. Thus, IDM420B≈i→the I405B≈2i is split in half and IDM420B≈IDM422B≈i→IDM424B≈i. Thus, I406B≈2i is also split in half→IDM424B≈IDM426B≈i→IDM426B≈i, which in effect is the scaled down (by 1/s) non-inverting current mirror of IDM438B≈s×i. Similar signal flow applies to the NMOSFET side of CNICM400B, where the M435B current is scaled down (by a factor of ‘k’) and mirrored onto M421B. If IDM435B≈0→IDM421B≈0→all of I405B≈2i flows onto IDM423B≈2i→IDM425B≈2i. Thus, all of I406B≈2i is taken by M425B→IDM427B≈0, which in effect is the (zero scale) non-inverting current mirror of IDM435B≈0. If IDM435B≈imax→IDM421B≈imax×1/k. Let's assume that imax×1/k>2i. Note here that the maximum current consumed in IDM421B is ‘2i’ that substantially lowers VDS of M421B, which substantially curbs the load-dependent sink current consumption of BUF400B. Thus, all of I405B≈2i is taken by M421B→IDM423B≈0→IDM425B≈0. Thus, all of I406B≈2i flows into M427B→IDM427B≈2i, which in effect is the rectified non-inverting current mirror of IDM435B≈imax. Note that the maximum (full scale) output current possible for M427B is 2i which corresponds to imax of M435B (and its scaled down and mirrored current imax×1/k of M421B). Also note that IDM427B that is monitored by MCS400B (via M433B) has its full scale current caped at 2i enabling it to respond more consistently to imax swings imposed on FET400B for sinking current, for example, for heavy external loads (e.g., imax>>2i×k). For steady state, again it is assumed that there is no external load, and that IDM435B≈k×i. Thus, IDM421B≈i→the I405B≈2i is split in half and IDM421B≈IDM423B≈i→IDM425B≈i. Thus, I406B≈2i is also split in half→IDM425B≈IDM427B≈i→IDM427B≈i, which in effect is the scaled down (by 1/k) non-inverting current mirror of IDM435B≈k×i.


In summary, BLOCK403B is CNICM400B which is the complementary non-inverting current mirror. One of the functions of CNICM400B is to monitor and mirror the currents in FET400B, and rectify the full scale (maximal) mirrored currents that feed the MCS400B. Given that the sink-source currents of FET400B can vary significantly, depending on the external load, the scaled down but un-rectified mirroring of FET400B raw currents can also vary a lot, and increase the current consumption of a (inverting) buffer driver. Mirroring and rectifying the sink-source currents of FET400B via CNICM400B, caps the full scale and curbs the maximal mirrored currents (e.g., to ‘2i’) before the sink-source current signals are fed to MCS400B for signal processing.


Therefore, arranging BUF400B comprising CNICM400B coupled with MCS400B coupled with ICMA400B provides the BUF400B with added benefit of lower (load-dependent) sink-source current consumption, as well as more consistent DC, AC, and transient performance, especially under varying external load conditions. As stated earlier, description of BUF400B is provided considering that it is part of AMP400B that is configured in unity gain.


Before describing the BUF400B operating in steady state and over drive (sink-source), the current equilibrium conditions are described at nodes 414A, 414B and 415A, 415B for steady state to persist. At nod 415B in FIG. 4B (which is connected to node 415A in FIG. 4A), KCL requires the sum of IDM416A, IDM417A, I408B, and IDM436B to converge to zero in order for steady state to hold. Also, at node 414B in FIG. 4B (which is connected to node 414A in FIG. 4A), KCL requires the sum of IDM418A, and IDM419A to converge to zero for steady state to hold. Moreover, for AMP400 unity gain loop (containing amplifier gain stage of FIG. 4A and BUF400B) to stay in steady state, (IDM417A≈ID419A)≈(IDM416A≈IDM418A) and I408B≈IDM429B. This is called “the conditions of current equilibrium for nodes 414B and 415B” which will be referred to below.


During steady-state conditions, IDM421B≈IDM427B≈i (with k≈s, and corresponding to IDM435B≈k×i), and IDM420B≈IDM426B≈i (with k≈s, and corresponding to IDM438B≈s×i). Therefore, IDM426B≈IDM428B≈IDM429B≈IDM431B≈i→VGSM427B≈VGSM429B≈VGSM431B. As explained in the previous sections above, with M429B and M427B (both with W/L=1X) operating at current ‘i’ while providing equal gate voltages for M431B and M433B that are in series (both with W/L=2X), causes the operating current of M431B and M433B to also be ‘i’. As such, the MCS400B selects current equality when min(M426B, M427B)≈min(M428B, M433B)≈M433B≈M428B≈i. The IDM433B×i is the output of MCS400B that is fed onto M430B which is the input to ICMA400B. Here, for the current mirror IDM430B≈i≈IDM432B→the current source I407B≈2i is split equally between IDM432B≈i≈IDM434B→the current mirror IDM434B≈IDM436B≈i. The IDM436B≈i is in balance with the current source I408B=i. This is how current equilibrium is held at node 415B, which sustains the controlled quiescent currents in the buffer driver output FETs, IDM438B≈IDM435B≈s×i (assume setting k≈s).


A brief summary is provided here for BUF400B (contained in AMP400B) of FIG. 4B, utilizing CNICM400B coupled with MCS400B coupled with ICMA400B, for the sink and source modes when there is large difference between IDM438B and IDM435B. As a reminder, k=s and IDM435B is rectified by CNICM400B and presented to MCS400B via IDM427B, and IDM438B is rectified by CNICM400B and presented to MCS400B via IDM426B.


In the sink mode, node 414B and node 415B are pulled up towards VDD, MN435B and MN421B are turned on→IDM421B>>i→initially before current equilibrium IDM427B≈2i. Also, MN438B and MN420B are turned off or are nearly off, initially→IDM426B≈ID428B≈IDM429B≈IDM431B≈0 which would block current flow in IDM433B→MCS400B selects the min(IDM427B, IDM426B)=IDM426B≈0<<i, which starves M433B, and initially→IDM432B≈IDM430B<<i→all of I407B=2i flows into M434B→IDM434B≈IDM436B≈2i at node 415B with I408B=i, which in a current imbalance at node 415B initially, and as such the AMP400B loop kicks to bring back the balance. In order for “conditions of current equilibrium for nodes 414B and 415B” to return, and to simplify the explanation for clarity of description, in FIG. 4A (IDM417A≈IDM419A) can get very slightly increased above (IDM416A≈IDM418A). Here, by AMP400B loop regulation, the voltages at both node 414B and node 415B are pulled down sufficiently enough until M420B (and its mirror M438B) turn back on enough for IDM420B≈IDM426B≈i→IDM428B≈IDM429B≈IDM431B≈i→by moving towards current balance, MCS400B inputs are regulated and as such min(IDM426B, IDM427B)≈IDM426B≈i. Such is the case since M435B is still sinking current, IDM435B>IDM438B→IDM421B>IDM420B→IDM427B>IDM426B. As explained in the previous section, note that in FIG. 4A, despite the very slight increase in (IDM417A≈IDM419A) above (IDM416A≈IDM418A) to regulate down the voltages at node 414B and 415B, ICMA400B (coupled and being fed via MCS400B and CNICM400B) have enough gain to regulate a slightly increased IDM436B (as compared to the fixed I408B≈i) to keep the voltage at node 415B high enough so that M435B can continue sinking high currents to the external load. Here, the “conditions of current equilibrium for nodes 414B and 415B” are held, and steady state conditions maintained, while M426D to continue sourcing current. Accordingly, IDM420B≈i→the operating current in the inactive source-FET is thus controlled and regulated at IDM431B≈s×i, while steady state condition is maintained and the sink FET, M435B, continues sinking extra current to an external load.


Conversely, in the source mode, when node 414B and node 415B of FIG. 4B are pushed down towards VSS, initially M438B and M420B turn on and M435B and M421B turn off→then initially before equilibrium, IDM426B≈IDM428B≈IDM429B≈IDM431B≈2i. Moreover, IDM427B≈0→VGM427B=0=VGM433B→M433B blocks flow of IDM431B→thus, before equilibrium, MCS400B selects the min (IDM427B, IDM426B)≈IDM426B≈0→output of MCS400B is IDM433B≈0 that is fed onto M430B which is the input of ICMA400B. IDM430B≈0≈IDM432B→all of current in I407B≈2i, before equilibrium, could flows in the current mirror IDM434B≈IDM436B≈2i>I408B≈1i, which creates a current imbalance at node 415B that is unsustainable and as such ICMA400B contained within the AMP400D loop kicks in to regulate back towards current equilibrium. In order for “conditions of current equilibrium for nodes 414B and 415D” to hold, and to simplify the explanation for clarity of description, (IDM416A≈IDM418A)≈(IDM417A≈IDM419A). Concurrently, voltage at node 415B gets regulated by being pulled up by IDM436B until IDM436B≈i=I408B. In order to arrive at this current equilibrium, ICMA400B along with AMP400B loop regulate the voltage of node 415B down sufficiently enough until IDM436B≈i=I408B→IDM421B≈IDM427B≈i. By moving towards current balance, MCS400B inputs are regulated and as such min (IDM426B, IDM427B)≈IDM427B≈i. Such is the case because M438B is still sourcing high currents to for an external load→IDM435B<IDM438B→IDM421B<IDM420B→IDM427B<IDM426B. As such, when the buffer driver sources current for an output load, then M435B which is the inactive sink FET, can operate at a controlled regulated current IDM435B≈k×i. Again, note that ‘i’ is a constant current source set by the main bias network of the amplifier, and ‘k’ is FET aspect ratio that is tightly controllable, which helps with the control of the quiescent current in the inactive sink FET.


In summary, when there is an imbalance between sink-source FETs currents, IDM438B and IDM435B, such imbalance is scaled down, mirrored, and rectified by the complementary non-inverting current mirror (CNICM400B). The outputs of CNICM400B, IDM427B and IDM426B, which are the rectified version of sink-source currents, are then fed onto the MCS400B. The MCS400B effectively selects the (scaled) minimum current between the output drive currents IDM438B and IDM435B. The selected minimum current, at the output of MCS400B, is then inputted onto ICMA400B which, together with the operations of amplifier's gain stage, regulates a minimum quiescent current in the inactive FET (e.g. for s=k, when IDM435B is sinking current for an external load→inactive source FET IDM438B=s×i, and when IDM438B is sourcing current for an external load→inactive sink FET IDM435B=k×i).


In conclusion, as noted in the previous section, the benefits of the FIG. 4D's BUF400D are equally applicable to that of FIG. 4B's BUF400B (comprising CNICM couple with MCS coupled with ICMA) which are briefly summarized here again. First, the function of ‘monitoring’, of sink-source currents, consume current itself. Sink-source currents can be high and have unpredictable patterns. The CNICM effectively curbs (e.g., rectify) the ‘monitoring’ current consumption. Also the rectifying of output FET currents, before they are fed to the MCS and ICMA, helps with consistency of DC, AC, and transient performance of the buffer driver, with more independence from external load variations. Second, the buffer driver is fast since the CNICM couple with MCS coupled with ICMA are all in principal operating in current mode, which is inherently fast. Third, it can operate with low power supply voltage since the minimum VDD is VGS+2VDS (low voltage) for CNICM, MCS, and ICMA. Fourth, it is made of a few transistor which makes it small and low cost. Fifth, buffer driver can operate with ultra low current and ultra low power, and utilization of CNICM helps curb the current consumption attributed with monitoring the sink-source currents when the amplifier drives different external loads. Sixth, it has wide input-output span of near rail-to-rail. Seventh, it can provide high sink-source drive capability and controlled quiescent current in the inactive output FET that benefits DC and dynamic performance. FIG. 4E is a simulation of a buffer amplifier containing buffer driver of FIG. 4B coupled with gain stage of FIG. 4A. This simulations indicates approximate and typical sink-source current capability for the buffer amplifier to drive external load resistors 900KΩ, 300KΩ, and 30KΩ. The simulation indicates current consumption for the buffer amplifier of roughly 120 nA for FIG. 4B coupled with gain stage of FIG. 4A. Eighth, buffer driver is minimally imposing on the amplifier's gain or the speed of the preceding amplifier's high-gain stage Ninth, it's based on standard CMOS process which is low cost, high quality, ready available). Tenth, the buffer driver arrangement utilizing CNICM coupled with MCS coupled with ICMA can be tailored to very high current and very high speed.


Section (XIV): Detailed Description of Buffer Driver Comprised of Complementary Non-Inverting Current Mirrors (CNICM), Loser Take All (LTA) and Current Mirror Amplifier (CMA), Illustrated in FIG. 4C (4C)



FIG. 4C is a circuit schematic showing a (inverting) buffer driver (BUF400C) comprising of CNICM400C and LTA400C coupled with CMA400C. This section builds on what was already described in previous section. A difference between the BUF400C embodiment illustrated in FIG. 4B in the previous section compared to the BUF400C embodiment illustrated in FIG. 4C is the utilization of LTA400C which is functionally equivalent to MCS400B. As such, the detailed description (e.g., regarding AMP400B, CNICM400B, CMA400B) that were provided in prior section pertaining to BUF400B of FIG. 4B are equally applicable to that of BUF400C of FIG. 4C.


Throughout the description of FIG. 4C, MOSFET scale factors (e.g., ‘s’, ‘k’, or W/L rations for M421C-M439C, M432C-M420C, M426C-M428C, M427C-M429C, etc.) and current source scale factors (e.g., for I405C-I406C-I407C) can be modified depending on cost-performance goals such as die size, current consumption, speed, and cross over distortion, and others. For example, the FIG. 4C disclosure, s=k=20 but ‘s’, ‘k’ can have wide ranges 0.01≤s≤1000000, 0.01≤k≤1000000.


The connections of the elements in FIG. 4C are described as follows. The body terminal of all NMOSFETs in FIG. 4C are connected to node 2 which is VSS, and the body terminals of PMOSFETs are connected to node 1 which is VDD. Bias current sources I407C upper terminals connected to node 1, which is VDD. The source terminal of NMOSFETs M421C M423C, M425C, M427C, M429C, M431C, M433C, M435C, M437C, and M439C are connected to node 2 that is VSS. The source terminal of PMOSFETs M420C, M422C, M424C, M426C, M428C, M430C, and M432C are connected to node 1 that is VDD. Node 414C is the vin1400C terminal of the buffer driver, and is connected to the gate terminal of M420C and the gate terminal of M432C. Node 415C is the vin2400C terminal of the buffer driver, and is connected to the gate terminal of M421C, the gate terminal of M439C, the drain terminal of M437C, and the lower terminal of current source I407C. Node 416C is connected to the drain terminal of M420C, the drain terminal of M422C, the gate terminal of M422C, the gate terminal of M424C, and the upper terminal of current source I405C. Node 417C is connected to the drain terminal of M421C, the drain terminal of M423C, the gate terminal of M423C, the gate terminal of M425C, and the lower terminal of current source I405C. Node 418C is connected to the drain terminal of M424C, the drain terminal of M426C, the gate terminal of M426C, the gate terminal of M428C, the gate terminal of M430C, and the upper terminal of current source I406C. Node 419C is connected to the drain terminal of M425C, the drain terminal of M427C, the gate terminal of M427C, the gate terminal of M429C, and the lower terminal of current source I406C. Node 420C is connected to the drain terminal of M429C, the drain terminal of M428C, the drain terminal of M431C, the gate terminal of M431C, and the gate terminal of M433C. Node 421C is connected to the drain terminal of M430C, the drain terminal of M433C, the gate terminal of M435C, the drain terminal of M435C, and the gate terminal of M437C. Node 422C is the Vo400C terminal of the (inverting) buffer driver, and is connected to the drain terminal of M432C and the drain terminal of M439C.


The buffer driver (e.g., inverting buffer drive) or BUF400C is comprised of BLOCK400C which is the output FETs, or FET400C (containing source PMOSFET M432C and sink NMOSFET M439C); plus BLOCK401C which is the inverting current mirror amplifier or CMA400C (containing NMOSFETs M435C, and M437C, and current sources I407C); plus BLOCK402C which is the loser take all or LTA400C (containing PMOSFET M428C, and M430C, and NMOSFETs M429C, M431C, and M433C); plus BLOCK403C which is the complementary non-inverting current mirror or CNICM400C (containing on PMOSFETs side M420C, M422C, M424C, M426C, and on the NMOSFET side M421C, M423C, M425C, M427C, and current sources I405C and I406C). As stated earlier, AMP400C includes an amplifier gain stage of FIG. 4A coupled with BUF400C, where node 414A is connected to node 414C, node 415A is connected to node 415C. AMP400C is configured in unity gain with node 412A (the amplifier gain stage negative input terminal) connected to node 422C (or Vo400C). Note that it would be possible to utilize a CMA that is complementary version of CMA400C (i.e., PMOSFET based or NMOSFET based). Also, it would also be possible to utilize an LTA that is a complementary version of LTA400C (i.e., primarily PMOSFET base or NMOSFETs based). It would also be possible that plurality of CMAs coupled with plurality of LTAs can be utilized, which are complementary to one another (e.g., NMSOFET based CMA coupled with primarily NMOSFET based LTA, utilized along with PMSOFET based CMA coupled with primarily PMOSFET based LTA). Moreover, it would be possible to design a buffer driver that utilizes other equivalent functional implementations of CNICM, LTA or CMA. For example, LTA can perform the equivalent function of MCS. Another example would be to utilize, current mirror amplifier (CMA) instead of CMA by properly arranging the circuit signal signs. This can be done, for example, by arranging an inverting LTA feeding a CMA, instead of LTA feeding a CMA. Another example would be to utilize a current amplifier (CA) or an inverting CA to perform the equivalent function of a CMA or CMA, respectively.


The detailed description of the CNICM400C, LTA400C, CMA400C were provided in previous sections, and their inner working as part of AMP400C loop will be be briefly discussed in this section. For clarity of description, non-idealities (e.g., offsets or mismatches) are set aside in these discussions. Before describing the BUF400C operating in steady state conditions and over drive (sink-source), the current equilibrium conditions are described at nodes 414A, 414C and 415A, 415C in order for the steady state conditions to persist. At node 415C in FIG. 4C (which is connected to node 415A in FIG. 4A), KCL requires the sum of IDM416A, IDM417A, I408C, and IDM436C to converge to zero. Also, at nod 414C in FIG. 4C (which is connected to node 414A in FIG. 4A), KCL requires the sum of IDM418A, and IDM419A to converge to zero. Moreover, for AMP400 unity gain loop (containing amplifier gain stage of FIG. 4A and BUF400C) to stay in steady state, (IDM417A≈IDM419A)≈(IDM416A≈IDM418A) and I407C≈IDM437C. This is called “the conditions of current equilibrium for nodes 414C and 415C” in this section, which will be referred to herein this description.


As explained in the previous section, BLOCK403C is CNICM400C which is the complementary non-inverting current mirror, and one of its function is to monitor by mirroring the FET400C currents and rectify them before feeding the LTA400C for signal processing. Mirroring and rectifying the sink-source currents of FET400C via CNICM400C, curbs the maximal mirrored currents (e.g., to ‘2i’) before they are fed to LTA400C. Setting k=s, for steady state conditions (with no external load), it is assumed that IDM439C≈k×i≈IDM432C≈s×i→IDM421C≈i≈IDM420C→the I405C≈2i is split in half→IDM421C≈IDM423C≈i≈IDM425C and IDM420C≈IDM422C≈i≈IDM424C→the I406C≈2i is split in half→IDM424C≈i≈IDM426C≈IDM428C≈IDM430C, and IDM425C≈i≈IDM427C≈IDM429C. The IDM426C and IDM427C represent the inputs and IDM435C represents the output of LTA400C. As indicated before, one of the tasks of minimum current selector (MCS) or its functional equivalent LTA400C (in the embodiment of BUF400C) here is to monitor the sink-source signals, in this case through CNICM400C that provides IDM426C and IDM427C which are then fed onto LTA400C that outputs the current IDM435C. In steady state, with IDM428C≈i≈IDM429C→at node 420C, IDM428C-IDM429C≈IDM431C≈0≈IDM433C there is no current left for M431C→IDM431C≈0≈IDM433C→at node 421C, with IDM430C≈i, and ID430C-IDM433C≈IDM435C≈i. In summary, in steady state, LTA400C responds to current equality with LTA (IDM426C, IDM427C)≈IDM426C≈IDM427C≈i, which is fed onto the input of ICMA400C via M435C. The functions of the CMA400C, as contained in the AMP400C, is to help regulate the steady-state current in FET400C. Accordingly, IDM435C≈i→IDM437C≈i≈I407C. Here, AMP400C loop is in steady-state, and node 414C and 415C currents are balanced with (IDM417A≈IDM419A)≈(IDM416A≈IDM418A) and I407C≈IDM437C.


Next will be described the sink conditions. In the sink mode, node 414C and node 415C are pulled up towards VDD, MN439C and MN421C are turned on→IDM421C>>i→initially before current equilibrium IDM427C≈2i, which feeds one of LTA400C inputs. Also, M432C and M420Care turned off or are nearly off, initially →IDM426C≈0 which feeds the other input of LTA400C→IDM426C≈IDM428C≈IDM430C≈0 and IDM427C≈2i≈IDM429C, which pulls down on the voltage at node 420C towards VSS→IDM431C≈0≈IDM433C→with IDM430C≈0≈IDM433C, then IDM430C-IDC433C≈0 is the net current output of the LTA400C that is fed onto M435C, as the input of CMA400C. In summary so far, before equilibrium, the LTA (IDM426C, IDM427C)≈IDM426C≈0 that is fed onto CMA400C via M435C→IDM435C≈IDM437C≈0<I407C=i at node 415C, which is a current imbalance, initially→AMP400C loop kicks to bring back the balance. In order for “conditions of current equilibrium for nodes 414C and 415C” to return here, and to simplify the explanation for clarity of description, in FIG. 4A (IDM417A≈IDM419A) can increase slightly over (IDM416A≈IDM418A)→AMP400C loop can regulate down V414C and V415C slightly enough until M420C (and its mirror M432C) turn back on enough for IDM420C≈IDM426C≈i≈IDM428C≈IDM430C. Moving towards steady state while, MN439C is sinking current for an external load→IDM421C>>i→IDM427C≈2i≈IDM429C→IDM429C≈2i>IDM428C≈i→M429C pulls down on the voltage at node 420C towards VSS→IDM431C≈0≈IDM433C→moving towards steady state, with IDM430C≈i, then IDM430C-IDM433C≈i is the net current output of the LTA400C that is fed onto M435C, as the input of CMA400C. Note that LTA(IDM426C, IDM427C)≈IDM426C≈i. In summary, this is the case for LTA400C since M439C is still sinking current, IDM439C>IDM432C→IDM421C>IDM420C→IDM427C>IDM426C. As explained in the previous section, note that in FIG. 4A, despite the very slight increase in (IDM417A≈IDM419A) over (IDM416A≈IDM418A) that can regulate down V414C and V415C, then CMA400C (coupled and being fed via LTA400C and CNICM400C) have enough gain to regulate IDM437C low enough (below the fixed I407C≈i) in order V415C to be high enough so that M439C can continue sinking high currents to the external load. Here, the “conditions of current equilibrium for nodes 414C and 415C” are held, and steady state conditions maintained, while M439C to continue sourcing current for an external load. Accordingly, IDM420C≈i→operating current in the inactive source-FET is thus controlled and regulated at IDM431C≈s×i, while steady state condition is maintained and the sink FET, M439C, continues sinking extra current to an external load.


Conversely, in the source mode, when node 414C and node 415C of FIG. 4C are pushed down towards VSS, initially M432C and M420C turn on and M439C and M421C turn off→then initially before equilibrium, IDM426C≈2i and IDM427C≈0. In the LTA400C, IDM426C≈2i≈IDM428C≈IDM430C and IDM427C≈0≈IDM429C→IDM428C−IDM429C≈2i≈IDM431C≈IDM433C→before equilibrium, IDM430C−IDM433C≈2i−2i≈0 is the net current output of the LTA400C that is fed onto M435C as the input of CMA400C→IDM435C≈0. In summary so far, before equilibrium, the LTA (IDM426C, IDM427C)≈IDM427C≈0 that is fed onto CMA400C→IDM435C≈IDM437C≈0 at node 415C with I407C=i, which in a current imbalance at node 415C initially, that is unsustainable→CMA400C, contained within the AMP400C loop, lifts V415C enough until IDM437C≈i=I407C→‘conditions of current equilibrium for nodes 414C and 415C’ can be maintained when (IDM416A≈IDM418A)≈(IDM417A≈IDM419A) and IDM437C≈i=I407C, which occurs after AMP400C loop regulates V415C high enough→IDM421C≈IDM427C≈i≈IDM429C, while MN432C countinues to source current for an external load→IDM420C>>i→IDM426C≈2i≈IDM428C≈ID430C→IDM428C−IDM429C≈i, which is fed onto IDM431C≈i≈IDM433C→moving towards steady state, IDM430C−IDM433C≈i, which is fed onto input of CMA400C that is IDM435C≈i≈IDM437C≈I407C≈i at node 415. In summary, note that since M432C is still sourcing current, while in equilibrium IDM432C>IDM439C→IDM420C>IDM421C→IDM426C>IDM427C→at equilibrim LTA(IDM426C, IDM427C)≈IDM427C≈i which is fed onto CMA400C via IDM435C≈i≈IDM437C≈I407C≈i. As such, when the buffer driver sources current for an output load, then M439C which is the inactive sink FET, can operate at a controlled regulated current IDM439C≈k×i. Again, note that ‘i’ is a constant current source set by the main bias network of the amplifier, and ‘k’ is FET aspect ratio that is tightly controllable, which helps with the control of the quiescent current in the inactive sink FET.


In summary, when there is an imbalance between sink-source FETs currents, IDM432C and IDM439C, such imbalance is scaled down, mirrored, and rectified by the complementary non-inverting current mirror (CNICM400B). The outputs of CNICM400C, IDM427C and IDM426C, which are the rectified version of sink-source currents, are then fed onto the LTA400C. The LTA400C effectively performs the equivalent function of selecting the (scaled) minimum current between the output drive currents IDM432C and IDM439C. The selected minimum current, at the output of LTA400C, is then inputted onto CMA400C which, together with the operations of amplifier's gain stage inside the AMP400C loop, regulates a minimum quiescent current in the inactive FET (e.g. for s=k, when IDM439C is sinking current for an external load→inactive source FET IDM432C=s×i, and when IDM432C is sourcing current for an external load→inactive sink FET IDM439C=k×i).


In conclusion, as noted in the previous section, the benefits of the FIG. 4B's BUF400B are equally applicable to that of FIG. 4C's BUF400c (comprising CNICM coupled with LTA coupled with CMA) which are briefly summarized here again. First, the function of ‘monitoring’, of sink-source currents, consumes current itself. Sink-source currents can be high and have unpredictable patterns. The CNICM effectively curbs (e.g., rectify) the ‘monitoring’ current consumption overhead. Also the rectifying of output FET currents, before they are fed to the LTA and CMA, helps with consistency of DC, AC, and transient performance of the buffer driver, with more independence from external load variations. Second, the buffer driver is fast since the CNICM couple with LTA coupled with CMA are all in principal operate in current mode, which is inherently fast. Third, it can operate with low power supply voltage since the minimum VDD is VGS+2VDS (low voltage) for CNICM, LTA, and CMA. Fourth, it is made of a few transistor which makes it small and low cost. Fifth, buffer driver can operate with ultra low current and ult low power. The utilization of CNICM helps curb the current consumption attributed with monitoring the sink-source currents when the amplifier drives different external loads. Sixth, it has wide input-output span of near rail-to-rail. Seventh, it can provide high sink-source drive capability and controlled quiescent current in the inactive output FET that benefits DC and dynamic performance. FIG. 4F is a simulation of a buffer amplifier containing buffer driver of FIG. 4C coupled with gain stage of FIG. 4A. This simulations indicates approximate and typical sink-source current capability for the buffer amplifier to drive external load resistors 900KΩ, 300KΩ, and 30KΩ. The simulation indicates current consumption for the buffer amplifier of roughly 110 nA for FIG. 4C coupled with gain stage of FIG. 4A. Eighth, buffer driver is minimally imposing on the amplifier's gain or the speed of the preceding amplifier's high-gain stage. Ninth, it's based on standard CMOS process which is low cost, high quality, ready available). Tenth, the buffer driver arrangement utilizing CNICM coupled with LTA coupled with CMA can be tailored to very high current and very high speed.


Section (XV): Detailed Description of Buffer Driver (BUF) Comprised of Minimum Current Selector (MCS) and Inverting Current Mirror Amplifier (ICMA), Illustrated in FIG. 4D (4D)



FIG. 4D is a circuit schematic showing a buffer driver (e.g., inverting buffer diver), BUF400D, comprising MCS400D coupled with ICMA400D. This last section is started by summarizing the differentiation of the last 2 sections, that is equally applicable to this section's disclosure regarding buffer driver MCS400D coupled with ICMA400D: (1) can operate at low power supply voltage, with VDD as low as VGS+2VDS, (2) principally operates in current mode which makes it fast and widens its dynamic range, (3) facilitates near rail-to-rail input-output operation, (4) it is simple and made of few transistors, making it inexpensive, (5) can sink and source large currents to heavy output loads, while keeping its internal current consumption ultra low (6) can control the inactive sink-source output FET current, and thus improve its DC and dynamic performance (7) can run at ultra-low currents, (8) impose minimal loading back into the preceding amplifier's high-impedance stage and as a result attain high-gain amplification, (9) impose minimal low frequency poles, that might otherwise burden the AC and transient response of the amplifier's gain stage, and as a result improve an amplifier's dynamic response (10) can be made using standard CMOS process which is inexpensive and high quality, and (11) the advantages of this teaching are not exclusive to ultra-low power, but this teaching can be tailored for buffer drivers that need very high-speed, very high drive current capability, and provide near rail-to-rail input-outputs, with low supply voltages as well.


While the buffer drivers sinks or source large currents for an external load, for continued proper performance including fast dynamic response, low cross over distortion, and low output impedance, buffer drivers generally run a minimum (well controlled) quiescent current in their inactive output FET drivers. In this embodiment, the MCS400D sense the BUF400D's output FET driver's (M426D-M431D) current signals and feed a proportional signal into ICMA400D. While the output FET is sinking current, the MCS400D coupled with ICMA400D, and in concert with AMP400D loop, regulate the minimum current in the (inactive) source output FET. Conversely, while the output FET is sourcing current, the MCS400D coupled with ICMA400D, and in concert with AMP400D loop, regulate the minimum current in the (inactive) sink output FET.


Throughout the description of FIG. 4D, MOSFET scale factors (e.g., ‘s’, ‘k’, or W/L ratios for M421D-M422D, M424D, M423D-M425D, etc) and current source scale factors (e.g., for I407D-I408D) can be altered depending on cost-performance goals such as die size, current consumption, speed, and cross over distortion, and other goals. For the FIG. 4D disclosure, s=k=20 but ‘s’, ‘k’ can have wide ranges 0.01≤s≤1000000, 0.01≤k≤1000000. The connections of the elements in FIG. 4D are described as follows. The body terminal of all NMOSFETs in FIG. 4D are connected to node 2 which is VSS, and the body terminals of PMOSFETs are connected to node 1 which is VDD. Bias current sources I407D and I408D have their upper terminals connected to node 1, which is VDD. The source terminal of NMOSFETs M421D, M423D, M425D, M427D, M429D, and M431D, are connected to node 2 that is VSS. The source terminal of PMOSFETs M420D, M422D, and M426D are connected to node 1 that is VDD. Node 414D is the vin1400D terminal of the buffer driver, and is connected to the gate terminal of M424D, the gate terminal of MM426D, the drain terminal of M429D, and the lower terminal of current source I408D. Node 415D is the vin2400D terminal of the buffer driver, and is connected to the gate terminal of M421D and the gate terminal of M431D. Node 416D is connected to the gate terminal of M420D, the drain terminal of M420D, the drain terminal of M421D, and the gate terminal of M422D. Node 417D is connected to the drain terminal of M422D and the source terminal of M424D. Node 418D is connected to the drain terminal of M424D, the gate terminal of M423D, the drain terminal of M423D, and the gate terminal of M425D. Node 419D is connected to the drain terminal of M425D, the gate terminal of M427D, the drain terminal of M427D, and the gate terminal of M429D, and the lower terminal of current source I407D. Node 420D is the VO400D terminal, which is the output of the buffer driver, and is connected to the drain terminal of M426D and the drain terminal of M431D.


The buffer driver (e.g., inverting buffer driver) is comprised of BLOCK400D which is the output FETs, or FET400D (containing source PMOSFET M426D and sink NMOSFET M431D), BLOCK401D which is the non-inverting current mirror amplifier or ICMA400D (containing NMOSFETs M423D, M425D, M427D, M429D, and current sources I407D, and I408D), and BLOCK402D which is the minimum current selector or MCS400D (containing NMOSFET M421D, and PMOSFETs M420D, M422D,and M424D). As stated earlier, AMP400D is comprised of amplifier gain stage of FIG. 4A coupled with BUF400D, where node 414A is connected to node 414D, node 415A is connected to node 415D. AMP400D is configured in unity gain with node 412A (the amplifier gain stage negative input terminal) connected to node 420D (or VO400D)


Note that it would be obvious for one skilled in the art to utilize a ICMA that is complementary version of ICMA400D (i.e., PMOSFET based or NMOSFET based). It would also be obvious for one skilled in the art to utilize an ICMA that is a complementary version of MCS400D (i.e., PMOSFET base or NMOSFETs based). Additionally, it would be obvious for one skilled in the art that plurality of ICMAs coupled with plurality of MCSs can be utilized, which are complementary to one another (e.g., NMSOFET based ICMA coupled with primarily PMOSFET based MCS, utilized along with PMSOFET based ICMA coupled with primarily NMOSFET based MCS). Moreover, it would be obvious for one skilled in the art to design a buffer driver that utilizes other equivalent functional implementations that are minimum current selector (MCS) or non-inverting current mirror amplifier (ICMA). For example, a loser take all (LTA) can perform the equivalent function of MCS. Another example would be to utilize, a current mirror amplifier (CMA) instead of an inverting current mirror amplifier (ICMA) by arranging the circuit signal signs. Another example would be to utilize a current amplifier to perform the equivalent function of a ICMA.


Describing the details of the circuit in FIG. 4D is as follows, setting aside the non-idealities (e.g., mismatches, offsets). Let's first briefly describe the operation of minimum current selector (MCS400D), in the boundary conditions (with inputs near the rails), independent of the ICMA400D for now. The drain currents IDM421D and IDM424D track the currents in FET400D, which are M431D (sink NMOSFET) and M426D (source PMOSFET), respectively. The output of MCS400D is ID424D. The function of MCS400D is to select the min(IDM421D, IDM424D). When vin1400D and vin2400D are both up near VDD, then M421D (tracking M431D) is hard on and M424D (tracking M426D) is off or nearly off. Here, IDM421D>>i and IDM424D<<i. Excluding role of ICMA400D for now, if vin1400D≈vin2400D both voltages are near VDD→MCS400D operates and selects the min(IDM421D, IDM424D)≈IDM424D<<i. Conversely, when vin1400D≈vin2400D both voltages are near VSS, then M421D (tracking M431D) is off or near off and gate terminal of M424D (tracking M426D) tends to turn it hard on, but M424D is deprived of current since M422D is off and blocking any current flow. Here, IDM421D≈IDM420D≈IDM422D<<i. Again, excluding role of ICMA400D for now, if vin1400D≈vin2400D both voltages are near VSS→MCS400D operates and selects the min(IDM421D, IDM424D)≈IDM421D<<i.


Before describing the BUF400D operating in over drive (sink-source modes) and steady state conditions, there is described the current equilibrium conditions at nodes 414 and 415 for steady state to hold. At node 414D in FIG. 4D (which is connected to node 414A in FIG. 4A), KCL requires the sum of IDM418A, IDM419A, ID408D, and IDM429D to converge to zero for steady state. At node 415D in FIG. 4D (which is connected to node 415A in FIG. 4A), KCL requires the sum of IDM416A, and IDM417A to converge to zero for steady state. For AMP400 unity gain loop (containing the amplifier gain stage of FIG. 4A and BUF400D of FIG. 4D) to hold in steady state, (IDM417A≈IDM419A)≈(IDM416A≈IDM418A) and I408D≈IDM429D. This condition will be called “the conditions of current equilibrium for nodes 414D and 415D” in this section, which will be referred to later.


In the sink mode, when node 414D and node 415D are pulled up towards VDD, MN431D (i.e., sinking current for the external load) and MN421D are turned on hard that cause IDM421D≈IDM420D≈IDM422D>>i. Here, M426D and M424D turn off (initially) and IDM424D<<i, which would block IDM422D. Thus, the output current of MCS400D selects the min(IDM421D, IDM424D)=IDM424D<<i, which flows onto M423D, and initially→IDM423D=IDM425D<<i→all of I407D=2i flows into M427D→IDM427D≈IDM429D=2i. For “conditions of current equilibrium for nodes 414D and 415D” to hold, node 414D gets regulated by being pulled down in part by IDM429D, that is initially before equilibrium runs at ‘2i’, until I408D=i≈IDM429D. In summary, in order for this current equilibrium to persist, ICMA400D contained within AMP400D loop regulate node 414D by lowering node 414D voltage sufficiently enough, which is when I408D=i≈IDM429D→IDM424D≈IDM423D≈IDM425D≈IDM427D≈IDM429D≈i≈(1/s)×IDM426D. As such, for BUF400D in current ‘sink mode’ when IDM431D>>IDM426D, the inactive source FET, M426D, is regulated to run at a controlled current ID426D≈s×i . Note that ‘i’ is a constant current source set by the main bias network of the amplifier, and ‘s’ is FET aspect ratio that is tightly controllable.


Conversely, in the source mode, when node 414D and 415D of FIG. 4D are pulled down towards VSS, and M426D is turned on hard, where initially M431D, M421D, M420D,and M422D are turned off or are near off. Although, gate terminal voltage of M424D is pushed down towards VSS, but M422D being initially off or near off, starves M424D. As such, MCS400D initially selects the min(IDM421D, IDM424D)=IDM421D<<i, that flows through M424D and onto MM423D, which is the input ICMA400D. As such, before equilibrium, IDM421D≈IDM420D≈IDM422D≈IDM424D≈MM423D≈IDM425D<<i→all of I407D=2i would initially flow through M427D→IDM427D≈IDM429D≈2i>I408D→creates a current imbalance at node 414B that is unsustainable→AMP400D loop kicks in to regulate back towards current equilibrium. In order for “conditions of current equilibrium for nodes 414D and 415D” to return and hold, and to simplify the explanation for clarity of description, in FIG. 4A (IDM416A≈IDM418A) can get very slightly increased initially above (IDM417A≈IDM419A) in FIG. 4A. Here, by AMP400D loop regulation, the voltages at both node 414D and node 415D are both initially pulled up until M421D (and its mirrors M431D) turn back on sufficiently enough for IDM421D≈i, which brings current equilibrium. Note in FIG. 4A that despite the very slight increase in (IDM416A≈IDM418A) over (IDM417A≈IDM419A) at node 414D, ICMA400D (coupled and fed via MCS400D) have enough gain to regulate a slightly increased IDM429A (compared to the fixed I408D≈i) to keep the voltage at node 414D low enough so that M426D can continue sourcing current to the external load. The only draw-back here shows up in AMP400 DC (steady state) gain reduction as the sink-source currents for an external load is increased (i.e., the smaller the resistive load, the smaller the AMP400 gain, which is not uncommon in inverting buffer drivers). Here, the “conditions of current equilibrium for nodes 414D and 415D” are held, and steady state conditions maintained, while M426D to continues sourcing current. Just to capture the current balance and steady state conditions here, note that with IDM426D>>IDM431D→IDM424D>>IDM421D→min(IDM421D, IDM424D)=IDM421D≈i→operating current in the inactive sink-FET, that is mirrored with M421D, is controlled and regulated at IDM431D≈k×i, while steady state condition hold and the source FET, M426D, continues sourcing extra current to an external load.


Again, ignoring non-idealities, here the mechanism for the BUF400D operating in steady state is described, when there is no output load. At steady state conditions, I408D≈IDM429D at node 414D in FIG. 4D, as well as in FIG. 4A where IDM418A≈IDM419A at node 414A and IDM416A≈IDM417A at node 415A, and IDM418A≈IDM416A≈IDM419A≈IDM417A. In steady state, the current output of the MCS400D selects the min (IDM421D, IDM424D)≈IDM421A≈IDM424A≈i→I408D≈IDM429D≈IDM427D≈IDM425D≈IDM423D≈i≈IDM424D. For clarification, note that the MCS400D is arranged as a translinear circuit. Here, the VGM420D≈VGM424D (relative to VDD). When M420D that has a W/L=1× and is in parallel with the series association of M424D-M422D, each with W/L=2x→IDM424D≈IDM422D≈i≈IDM420≈IDM421. Note that VGSM420D≈VGSM426D and VGSM421D≈VGSM431D. As such, by setting FET400D scale factors k=s, the (inverting) output FETs currents, in the buffers driver, operate with a controlled quiescent current of IDM431D≈IDM426D≈s×i while in steady state.


In summary, a near rail-to-rail input-output (i.e., inverting) buffer driver is proposed utilizing a novel current mode output stage comprising of minimum current selectors (MCS) and non-inverting current mirror amplifiers (ICMA) which are fast and operate at low VDD. The preceding gain amplifier stage high-impedance high-gain node, dominates in setting the AC and transient response, with minimal speed degradation caused by the fast buffer driver. Such is the case in part because of the (inverting) buffer driver wide bandwidth since it operates chiefly in current mode makes its fast inherently. The (inverting) buffer driver is based on the main-stream, standard, low cost, and rugged standard digital CMOS manufacturing platform.


In conclusion, the benefits of the FIG. 4D's BUF400D comprising MCS400D and ICMA400D include the following. First, the (inverting) buffer driver can work with minimum VDD of VGS+2VDS (low voltage). Second, it principally operates in current mode which makes it inherently fast. Third, it is simple and made of a few transistor which makes it small, low cost, and higher quality. Fourth, it can run at ultra low currents, which makes its power consumption ultra-low when combined with its ability to perform to specifications with low power supply voltage. Fifth, it can sink and source significantly higher currents (compared with its own steady-state operating current) for an external load, and have regulated and controlled quiescent current in its inactive output FET which helps with the (inverting) buffer drivers performance, including more consistent DC and dynamic response, amongst other. FIG. 4G is a simulation of a buffer amplifier containing buffer driver of FIG. 4D coupled with gain stage of FIG. 4A. This simulations indicates approximate and typical sink-source current capability for the buffer amplifier to drive external load resistors 900KΩ, 300KΩ, and 30KΩ. The simulation indicates current consumption for the buffer amplifier of roughly 120 nA for FIG. 4D coupled with gain stage of FIG. 4A. Sixth, the MCS that monitors the output FET's (BLOCK-400D) operates in current mode. Also, the ICMA, while running chiefly in current mode, regulates the operating currents in the output (inactive) FETs. Running MCS and ICMA chiefly in current mode enables them to be inherently fast, but additionally, processing signals in current mode reduces internal voltage swings that helps with operating at lower voltages, which widens the input-output voltage span closer to the rails. This is beneficial to low power supply environments where there little signal to noise cushion available to waste for overhead at input or output terminals. Seventh, while utilization of current mirror inverter lowers the input impedance at the internal inputs of ICMA that help with faster signal processing speeds within ICMA. Moreover, the ICMA has a high output impedance that, when connected to an amplifier's gain stage, it neither substantially loads the amplifier's high impedance node nor does it dynamically slow the amplifier. Eighth, (inverting) buffer driver needs no special process and can be manufactured on rugged and time tested main-stream CMOS process that is inexpensive, readily available at many foundries, and with high quality. Ninth, this teaching can be tailored for (inverting) buffer drivers that need very high-speed, very high current drive capability, and near rail-to-rail input-outputs with low supply voltages s as well (i.e., can be very high current and very high speed).


The definitions of the words or elements of the claims shall include not only the combination of elements which are literally set forth, but all equivalent structure, material or acts for performing substantially the same function in substantially the same way to obtain substantially the same result.


All references, including publications, patent applications, patents and website content cited herein are hereby incorporated by reference to the same extent as if each reference were individually and specifically indicated to be incorporated by reference and was set forth in its entirety herein.


The words used in this specification to describe the invention and its various embodiments are to be understood not only in the sense of their commonly defined meanings, but to include by special definition in this specification any structure, material or acts beyond the scope of the commonly defined meanings. Thus if an element can be understood in the context of this specification as including more than one meaning, then its use in a claim must be understood as being generic to all possible meanings supported by the specification and by the word itself.


Recitation of ranges of values herein are merely intended to serve as a shorthand method of referring individually to each separate value falling within the range, unless otherwise indicated herein, and each separate value is incorporated into the specification as if it were individually recited herein. Therefore, any given numerical range shall include whole and fractions of numbers within the range. For example, the range “1 to 10” shall be interpreted to specifically include whole numbers between 1 and 10 (e.g., 1, 2, 3, . . . 9) and non-whole numbers (e.g., 1.1, 1.2, . . . 1.9).


Neither the Title (set forth at the beginning of the first page of the present application) nor the Abstract (set forth at the end of the present application) is to be taken as limiting in any way as the scope of the disclosed invention(s). The title of the present application and headings of sections provided in the present application are for convenience only, and are not to be taken as limiting the disclosure in any way.

Claims
  • 1. A method of arranging an ultra low power Buffer Amplifier (BA), containing an Amplifier (A) coupled with a Buffer Driver (BD), comprising: increasing gain and widening the input-output voltage span of the Amplifier (A) utilizing a plurality of regulated cascode (RGC) current mirrors (RGC-CM) where each RGC-CM is made of at least one of the following three circuits: 1) a diode connected self cascode (DCSC) coupled with a common source amplifier (CSA); 2) a current mirror amplifier (CMA) that contains a common source amplifier (CSA); and 3) a CSA coupled with one common gate amplifier (CGA) wherein the common gate terminal of the CGA is connected to the common source terminal of the CSA;lowering the minimum operating power supply and reducing offset of the Amplifier (A) by utilizing a current equalizer circuit that emulates the function of a floating current source (FCS) having at least two complementary cascode current sources made of field effect transistors (FETs), wherein the middle cascoded FET's gate to source voltages (VGS) are held constant by regulating the VGS of the upper and lower FETs wherein the source terminals of the said upper and lower FETs are connected to the power supplies, and the upper and lower FET currents are equalized and mirrored into the Amplifier's (A) bias network;lowering output noise of the Amplifier(A), by narrow banding the Amplifier(A)'s high gain node, while concurrently rejuvenating the narrow banded Amplifier(A)'s speed by utilizing one of the following circuits: 1) minimum current select (MCS); and 2) loser take all (LTA) circuits to generate dynamically boosted operating current when the Amplifier(A)'s inputs are intermittently imbalanced; andlowering the minimum operating power supply, having near rail-to-rail input output voltage span, having high-speed, providing large sink-source current for output load, while regulating the operating current in the inactive sink-source transistor of the Buffer Amplifier (BA) by utilizing a buffer driver (BD) that contains at least one of the following circuits: 1) MCS; 2) LTA; 3) current mirror amplification (CMA); and 4) complementary current mirror that enable the Buffer Driver (BD) to chiefly operate and process signals in current mode.
CROSS-REFERENCE TO RELATED PATENT APPLICATIONS

This application claims the priority benefit of U.S. Provisional Patent Application Ser. No. 62/304,373, filed Mar. 7, 2016 and entitled “Class AB amplifier low power and fast”; and U.S. Provisional Patent Application Ser. No. 62/320,512 filed Apr. 9, 2016 and entitled “Class AB amplifier high gain”; and U.S. Provisional Patent Application Ser. No. 62/415,496 filed Oct. 13, 2016 and entitled “Class AB Amplifier Low Noise”; Each of the above applications are herein specifically incorporated by reference in their entirety.

US Referenced Citations (41)
Number Name Date Kind
4677369 Bowers et al. Jun 1987 A
5283579 Tasdighi et al. Feb 1994 A
5294927 Levinson et al. Mar 1994 A
5640084 Tero et al. Jun 1997 A
5668710 Caliboso et al. Sep 1997 A
5734260 Tasdighi et al. Mar 1998 A
5734292 Tasdighi et al. Mar 1998 A
5814995 Tasdighi Sep 1998 A
5923208 Tasdighi et al. Jul 1999 A
6005374 Tasdighi Dec 1999 A
6054823 Collings et al. Apr 2000 A
6122284 Tasdighi et al. Sep 2000 A
6281743 Doyle Aug 2001 B1
6366071 Yu Apr 2002 B1
6384683 Lin May 2002 B1
6489831 Matranga Dec 2002 B1
6657480 Ochi Dec 2003 B2
6885178 Marinca Apr 2005 B2
7211993 Marinca May 2007 B2
7612606 Marinca Nov 2009 B2
20020093325 Ju Jul 2002 A1
20040124822 Marinca Jul 2004 A1
20050264345 Ker Dec 2005 A1
20050285666 Garlapati Dec 2005 A1
20060224662 Haiplik Sep 2006 A1
20070052473 McLeod Mar 2007 A1
20070076483 Colman Apr 2007 A1
20070132506 Fujisawa Jun 2007 A1
20070200620 Iriguchi Aug 2007 A1
20080018316 Chang Jan 2008 A1
20080278137 Harvey Nov 2008 A1
20090243711 Marinca Oct 2009 A1
20100164467 Jo Jul 2010 A1
20100225384 Hirose Sep 2010 A1
20100237926 Kikuta Sep 2010 A1
20110025666 Baek Feb 2011 A1
20110068767 Carper Mar 2011 A1
20110102049 Conte May 2011 A1
20110205193 Nishimura Aug 2011 A1
20120176186 Chen Jul 2012 A1
20150160680 Marinca Jun 2015 A1
Non-Patent Literature Citations (31)
Entry
Sansen, Willy M., “A CMOS Temperature-Compensated Current Reference”, IEEE Journal of Solid-State Circuits, Jun. 1988,p. 821, vol. 23.
De La Plaza, Alejandro,“Nanoampere Supply Independent Low-Voltage Current Reference”, Faible Tension Faible Communication, May 2011,p. 9.
Fayomi, Christian, “Sub 1V CMOS bandage reference design techniques a survey”, Analog Integrated Cir Sig Process, Feb 2010, p. 141.
Vittoz, Eric, “A Low-Voltage CMOS Bandgap Reference”, IEEE Journal of Solid-State Circuits, Jun. 1979, vol. SC-14, No. 3, Feb. 2010, p. 141.
Galeano, Edgar,“A Very Low Power Area Efficient CMOS Only Bandgap Reference”, Integrated Circuits and Sys Design, Aug. 2012, p. 1.
Osaki, Yuji, “A 95-nA, 523 ppm/C, .6-uW CMOS Current Reference Circuit w/ Subthreshold MOS Resistor Ladder”, ASPDAC '11, Jan. 2011, p. 113.
Ueno, Ken, “A 1-uW 600-ppm/C Current Ref Circuit Consisting of Subthreshold CMOS Circuits”, IEEE Transaction on Circuits, Sep. 2010, p. 681.
De Vita, “A 109 nW, 44 ppm/C CMOS Current Reference with Low Sensitivity to Process Variations”, Circuits and Systems, Feb. 2007, p. 3804.
Ledesma, F., “Comparison of New and Conventional Low Voltage Current Mirrors”,Circuits and Systems, Aug. 2003, p. 49-52.
Hihose, Tetsuya, “A Nano-Ampere Current Reference Circuit and Its Temperature Dependence,” Proceedings of the EECIRC, Oct. 2010, p. 114.
Camarcho-Galeano, Edgar, “A 2-nW 1.1V Self-Biased Current Reference in CMOS Technology”, On Circuits and Systems, Feb. 2005,p. 62.
Camarcho-Galeano, Edgar,“Temperature Performance of Sub-1V Ultra-Low Power Current Sources”, Circuits and Systems, May 2008, p. 2230.
Osaki,“Nano-Ampere CMOS Current Reference with Little Temp. Dependence Using Small Offset Voltage”, Proc. at 53rd IEEE Intl, Aug. 2010, p. 668.
Gallegos, S.,“A 128.7nW Neural Amplifier and Gm-C Filter for EEG, Using gm/ID Methodology..”, Circuits and Systems, Aug. 2014, p. 876.
Cordova, David,“A Low-Voltage Current Reference with High Immunity to EMI”, Integrated Circuits and Systems Design, Sep. 2014, p. 1-6.
Annema, Anne-Johan,“Low-Power Bandgap Heferences featuring DTMOST's”, IEEE Journal of Solid-State Circuits, Jul. 1999, p. 949.
Buck, Ame, “A CMOS Bandgap Reference without Resistors”, Solid-State Circuits Conference, Dec. 2000, p. 442-443.
Chouhan, Shailesh, “A moditied CMOS Nano-Power Resistorless Current Reference Circuit”, Microelectronics and Electronics, Jun. 2014,p. 1.
Guiges, Fabrice, “Sub-1V Oguey's Current Reference Without Resistance”, Circuits and Systems, Dec. 2006, pp. 288-291.
Guistolisi,G.,“A Low-Voltage Low Power Voltage Reference Based on Subthreshold MOSFETS”, Journal of Solid-State Circuits, Jan. 2003, p. 151.
Hirose, T., “A CMOS Bandgap and Sub-Bandgap Voltage Reference Circuits for Nanowatt Power LSIs”, IEEE Asian Solid-State, Nov. 2010, p. 77.
Kussener, F, “50 nA, 1V Nanowatt Resistor-Free Compact CMOS Current References”, Electronics, Circuits and Systems, Dec. 2010, p. 1212.
Oguey, H., “CMOS Current Reference Without Reference”, IEEE Journal of Solid-State Circuits, Jul. 1997, p. 1132, vol. 32.
Serra-Graelis, F.,“Sub-1V CMOS Proportional-to-Absolute Temp. References”, Journal of Solid-State Circuits, Jan. 2003, p. 84.
Ueno,Ken,“A 300 nW,15 ppm/C,20 ppm/V CMOS Voltage Ref Circuit . . . ”, Journal of Solid-State Circuits, Jun. 2009, p. 2097.
VanPetegham, Peter, “On a New CMOS Resistor Implementation for Linear Analog IC Signal Processing Applications,” IEEE, Jun. 1998, p. 163-166.
Vittoz, Eric, “CMOS Analog ICs Based on Weak Inversion Operation”, Journal of Solid-State Circuits, Jun. 1977, p. 224.
Wilson,G., “Novel Voltage-Controlled Grounded Resistor”, Electronic Letters, Dec. 1989, pp. 1725-1726.
Huang, Zhangcai, “A CMOS Sub-1-V Nanopower Current and Voltage Ref with Leakage Compensation”, Circuits and Systems, Jul. 2010, p. 4069.
Columbo, Dalton, “A CMOS 25.3 ppm/C Bandgap Voltage Ref using Self-Cascode Composite Transistor”, Circuits and Systems, Apr. 2012, p. 1.
Colombo, Dalton, “A CMOS 25.3 ppm/C Banagap Voltage Ref Using Self-Cascade Composite Transistor”, Circuits and Systems, Dec. 2012, p. 1.
Provisional Applications (3)
Number Date Country
62304373 Mar 2016 US
62320512 Apr 2016 US
62415496 Oct 2016 US