1. Field of the Invention
The present invention relates to current amplifier circuits, and in particular to a programmable multi-gain current amplifier.
2. Description of the Related Art
Filters utilizing current amplifiers (CAs), unlike the second generation current conveyer (CCII) and its extended versions are pure current-mode circuits wherein the primary signal variable is current. Basically, a CA is equivalent to a current controlled current source with gain α and ideally zero input impedance and infinite output impedance. It is often defined as a two terminal device that amplifies a current input signal applied at its low impedance input terminal X and conveys it to a high impedance output terminal Z. A current follower (CF) is a unity gain CA. From a certain point of view, it is the simplest current-mode device since it is a subpart of several devices such as CCII, current feedback amplifier (CFA) and transresistance amplifier (TRA).
In the area of filter design, CA based topologies avoid the use of any explicit or implicit voltage-mode active component leading to true current-mode signal processing circuits having the potential to operate at relatively high frequency and large signal swings. In fact, the bandwidth and the closed-loop gain are almost independent and a high voltage swing is not usually required. Known configurations include a current-mode Sallen-Key low-pass filter based on the CA and other filters based on CFs.
A current controlled current conveyor is widely used to provide the missing tuning feature of its CCII based counterparts. A programmable CA would be obtained from a CCCII by grounding its Y terminal. A programmable integrator can be obtained from a single output CCCII but this would require applying an input voltage (instead of the usual current signal) at the X-terminal of the CCCII. On the other hand, a current-mode programmable integrator can be obtained by using a dual-output CCCII adopting the topology described previously for the CA, or it can be obtained by using two CCCII.
Alternatively, electronically controllable CCIIs (ECCIIs) are obtained with the help of small-signal current amplifiers as suggested in the prior art. Basically, the sensed output current (ix) of the voltage buffer stage is applied to the input of a current amplifier. The current amplifier amplifies ix and makes it available from a high output impedance terminal Z. The current gain of these current amplifiers is often a function of ratio between two biasing currents. Thus, the gain is controlled by varying these biasing currents. A BJT based ECCII is known in the art and complementary metal oxide semiconductor (CMOS) based ECCIIs are also known. In the known CMOS circuits, however, the maximum gain is limited by the size of some current mirror transistors (e.g. with equal size current mirrors the gain is limited to 2). Large transistors require large silicon area and result in limited bandwidth as they would be associated with larger practice capacitances. The known programmable CCII replaces the current amplifier with several cascaded current division cells. Each cell consists of seven transistors with the first two cells (among n cells) consuming 300 μA leading to an area and a power inefficient solution.
A CCII with both voltage and current gain called VCG-CCII is also known in the art. The voltage gain Vx/Vy, just like any voltage amplifier, suffers from gain-bandwidth product problems. Whereas the current gain Iz/Ix is proportional to the small signal transconductance or output resistance of current mirroring transistors. The bandwidths of the current transfer characteristics of the VCG-CCII of these known works are limited to approximately 20 kHz and 1 MHz, respectively. In addition, the operation of these circuits is valid only for small signal processing, limiting their linearity.
A highly linear programmable CCII based on a programmable current mirror of [25] is also known in the art. However, this technique employs transistors operating in moderate inversion which results in low bandwidth (17 MHz) and limited tuning range. Adjustable current mirrors such as the known cell in the art can also be utilized but extending the design to produce multi-output would be associated with increased circuit complexity. Additional known art presents a single output programmable current amplifier employing two SINH or two TANH blocks, for example. But this known topology is clearly inefficient, since each of the two blocks consists of a CCII and a CM with adjustable gains. Another known technique is the design of a digitally controlled fully differential current conveyor (DCFDCCII). It utilizes two current division networks (CDN) in order to realize a single DCFDCCII.
In known current mirror CM based circuits, the input stage transconductance (basically 1/gm) is used for electronic programmability. Hence, it is possible to realize active-C filters based on CMs under the condition that each CM in the circuit topology is loaded by series resistance at the X terminal. In this case, the passive resistors would be replaced by the internal input resistance of the CM. An adjustable biasing current is used to vary the parasitic resistance (1/gm) of the CM's X-terminal. However, this approach cannot be employed to provide independent tuning characteristics since it is impossible to obtain different gains from the same device. This leads to filters exhibiting dependent pole frequencies and pole quality factors.
The second unattractive feature of these known filters is that various gains are often fixed to unity. A third problem with this known approach is that 1/gm is naturally nonlinear which significantly limits the linearity and the tuning range. Also, it can be seen that filters based on CM devices inherently cannot provide virtual ground input impedance (i.e., 1/gm is not ideally zero). Yet there remains the problem of gain programmability and a number of independently programmable outputs.
Thus, a programmable multi-gain current amplifier addressing the aforementioned problems is desired.
A novel multi-output current amplifier exhibiting independent electronically controllable gains is given. Its circuitry includes a pair of metal oxide semiconductor (MOS) transistors that set the voltage at X terminal to zero and with the help of negative feedback, formed basically by a third MOS transistor, an input resistance in the range of few tens of ohms is achieved. The input current ix, which is forced by the biasing circuitry, is conveyed to the output port Z by a source-coupling a complementary output pair of MOS transistors instead of using current mirroring. Since this coupled pair is biased with a constant tail current, the drain current changes will be equal but with opposite sign, regardless of their matching resulting in negative type CA with unity gain (iz=ix). In this design of the CA, programmability feature is achieved utilizing the output stages instead of the input stage. For example, when a second differential pair is connected in parallel, it provides two additional current outputs. When the two differential pairs are biased with different tail currents (IT1 and IT2), the new two outputs (izP1 and izN1) would approximately be given by
izP1=izN1=(IT2/IT1)1/2ix. (1)
Thus, the new outputs exhibit electronically controlled gains which can be programmed by adjusting the tail currents. Extra output currents with different gains can be obtained by adding more output stages with each stage providing both positive and negative signals.
These and other features of the present invention will become readily apparent upon further review of the following specification and drawings.
Unless otherwise indicated, similar reference characters denote corresponding features consistently throughout the attached drawings.
The programmable multi-gain current amplifier circuitry includes a power efficient current amplifier circuit (CA) 100 in which, as shown in
A second differential pair or set of transistors, including transistors M12-M13 is connected in parallel with second set of transistors that includes the transistors M3-M4 to provide two additional current outputs, for example. When the two pairs or sets are biased with different tail currents, it can be shown that the large signal current relationship will be given by:
where Vd=Vc−Vg3 (i.e., the differential voltage of the two source coupled pairs), K=0.5 μCoxW/L with μ as the surface carrier mobility, Cox is the gate oxide capacitance per unit area, and W and L are the width and length of the channel. Thus, for small signals Vd<<2[min. (IT1,IT2)/K]1/2, the relationship simplifies to (1) above. Thus, the new outputs would exhibit electronically controlled gains which can be programmed by adjusting the tail currents. Extra output currents with different gains can be obtained by adding more output stages with each stage providing both positive and negative signals. The original output current iz is of negative type and associated with unity gain, and, hence, it can be utilized in the local feedbacks needed to form the lossless integrators. This is considered as an additional inherent advantage that makes the structure of
The filter 200 in
The basic terminal characteristics of CAs can be expressed as Vx=0, Izp=αPijIx for positive outputs (the two currents have same direction with respect to the device) and Izn=αNijIx for negative outputs. The notations are selected such that the first number (i) in the subscript identifies the CAs whereas, the second number (j) denotes the output terminal assuming that unequal gains must be obtained from different output stages (e.g. αN01 is not used since it is assumed to be equal to αP01). Assuming unity gains for the negative outputs of the CAs employed in local feedbacks needed to form lossless integrators (their non-ideal effects are investigated infra), leads to the following transfer functions where Ilp is a low pass current output, Ibp is a band pass current output, Iin is a source current, s is a Laplace transform parameter representing the complex frequency jω, the alpha-P (αP) are programming constants applied to positive differential outputs of the current amplifier circuit, alpha-N (αN) are programming constants applied to negative differential outputs of the current amplifier circuit, the (αP) and (αN) programming constants taking on values corresponding to respective differential pair tail currents, ω0 is a programmable pole frequency, Q0 is a programmable pole quality factor, Glp is a programmable low pass gain, Gbp is a programmable band pass gain, and where C1R1 is a product of a first integration RC circuit connected to the current amplifier, and C2R2 is a product of a second integration RC circuit connected to a second current amplifier:
Thus, the filter exhibits a pole frequency, a pole quality factor and gains of:
It can be seen from (5) through (8) that all parameters are programmable. The parameter Qo for LPF can be tuned independently by changing αP11 while for BPF by changing αP11 and αP13 together. The parameter ωo can be programmed by changing αN12 and αP01 while αP11 and αN02 must be also tuned simultaneously to maintain constant Qo and gain, respectively. The gain of the LPF and BPF outputs can also be changed independently by αN02 and αP13, respectively. On the other hand, if each CA were assumed to have equal gain (i.e. αN12=αP11=αP13=α1 and αP01=αN02=α0), then only ωo can be tuned by changing α1 and α0 simultaneously, whereas all other parameters will typically be uncontrollable.
The filter of
It can be shown that the filter of
Therefore, the filter exhibits a pole frequency, a pole quality factor and a gain given by:
It can be seen from (12) through (16) that the proposed KHN has relatively complete independent control of various parameters. In this regard, the parameter ωo can be programmed by changing αN12 and αN21, simultaneously. Qo for LPF can be tuned independently by changing αP11 and for BPF by changing αP11 and αP13 together. The gain of LPF, BPF and HPF can also be adjusted independently by αN02, αP13 and αP22, respectively. The current gain αP01 can be set to unity as it is not used for tuning. When the programmability requirements are relaxed, the required output stages can be reduced. For example, if gain tuning is unneeded, then the gains of CA2 and CA0 can be selected to be equal whereas only two different gains would be needed from CAL
The non-ideal ac response of the filters can be found by considering the non-ideal effects of the CAs characterized by input parasitic impedance (Zx) and output parasitic conductance (Yz). Since the CA is designed to exhibit relatively low input impedance and relatively high output impedance, Zx and Yz are dominated by series resistance (rx) and parallel capacitance (Cz), respectively. For the proposed filters, the effect of Cz, for all CAs can be easily observed as they are in parallel with the passive capacitances and their values can be absorbed. Also, rx for all CAs, except that of CA2 of
Denoting these gain errors by βNi=1−εNi for respective CAs where |εNi|<<1, results in the following transfer functions for the TT topology, where a gain error |εNi|<<1, where Q is a programmable quality factor, ω0 is a programmable pole frequency, Ilp is a low pass current output, Ibp is a band pass current output, lin is a source current, the alpha-P (αP) are programming constants applied to positive differential outputs of the current amplifier circuit, alpha-N (αN) are programming constants applied to negative differential outputs of the current amplifier circuit, the (αP) and (αN) programming constants taking on values corresponding to respective differential pair tail currents, and where C1R1 is a product of a first integration RC circuit connected to the current amplifier, and C2R2 is a product of a second integration RC circuit connected to a second current amplifier, and s is a Laplace transform parameter representing the complex frequency jω:
Similarly, it can be shown that the non-ideal transfer functions for the proposed KHN, wherein a first CMOS current amplifier includes a band pass current output IBP, a second CMOS current amplifier includes a low pass current output ILP, and a third CMOS current amplifier includes a high pass current output IHP are given by:
Thus, it can be seen that various errors in βNi can lead to some deviations in the denominator's coefficients but without introducing any new pole. These errors can be compensated for through the tuning features. However, the main problems come from the error terms appearing in the numerators of the HPF and BPF due to various εi. Although these errors cannot be remedied, they are found to result in small deviations in low frequency bands. Plot 400 of
A CA based on the circuit 100 of
Plot 600 of
The linearity of the filter was determined by finding the input third-order intercept point (IIP3) determined by performing several intermodulation (IM3) tests using 800 kHz and 900 kHz signals. The filter can use a relatively low power consumption of 0.825 mW and can provide 5 functions simultaneously, for example. IIP3 estimation for in-band signals measured at the LPF output is found to be approximately −16 dBm (referenced to 50Ω). The spectrum of the output signal used to measure the dynamic range (DR) is shown in the plot 700a of
Embodiments of a programmable multi-gain current amplifier as a multi-output CA exhibit independently programmable gains and do not typically require a condition on the circuit topologies. Therefore, embodiments of a programmable multi-gain current amplifier are relatively more versatile than CM or CCIII counterparts, for example. Further advantages of the programmable multi-gain current amplifier can be demonstrated through the designs of two-integrator-loop biquads. Besides employing a true current-mode active element, the embodiments of the programmable multi-gain current amplifier filters use a relatively minimum number of active components of their kinds and, hence, can achieve optimum cost-efficient and relatively low power solutions. Experimental results obtained from 0.18 μm CMOS process of embodiments of the programmable multi-gain current amplifier also show an independent control of various parameters electronically. The design of a high-order filter, such as based on the biquad circuitry 200 of
It is to be understood that the present invention is not limited to the embodiments described above, but encompasses any and all embodiments within the scope of the following claims.
Number | Name | Date | Kind |
---|---|---|---|
6121830 | Manninen et al. | Sep 2000 | A |
6617926 | Casper et al. | Sep 2003 | B2 |
7425867 | Aemireddy et al. | Sep 2008 | B2 |
7583147 | Ruck | Sep 2009 | B2 |
7902924 | Cao | Mar 2011 | B2 |
7965118 | Seefeldt | Jun 2011 | B2 |
20070024369 | Cao | Feb 2007 | A1 |
20080106336 | Pera | May 2008 | A1 |
20100013557 | Cao | Jan 2010 | A1 |
Number | Date | Country |
---|---|---|
2008-211535 | Sep 2008 | JP |
I332750 | Nov 2010 | TW |
Entry |
---|
Jetsdaporn Satansup, Tattaya Pukkalanun and Worapong Tangsrirat,“Current-Mode KHN Biquad Filter Using Modified CFTAs and Grounded Capacitors” Proceedings of the International MultiConference of Engineers and Computer Scientists 2011 vol. II. IMECS 2011, Mar. 16-18, 2011, Hong Kong, 4 pages. |
Alzaher, H., and Ismail, M.: “Robust low-distortion wideband CMOS current-follower,” Electron. Lett., Dec. 1999, vol. 35, No. 25, pp. 2203-2204. |
Alzaher, H.: “A CMOS digitally programmable universal current-mode filter,” IEEE Trans. Cir. Syst. II, 55, (8), Aug. 2008, pp. 758-762. |
Alzaher, H., Alees, O., Tasadduq, N: “Programmable multi-gain current amplifier,” IEEE Xplore Digital Library, Nov. 2012 and Jan. 3, 2013, pp. 421-428, and Abstract. |
Kumngern, M., Moungnoul, P., Junnapiya, S., Dejhan, K.: “Current-mode universal filter using translinear current conveyors,” 5th Int. Conf. on Electrical Engineering/Electronics, Comp., Telec. and Information Tech., 2008, 2, pp. 717-720. |
Number | Date | Country | |
---|---|---|---|
20150028952 A1 | Jan 2015 | US |