Patent Application
20090072800
The present invention relates to circuitry and methods for providing a regulated output voltage as a voltage supply for use in an integrated circuit. The circuits and methods are particularly useful as on-board voltages in integrated circuits including analog and digital circuitry and the like. A stepped-down output voltage is provided from an input voltage using a highly efficient power regulator circuit based on a switched capacitor arrangement and incorporating digital modulation circuitry to rapidly regulate the output. The methods and circuits provide efficient regulation even under minimal or no-load situations, such as power down or sleep modes of operation in battery-operated applications. The circuits are compatible with semiconductor processing and may be incorporated into a system integrated circuit (IC), for example, when the integrated circuit is used to implement a portable device.
Systems implemented on highly integrated semiconductor circuits are increasingly important, particularly in producing circuits used to implement battery-operated devices, such as cell phones, portable computers, such as laptops, notebooks and PDAs, wireless e-mail terminals, MP3 audio and video players, portable wireless web browsers, and the like. Sometimes these integrated circuits are referred to as “SOIC” (for “systems on an integrated circuit”) or “SOC” (for “systems on a chip”) devices. SOC devices increasingly include analog circuitry as a portion of an integrated function that also includes a large digital portion, for example, a microprocessor, DSP, RISC processor, or another large digital portion, often including on-board data storage. As is known in the art, SOCs have portions that operate as part of a system, for example, a portion of the integrated circuit may implement a so-called analog front end or AFE, the analog portion of a wireless receiver or a transmitter. Sensors such as anti-locking braking sensors, pressure sensors, temperature sensors, and other input sensors are often implemented using a large analog circuit and an analog to digital converter circuit that is then coupled to an on-board processing unit that is formed from digital logic circuitry. These highly integrated circuits require a variety of on-board supply voltages. Often these on-board supply voltages are created as “stepped down” voltages from a higher off board supply voltage, although “step up” circuits are also known. Many portable devices now being produced are powered by a battery supply, which creates a first supply voltage, and various on-board voltages are needed within the integrated circuit to power different functions. Typically these batteries are rechargeable so that even when recharging, the voltage that operates the circuit is taken from the DC battery supply.
In the prior art, a regulated supply voltage output is often provided using an LDO or “low drop out” voltage regulator circuit. As is well known in the art, the LDO relies upon the control of a power FET, operated in a continuous-time fashion, coupled between the input voltage and a lower output voltage. Feedback is used to compare a reference voltage taken across a resistor at the output to a regulated reference voltage. If the output voltage is rising above the desired output (e.g., the power demanded by the load, in the form of load current, is dropping), the bias voltage to the gate of the FET is raised (if it is a PMOS device) or lowered (if it is an NMOS device). Conversely, if the output voltage is dropping below the desired output, the bias voltage to the gate of the FET is lowered (if it is a PMOS devise) or raised (if it is an NMOS devise) to provide additional current to the load so that the output voltage rises to the desired level. Thus, the LDO is a linear circuit.
The efficiency of the LDO circuits of the prior art is higher with a higher output voltage (relative to the input voltage) but can be quite low with lowered output voltages. In applications where the output voltage of the LDO is low, for example, less than 1V with a 1.8V input supply, the efficiency may be quite low, which means an increase in wasted power. This inefficiency leads to shorter battery life, or shorter battery operating time between charges for rechargeable devices, for example. Present circuits may often operate at even lower operating voltages and this inefficiency becomes more problematic as operating voltages for integrated circuits decrease.
Another prior art DC-to-DC converter solution is to use an off-chip switching regulator with off-chip components, or an on-chip controller with off-chip components. Switching regulators are known to be able to achieve increased efficiencies at lower output voltages than LDO circuits. As integration increases and the size of the devices is reduced, the use of off-chip regulators, and/or of off-chip components, becomes undesirable for several reasons. Further, the use of switching regulators with typical switching frequencies in wireless or cellular device applications can create tone frequencies and noise problems that are unacceptable.
The implementation of a fully integrated on-chip switching regulator, such as that of
Silicon area is the biggest problem in implementing the fully integrated synchronous buck converter of
A study was also made using the synchronous rectifier or buck converter described above with the smaller on-chip 10 nH inductor in discontinuous conduction mode but using a Pulse Frequency Modulation (PFM) control scheme instead of the PWM control scheme. Using this technique improved the efficiency, and the ripple obtained was similar to an LDO. The efficiency of such an approach at a 1V output voltage with a 10 milliamp load current was 61%. The efficiency at a 0.7V output voltage with a 7 milliamp load current was 60%. However the use of a PFM control scheme is not practical for many applications because the varying frequency of the switching circuitry produces an unpredictable ripple voltage spectrum at the output that is a function of the load current, and thus cannot be easily filtered. In wireless applications in particular, or in other noise-sensitive applications, using a PFM control scheme is highly undesirable due to its unpredictability.
DC-DC converters arranged without inductors are also known in the prior art. A switched capacitor circuit may be used to provide a stepped-up, stepped-down, or unity gain configuration DC-DC converter. In a switched capacitor circuit, a “flying” or charge transfer capacitor is alternatively switched between being coupled to an input voltage (typically a battery or other DC voltage) and to the load, while a holding or load capacitor is used to maintain the output voltage at the load. The discharge of the two capacitors at the output voltage terminal will provide output load current. The load or holding capacitor is discharged along with the flying capacitor during the “gain” portion of the cycle, and the charge transfer or flying capacitor is charged by the input supply during the “common” portion of the cycle. Switches are used to reconfigure the circuit in a clocked arrangement. Typically, two non-overlapping clocks are applied to the circuit to drive the switches.
The switched capacitor circuit is operated by clocking the switches labeled SW1A, SW1B, SW2A and SW2B with non-overlapping continuous waveforms or clocks φ1 and φ2. In this manner, the capacitor C1, the charge transfer or “flying” capacitor, is alternatively charged by the battery or discharged into the load, and the current is transferred to the load by the transfer capacitor CL. Simple digital circuitry is used to build the non-overlapping clocks φ1 and φ2 from a clock signal, and typical frequencies may vary from 10 kHz to 100 MHz.
a and 5b depict representative simplified circuitry illustrating the two phases of operation of the switched capacitor circuit of
As is also well known in the art, and as further described by Ramadass, the use of different capacitor topologies in the switched capacitor circuit can provide a wide range of gain ratios, including stepped-up and stepped-down ratios.
During the phase of operation when clock φ1 is high, switches SW1A, SW1B, SW1C and SW1D are closed. An examination of the circuit of
During the second phase of operation, when the clock φ1 is low, the clock φ2 is high (again, non-overlapping with φ1. Switches SW2A, SW2B and SW2C are closed. Examination of the circuit then reveals that capacitors C2 and C3 are now connected in series between the ground terminal and the output voltage Vout, and the series combination is in parallel with capacitor CL, which is discharging to the load. Thus, the output voltage Vout is now ⅔ of VBAT, instead of ½, as in the circuit of
To provide a practical regulated supply voltage, the output voltage Vout must be regulated. Feedback or hysteretic control circuitry is typically used to provide the control loop for the DC-DC converter function. Because the output voltage Vout depends on how many cycles there are (pulse frequency) and the length of time the clocks φ1 and φ2 are active (pulse width), modulation schemes are known in the prior art using voltage monitoring circuitry (typically a sense resistor at the output provides a proportional monitoring voltage) and op amp comparators or other comparator circuitry to detect when more, or less, current is needed to maintain the output voltage Vout at a desired DC level. When the comparator circuitry determines more current is needed by the load (output voltage Vout is falling), the control circuitry will either increase the switching frequency (in PFM mode) or increase the amount of time the charge transfer capacitor C1 is charged by the VBAT by modulating the pulse width of the clocks φ1 and φ2 (PWM) until the output voltage Vout increases. When the load is demanding less current, the output voltage Vout will rise, and the control circuitry will provide correspondingly less energy from the VBAT by reducing the switching frequency (in PFM mode) or by reducing the pulse width (in PWM mode) in order to maintain the output voltage Vout within a desired range.
DC-DC switched capacitor circuits using PFM or PWM control schemes are practical solutions for many applications. The efficiency of these circuits can be improved using various known schemes including “pulse skipping” for no load or light load conditions or “stand by” or “sleep” modes of operation for remaining conditions. Some known approaches also use gain hopping. Since the efficiency of the switched capacitor circuit increases when the “no load” voltage is very close to the desired output voltage Vout, changing the gain can improve efficiency. For example, duty cycle control is described in a paper entitled “Duty-Cycle Control Boosts DC-DC Converters,” by Cheong et al., IEEE Circuits and Devices Magazine, vol. 9, pp. 36-37, March 1993. However, the use of PFM control schemes creates tones or spurs in the output voltage Vout that are not predictable, making effective filtering of these tones impracticable. Thus, the use of this kind of control to achieve voltage regulation in a switched capacitor DC-DC converter is not preferable for many applications where the unpredictable tone frequency is not acceptable, for example, in certain analog, wireless, cellular or other transmitting and receiving technologies where the tone noise would make operation of the circuitry coupled to the output voltage Vout in the load impossible.
Another approach to regulate output voltage Vout of switched capacitor DC-DC converters uses fixed frequency approaches. U.S. Pat. No. 6,995,995, issued Feb. 7, 2006 to Zeng, et al., describes the use of segmented switches to regulate the output voltage Vout provided during the charging or gain phase of the clock cycle. In this segmented switch mode, the size of the switches (the current path) is modulated while the switching frequency is left constant. While these approaches will not have the same tone or noise unpredictability problems of the PFM control scheme, other problems arise. By using PWM control schemes or segmented switch modes (changing the current supplied to the capacitors by changing switch sizes to regulate the output voltage), these approaches rely on a fixed switching frequency regulation scheme. However, the loss mechanisms associated with switched capacitor circuits do not scale with the load power in these schemes. Also, effective output regulation in response to wide variations in load power cannot be obtained using switched capacitor DC converters regulated with these approaches.
A need thus exists for an area-efficient and cost-effective solution to the problem of providing a regulated DC output voltage from a DC input supply voltage with high efficiency even at low regulated voltages or no load conditions. The circuit must be practical to integrate on the same integrated circuit with other digital and analog circuitry, be silicon area efficient, and must provide small voltage ripple. The use of a fixed frequency of operation is needed to produce tone spurs or noise only at predictable, filterable frequencies in the output so as to be compatible with analog, cellular and wireless applications, while maintaining the output voltage within a narrow regulation variance over a wide range of load power demand conditions.
These and other problems are generally solved or circumvented, and technical advantages are generally achieved, by embodiments of the present invention, which provide circuits and methods of forming a digitally controlled switched capacitor circuit for outputting a regulated, stepped-down supply voltage to power a load circuit.
In one preferred embodiment, a switched capacitor DC-to-DC converter circuit is used to provide regulated output voltage to a load. A load capacitor is coupled to the output terminal. A segmented, switched capacitor matrix is used to provide a digital capacitance modulation circuit for the transfer capacitor. A pair of fixed frequency, non-overlapping clocking signals is provided to drive the switches in the switched capacitor circuitry. A hysteretic control loop is established using a voltage feedback path from the output. Control circuitry increases, or decreases, the output voltage by increasing, or decreasing, the transfer or flying capacitor in response to comparators that monitor the output voltage. This control loop uses a digital counter to add and subtract a value that in turn generates proportionally weighted control lines used to modulate the transfer capacitance.
In another preferred embodiment, a fine bank of segmented switched capacitors and a coarse bank of segmented switched capacitors are provided. In one preferred embodiment, the coarse bank implements a binary weighed scale of unit capacitance values. The fine bank implements a plurality of fractional values of the unit capacitance values. Control signals are generated to decrement, or increment, the weight of the transfer capacitance during operation of the regulator. If the output voltage is a great deal from the desired value, a coarse mode of adjustment is used to rapidly increment, or decrement, the transfer capacitance, thereby regulating the output rapidly. If the output voltage is near the desired voltage, a fine adjustment is made to slowly increment, or decrement, the transfer capacitance, thereby making efficient regulation of the output voltage.
In another preferred embodiment, a low load or no load mode of operation is provided. Circuitry to replace the fixed frequency clock pulses with a lower variable frequency signal is used to enable the switching capacitor converter to efficiently regulate the output voltage when the load is not consuming power or is consuming very little power, such as for sleep or stand by modes of operation of battery-powered devices.
In one preferred embodiment, the capacitors are implemented using N-poly N-well capacitors. In another preferred embodiment, the capacitors are implemented using metal-dielectric-metal capacitors above the substrate.
Additional preferred embodiments provide control circuitry to couple unused capacitors in the segmented capacitor switch matrix to the load capacitor to further increase the value of the load capacitor and to reduce voltage ripples at the output terminal, thereby increasing circuit performance.
A preferred method of the present invention provides a load capacitor coupled between an output voltage and a ground reference potential. A switched capacitor circuit including a transfer capacitor coupled to the load capacitor and selectively coupled to a voltage supply and selectively coupled to the ground reference potential is provided to clock the switch capacitor at a fixed frequency to alternatively charge the transfer capacitor by coupling it between the battery supply and the output terminal and then to discharge the transfer capacitor by coupling it between the output terminal and the ground reference potential. The method also monitors the output voltage and determines when it is above a first threshold and determines when it is below a second threshold, and digitally modulates the value of the transfer capacitor responsive to the monitoring of the output voltage to maintain the output voltage between the second threshold and the first threshold.
In additional preferred methods, providing the transfer capacitor further comprises providing a switch matrix of segmented capacitance circuits that are selectively enabled responsive to control lines, and providing control circuitry for increasing the transfer capacitance to increase the output voltage and for decreasing the transfer capacitance to decrease the output voltage by enabling and disabling segments of the segmented capacitance circuits in the switch matrix, responsive to the monitoring circuitry. In additional preferred methods, providing the control circuitry further provides a coarse mode of adjustment and a fine mode of adjustment, wherein, in the coarse mode of adjustment, the transfer capacitor is adjusted in increments or decrements of a unit capacitor value each cycle and, in the fine mode of adjustment, the transfer capacitance is adjusted in increments or decrements of fractions of the unit capacitor value.
In additional preferred methods of the invention, providing segmented capacitor values for the transfer capacitor further provides a plurality of weighted switched capacitor circuitry that implements a binary scale. In further preferred methods, providing the segmented capacitor values further provides switched capacitor circuitry that is weighted in fractions of a unit capacitor value.
In further preferred methods, providing segmented capacitance circuits further provides switched capacitor circuitry that selectively provides a first gain and a second gain. In another preferred method, the first gain is a first ratio and the second gain is a second different ratio, and control circuitry is provided to perform as a stepped-down DC-DC converter by outputting an output voltage that is less than the input supply voltage.
Additional preferred methods provide digital capacitance modulation circuitry by providing digital control logic to control the switches and to clock the transfer capacitance circuitry. In another preferred method, a low power mode of operation is provided using pulse frequency modulation to lower the switching frequency when the control circuitry and the output voltage monitoring indicate there is no power demanded at the output. In another preferred method, the low power mode of operation is entered when the control circuitry indicates the segmented transfer capacitance should go below its lowest setting. In another preferred method of operation, the low power mode of operation is needed when the control circuitry indicates that the transfer capacitance should be increased above its lowest setting over several cycles when the low power mode is enabled.
In an additional preferred method, a coarse mode of adjustment is entered at power up. In an additional preferred method, a coarse mode of operation is entered into when the output voltage falls below a predetermined threshold voltage set by a user. In another preferred method, the coarse mode of operation is ended and a fine mode of operation begins when the control circuitry detects an output voltage above the first threshold on one clock cycle and below the second threshold on a subsequent clock cycle. In the fine mode of operation, adjustments to the transfer capacitance are made in fractions of a unit capacitance.
The foregoing has outlined rather broadly the features and technical advantages of the present invention in order that the detailed description of the invention that follows may be better understood. Additional features and advantages of the invention will be described hereinafter, which form the subject of the claims of the invention. It should be appreciated by those skilled in the art that the conception and specific embodiments disclosed might be readily utilized as a basis for modifying or designing other structures or processes for carrying out the same purposes of the present invention. It should also be realized by those skilled in the art that such equivalent constructions do not depart from the spirit and scope of the invention as set forth in the appended claims.
For a more complete understanding of the present invention, and the advantages thereof, reference is now made to the following descriptions taken in conjunction with the accompanying drawings, in which:
a and 12b depict efficiency graphs for preferred embodiments of converter circuits incorporating the features of the invention, wherein
The making and using of the presently preferred embodiments are discussed in detail below. It should be appreciated, however, that the present invention provides many applicable inventive concepts that can be embodied in a wide variety of specific contexts. The specific embodiments discussed are merely illustrative of specific ways to make and use the invention, and do not limit the scope of the invention.
Preferred embodiments of the invention provide a DC-DC converter for outputting a highly regulated DC output voltage that is user selectable, using an area-efficient and semiconductor process compatible integrated switched capacitor circuit and associated control circuitry. The circuit is easily integrated into existing commercial semiconductor processes, uses only on-chip components, and is free from the frequency spectrum problems and loss problems of prior art switched capacitor circuitry. Preferred embodiments of the invention provide additional increases in efficiency by using a variable DC gain architecture, and by switching from a fixed frequency mode to a PFM or pulsed frequency modulation scheme under “stand by” or sleep conditions where the load is consuming little or no power.
In preferred embodiments, digital control circuitry provides automatic detection and regulation of the output voltage, automatic switching to the PFM mode, and in other preferred embodiments, automatic switching between a coarse and a fine adjustment mode to provide rapid response to rapid changes in output power loading and better regulation in a narrow regulation window when the load power is constant. Additional preferred embodiments provide a method for further reducing the ripple current obtained from the circuit by switching unused capacitors from the transfer capacitance to the load to still further improve performance without any increase in silicon circuit area.
The benchmark or target performance criterion considered for the preferred embodiments of the invention and methods include providing an integrated DC-DC step-down circuit and a method for converting a supply voltage to a lower output voltage with excellent regulation. The circuit and methods are shown to be more power efficient than an LDO solution as is known in the prior art, to be silicon area efficient, and to have similar performance to an LDO circuit, e.g., characteristics such as power supply rejection ratio, ripple, and other performance characteristics should be similar to the performance of the LDO. A fixed frequency mode of regulation of the switched capacitor converter is used in the preferred embodiments to eliminate the tone or spur problems of PFM and PWM approaches of the prior art converters.
In order to maximize the efficiency of a converter using switched capacitors, the gain should be adjusted so that the output load voltage is close to the “no-load” voltage. The efficiency that can be obtained from a switched capacitor converter is limited to a ratio of the load voltage to the no-load voltage of the gain setting being used.
The efficiency can be shown to be limited to:
ηlim=(1−ΔV/VNL). (1)
Thus, to improve efficiency, the voltage ΔV (the difference between no-load voltage VNL and the load voltage) should be minimized. In preferred embodiments of the present invention, in addition to other control circuitry, the gain of the converter circuit is selected from at least two settings, for example a ½ and a ⅔ VO/VBAT supply gain setting, to place the no-load output voltage VNL as close to the output voltage desired at the load VL as possible. Once the gain is set for a particular application, preferably by automatic selection based on the output voltage desired by the user, it is typically not changed (unless the output voltage setting selected is also changed). Each of the switched capacitor circuits described below in discussing the preferred embodiments has this selectable two gain feature.
The overall design approach of the preferred embodiments may be better understood with an understanding of the power delivery equations that describe the load power delivered by a switched capacitor converter circuit.
Taking for example the switched capacitor circuit of
E
L=8CBVL(VNL−VL) (2)
and that the power delivered to the load for this gain setting is:
P
L
=E
L
f
S=8CBVLfS(VNL−VL) (3)
Thus, in order to modify the power to the load (e.g., to regulate the output voltage Vout as the load current demand increases or decreases), as is readily seen from Equation (3), the available adjustment settings are EL, fS, and the value of the transfer capacitance CB. As described above in the background, some prior art control methods approach the regulation by using a variable switching frequency modulation or PFM to adjust the switching frequency fS. Alternatively, the energy delivered to the load may be adjusted, for example, using duty cycle control approaches such as PWM control to adjust the pulse width or switch segment control to modulate the output current being supplied to the load. The problems with these approaches are discussed above.
The preferred embodiments of the present invention use a novel approach to voltage output regulation. In contrast to the control approaches of the prior art, in the preferred embodiments of the present invention, the transfer capacitance value is dynamically adjusted to regulate the power to the load. This is accomplished by using a technique referred to as digital capacitance modulation (DCM). In additional preferred embodiments, this approach is combined with a DC gain adjustment to further set the overall circuit efficiency as high as possible. Other preferred embodiments provide rapid adjustments to the transfer capacitance in certain situations, and a special low load operating mode to increase efficiency even further.
The approach used in a first preferred embodiment is to segment the transfer capacitance into a switched matrix of capacitance banks, and to use a hysteretic control loop approach to dynamically adjust the transfer capacitance (by turning different switches on and off to control the transfer capacitance value), while maintaining a constant switching frequency fS, in order to regulate the power output.
The many advantages of such an approach to providing a regulated DC voltage may be better understood if the loss mechanisms for the converter circuit are considered. The efficiency of a switched capacitor converter circuit is impacted by at least three primary loss mechanisms: (1) a linear conduction loss that occurs due to the act of charging the transfer capacitor through a switch; (2) loss caused by the bottom plate parasitic capacitance due to the bottom plate of the charging capacitor being shunted (switched) to ground every cycle; and (3) loss due to the switches themselves, that is loss due to charging and discharging the switches (typically N-MOS, P-MOS FETs, or transfer gate switches formed from both an N-MOS and a P-MOS FET).
An expression for the overall efficiency may be obtained:
In the overall efficiency equation (4), the first term or preterm is due to conduction losses. The first term in the denominator is due to bottom plate loss characteristics, and the second term is due to switching losses. Examination of the terms reveals that the efficiency is apparently not dependent on the area of the capacitors, and the switching loss can be reduced by lowering fS, which may be done by increasing the capacitor area by the same amount (due to term KSfS).
The approaches of the prior art, e.g., pulse frequency modulation (PFM) and duty cycle regulation approaches (PWM), or segmented switch regulation, all suffer from loss mechanisms that do not scale well with lower output power. In contrast, the digital capacitance modulation (DCM) approach of preferred embodiments of the present invention disclosed herein provides loss mechanisms that scale well with load power. As the output power is decreased in switched capacitor circuits using the DCM technique, the number of capacitor banks used will also decrease and the switches required will decrease, so that losses due to switching and bottom plate parasitic effects will also decrease. Further, the use of capacitor ratios to regulate the output voltage produces an output voltage regulation that is process independent, that is, process variations in the capacitors will tend to cancel out. The prior art approaches tend to be very dependent on process and thus subject to process variation problems. This is not so with the preferred embodiments of the present invention.
During the second phase of operation, when clock φ1 is low, clock φ2 is high (again, non-overlapping with clock φ1). Switches SW2A and SW2C are closed. Switches SW2B and SW2D are closed when the selection has been made to place the gain at ½ (G1BY2) as indicated by the signal label φ2, G1BY2 in
In operation, control switching signal φ2, G1BY2 of
One skilled in the art will recognize that the weighting scheme of
However the segmentation of
The multiplexer 21 enables the circuit 11 to implement an additional preferred embodiment using PFM regulation control. This is done to increase the efficiency of the circuit 11 at very low loading levels, such as when no load circuit power is used, as in sleep or stand-by modes used in battery-powered devices. This PFM feature is not required in other preferred embodiments and so multiplexer 21 may be omitted for those embodiments. The clock generator 23 receives the clock input, for example a 60 MHz clock is indicated, however the invention is not limited to any specific frequency and other frequencies may be used. The clock generator 23 generates clocks φ1 and φ2 as described above, which are non-overlapping clock pulses. Add/subtract block 19 outputs the “enable” signals to the switch matrix 25. As will be further described, one of the “enable” signals goes to each of the segments of the segmented transfer capacitor in switch matrix 25, each enabling one of the weighted segments to be included in the transfer capacitance as shown in
Logic block 17 is implemented using digital logic circuitry in a control function. Known techniques such as digital synchronous state machines, programmable micro controllers, processors, DSPs, RISC cores, gate arrays, ROMs, EPROMS, flash and the like may be used to implement this block. Typically a clocked state machine and digital logic circuitry will be used.
Comparators 13 and 15 are used to set the bounds needed to implement the hysteretic control loop. An allowable voltage threshold variance ΔV is provided. A reference voltage VREF is also provided, for example from a bandgap reference. When the sensed feedback voltage from the output VFB is less than the trigger voltage (VREF−ΔV), the comparator 13 puts out a GO_UP signal. When the sensed voltage VFB is greater than the trigger voltage (VREF+ΔV), comparator 15 puts out a GO_DOWN signal. These comparisons are used with logic block 17 to provide the control loop of the circuitry.
Comparator 27 provides additional control. When the observed output voltage VFB is less than the voltage difference (VREF−VOFF), the output voltage has fallen rapidly and is quite a bit lower than desired. In order to enable the circuit to rapidly resume proper output regulation, the logic block 17 will respond to the COARSE_EN signal by rapidly adjusting the transfer capacitance using the enable lines enMSB<0:3> to the switch matrix 25.
The operation of the preferred embodiment of
Logic block 17 receives the signals GO_UP and GO_DOWN from the comparators 13 and 15, respectively. When the observed voltage VFB is below the hysteretic band, the comparator 13 puts out the GO_UP signal. The logic block 17 then outputs a signal to the add/subtract block 19 to cause it to add to the enable value that determines the value of the transfer capacitance in the switch matrix 25. The circuit 11 then causes the add/subtract block 19 to increment either the output enMSB <0:3> (for coarse mode control) or the output enLSB <0:1> (for fine mode control) to the switch matrix 25 to increase output voltage Vout. When the observed voltage VFB is above the hysteretic band, the comparator 15 puts the GO_DOWN signal out to logic block 17. The logic block 17 then outputs a signal to the add/subtract block 19 to cause it to decrease the value enMSB <0:3> (for coarse mode control) or enLSB <0:1> (for fine mode control) to the switch matrix 25. This will cause a decrease in the output voltage Vout because the enable signals determine the active segments in the segmented transfer capacitance of switch matrix 25. In this manner, the circuit 11 attempts to maintain the output voltage Vout in the desired hysteretic range over a variety of conditions of load power demand.
The preferred embodiment of the converter circuit of
When the output voltage Vout falls rapidly, the load has suddenly demanded an increase in power. If the output voltage Vout falls beneath the value VFB−(VREF−VOFF), the signal COARSE_EN is active, indicating a coarse mode of regulation is enabled. In this mode, the adjustments made by the add/subtract block 19 will be made by incrementing (GO_UP signal is active) or decrementing (GO_DOWN signal is active) the four-bit control word enMSB<0:3>. This mode of operation allows for fast adjustment of the transfer capacitance in increments or decrements of the unit value 1 CB, thereby allowing the circuit 11 to rapidly settle to the desired output value.
Once coarse mode regulation is achieved and the feedback voltage VFB is within the hysteretic range, the fine mode of operation is used and signal COARSE_EN is no longer active. In fine mode, the transfer capacitance of switch matrix 25 is adjusted, in response to a GO_UP or GO_DOWN signal, in increments or decrements of value 2CB/7, using the control word bits enLSB<0:1>. Taking the two control signals enMSB and enLSB together, the six-bit control word is adjusted incrementally to achieve voltage regulation. This fine step size, used when the output voltage Vout is nearly perfectly regulated and when the output power load is not changing rapidly, enables the circuit 11 to settle within narrow hysteretic bands without unwanted oscillations.
The transition between the coarse mode and the fine mode of operation may be automatically determined using the existing signals. When the output voltage Vout is nearly regulated, the logic block 17 may detect a situation where there is a GO_UP signal followed immediately by a GO_DOWN signal. In this situation the output voltage Vout is very nearly regulated at the required value, that is, the load voltage is transitioning from a falling to a rising mode, a local minimum has been reached, and the fine mode of regulation may be efficiently used. Another transition from coarse mode to fine mode is made when the four bits of the control word enMSB<0:3> are all zero and a GO_DOWN signal subsequently occurs. In this case, the load power is very low and again the fine mode of regulation may be efficiently used.
The signal COARSE_EN is active when the converter transitions from a fine mode of regulation to a coarse mode. This happens, for one example, when the circuit 11 needs to rapidly adjust the output voltage Vout, for instance at power up. The big increase in load power in these situations leads to a fall off in the output voltage Vout so that feedback voltage VFB falls below the trigger value VREF−VOFF. The change to coarse mode enables the circuit 11 of
The use of two modes of regulation is advantageous. If only the coarse mode is used, then the selected hysteretic band has to be wide enough to prevent the circuit from oscillating from a GO_UP signal to a GO_DOWN signal on alternating cycles when the output voltage Vout is more or less regulated. These oscillations would appear on top of the regulated DC output voltage at a frequency which is a complex function of load current, the value of the capacitors CL and CB, and the output voltage Vout. Oscillations produce unwanted noise or tones that signaling applications, such as cellular or wireless transceiver circuitry, cannot tolerate. The introduction of the fine mode of operation in the preferred embodiment prevents this unwanted oscillation and allows a narrower hysteretic band to be used efficiently for better regulation of the output voltage Vout.
The circuit 11 of
To further enhance the operation of the circuit 11 in these low load or no-load conditions, the preferred embodiment of
The transition to this low load mode of operation may be automatically determined by the logic block 17 from the existing signals. When the enMSB and enLSB control words are all zero, and a GO_DOWN signal is output, the output voltage Vout is continuing to rise even though the flying capacitance is at its lowest available value (CB/7 in this example). Thus the load is not demanding any current. When the logic block 17 detects this situation, it will transition to the low load mode of operation and output a zero on the control line DCM/PFM, which causes the clock signal CLK to be replaced with the GO_UP signal.
As the output voltage Vout falls, the circuit 11 will periodically clock the non-overlapping clock signal CLK when GO_UP is high. If the GO_UP signal repeats for several cycles, this indicates the load is actively demanding current and the logic block 17 then transitions back to the normal, DCM, mode of operation by outputting a “1” on the logic signal DCM/PFM.
Normally the use of PFM regulation, as discussed in the background section, produces unwanted tones or spurs in the output. However, in cases where the load is not operating, this noise source is acceptable, and the increase in efficiency is simply obtained by the logic block 17 observing already existing signals. Once operation of the load resumes, the control circuitry will transition the operation back to the normal DCM mode of operation.
Similarly, block 81 of weight 4× is enabled when the control signal enMSB<1> is high. Examination of
Those skilled in the art will recognize that this implementation is but one possible implementation, and other implementations for the switch matrix 25 are contemplated as part of the present invention and are within the appended claims.
A further preferred embodiment of the switched capacitor stages is now presented which, when used with a circuit such as the circuit 11 embodiment illustrated in
The advantage of this additional preferred embodiment is that increasing the load capacitor value CL reduces the output voltage ripple as the load voltage falls. By using the same switches that are already used to enable the capacitor segment, the total area occupied by the on-chip converter circuit is not increased for this additional enhancement.
In one exemplary implementation of the preferred embodiment of
A preferred method of the present invention thus provides a load capacitor coupled between an output voltage and a ground reference potential. A switched capacitor circuit including a transfer capacitor coupled to the load capacitor and selectively coupled to a voltage supply and selectively coupled to the ground reference potential is provided to clock the switch capacitor at a fixed frequency to alternatively charge the transfer capacitor by coupling it between the battery supply and the output terminal and then to discharge the transfer capacitor by coupling it between the output terminal and the ground reference potential. The method also monitors the output voltage and determines when it is above a first threshold and determines when it is below a second threshold, and modulates the value of the transfer capacitor responsive to the monitoring of the output voltage to maintain the output voltage between the second threshold and the first threshold.
In additional preferred methods, providing the transfer capacitor further comprises providing segmented capacitance circuits that are selectively enabled responsive to control lines, and providing control circuitry for increasing the transfer capacitance to increase the output voltage and for decreasing the transfer capacitance to decrease the output voltage by enabling and disabling segments of the segmented capacitance circuits responsive to the monitoring circuitry. In additional preferred methods, providing the control circuitry further provides a coarse mode of adjustment and a fine mode of adjustment wherein, in the coarse mode of adjustment, the transfer capacitor is adjusted in increments or decrements of a unit capacitor value each cycle and in the fine mode of adjustment, the transfer capacitance is adjusted in increments or decrements of fractions of the unit capacitor value.
In additional preferred methods of the invention, providing segmented capacitor values for the transfer capacitor further provides a plurality of weighted switched capacitor circuitry that implements a logarithmic scale. In further preferred methods, providing the segmented capacitor values further provides switched capacitor circuitry that is weighted in fractions of a unit capacitor value.
In further preferred methods, providing the segmented capacitance circuits further provides switched capacitor circuitry that selectively provides a first gain and a second gain. In another preferred method, the first gain is a first ratio and the second gain is a second different ratio, and control circuitry is provided to perform as a stepped-down DC-DC converter by outputting an output voltage that is less than the supply voltage.
Additional preferred methods provide digital capacitance modulation circuitry by providing digital control logic to control the switches and to clock of the transfer capacitance circuitry. In another preferred method, a low power mode of operation is provided using pulse frequency modulation to lower the switching frequency when the output voltage monitor indicate there is no power demanded at the output. In another preferred method, the low power mode of operation is entered when the control circuitry indicates the segmented transfer capacitance should go below its lowest setting. In another preferred method of operation, the low power mode of operation is needed when the control circuitry indicates that the transfer capacitance should be increased above its lowest setting over several cycles when the low power mode is enabled.
In an additional preferred method, a coarse mode of adjustment is entered at power up. In an additional preferred method, a coarse mode of operation is entered into when the output voltage falls below a predetermined threshold voltage set by a user. In another preferred method, the coarse mode of operation is ended and a fine mode of operation begins when the control circuitry detects an output voltage above the first threshold on one clock cycle and below the second threshold on a subsequent clock cycle. In the fine mode of operation, adjustments to the transfer capacitance are made in fractions of a unit capacitance.
The advantages of a practical circuit incorporating preferred embodiments of the invention may be seen from
As can been seen in
However using the N-poly N-well capacitors as circuit elements uses a substantial amount of silicon area. In a typical semiconductor process this represents a 0.082 mm area penalty over the LDO. This can be further improved upon by using, in another preferred embodiment, so called “over IC” capacitors formed above the substrate in the metal dielectric layers. For example, MiM (metal, insulator, metal) or MoM (metal, oxide, metal) capacitors can be used, and multilayer capacitors can be used. In one example implementation, capacitors of 50 femtoFarads/um2 were obtained.
b illustrates the efficiency results obtained in this example of a preferred embodiment digital capacitance modulated circuit, again implemented in an exemplary 65 nanometer semiconductor process. In
Thus advantages of the use of the preferred embodiments are that the method and circuitry provides a semiconductor process compatible, on-board, DC-DC converter circuit that is efficient at a wide range of operating voltages, is efficient under power down or sleep mode, or no-load, conditions, and rapidly adjusts to changes in load power. Efficiency obtained is greater than an LDO circuit of the prior art. If high density capacitors are used, an area savings in silicon is also obtained over the prior art. No external components are required. Losses due to the operation of the circuitry scale with the load voltage and the circuit remains efficient even at low load voltages. Fixed frequency switching is used in active operation so that the unpredictable frequency spectrum tones produced by the PFM modulation switched capacitor circuits of the prior art are not present.
Although the present invention and its advantages have been described in detail, it should be understood that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the invention as defined by the appended claims. For example, it will be readily understood by those skilled in the art that the methods may be varied while remaining within the scope of the present invention.
Moreover, the scope of the present application is not intended to be limited to the particular embodiments of the methods and steps described in the specification. As one of ordinary skill in the art will readily appreciate from the disclosure of the present invention, processes, or steps, presently existing or later to be developed, that perform substantially the same function or achieve substantially the same result as the corresponding embodiments described herein may be utilized according to the present invention. Accordingly, the appended claims are intended to include within their scope such processes or steps.
This application claims the benefit of U.S. Provisional Application No. 60/972,076, filed on Sep. 13, 2007, entitled Switched Capacitor Step-Down Power Converter to Replace Linear Regulators in Embedded Power Management Applications, which application is hereby incorporated herein by reference.
| Number | Date | Country | |
|---|---|---|---|
| 60972076 | Sep 2007 | US |
