System for distributing phase current of a current-mode DC-DC converter and improving dynamic response thereof with active phase count.

IN604791BActive Publication Date: 2026-10-01INDIAN INSTITUTE OF TECHNOLOGYKHARAGPUR
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
IN202231010977
Authority / Receiving Office
IN · IN
Patent Type
Patents
Current Assignee / Owner
Filing Date
2022-03-01
Publication Date
2026-10-01

AI Technical Summary

Technical Problem

Current-mode DC-DC converters with active phase count (APC) face challenges in achieving predictable current distribution among phases, leading to arbitrary current distribution and degraded transient responses, which are critical for modern application processors with high power demands and stringent load current slew-rates.

Method used

A system that includes a phase current distribution unit with a summing amplifier and flash ADC to selectively turn on active phases, ensuring equal current distribution among phases and improving transient responses by forcefully turning phases on/off during extreme load transients, using a slew rate enhancement block with hysteretic comparator and pulse generating circuit to manage load steps and overshoots/undershoots.

Benefits of technology

The proposed solution achieves nearly theoretical limits for undershoot and overshoot in transient responses and maintains high efficiency across the entire load current range, ensuring predictable current distribution and improved dynamic load transient response.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

Abstract Title: System for distributing Phase Current of a Current-Mode DC-DC Converter and Improving Dynamic Response thereof with Active Phase Count. The present invention discloses a system for improving current-mode DC-DC converter operation involving high load current slew-rate and fast dynamic load transient response comprising a phase current distribution unit operable with the current-mode DC-DC converter for extending the current-mode DC-DC converter from single-phase to APC configuration. The phase current distribution unit is operatively connected with plurality of power stages, whereby the phase current distribution unit includes a summing amplifier followed by a flash ADC for selectively turning on active phases in the power stages in accordance with load current range for sharing the load current among the turned on active phases and maintain overall power efficiency in entire load range.
Need to check novelty before this filing date? Find Prior Art

Description

Field of the invention:The present invention relates to fixing current distribution in different active phases of a current-mode DC-DC converter with active phase count (APC). More specifically, the present invention is directed to develop an improved current mode DC-DC converter with a novel system for fixing distribution of load current in the different active phases and prohibits any arbitrary distribution of the current. The present invention is also directed to improve transient responses of the current-mode DC-DC converter while maintaining high efficiency throughout entire load current range and achieves almost the theoretical limit of the undershoot and the overshoot in the current-mode DC-DC converter with APC including forcefully and selectively turning ON-OFF the phases during the extreme load transient.Background of the invention:With the rapid advancement of the scaling of the integrated technology, modern application processors (APs) for high-performance battery-operated portable electronics are implemented in the lower technology node, which led to a dramatic increase in its clock frequency, the number of cores, and the total power dissipation. Also, these modern APs demand a tightly regulated supply voltage even with an unprecedented load current slew rate under dynamic operation. As a result, the design of the DC-DC converter that is targeted to supply the power to these APs becomes very challenging. As the dynamic range of the output power is increased, the converter has to maintain a high power efficiency over a wide range. The converter has to meet the unprecedented load current slew-rate (>1A / ns) when it switches from the sleep mode to the active mode (or vice-versa) as reported in the article no SNVA589 by J. Y Wong. At the same time, it also has to maintain a tight load regulation. Therefore, the DC-DC converter powering these APs must have a fast dynamic load transient response. Various control methods for the DC-DC converter are available in the literature. Among them, the current-mode DC-DC converter topology provides a fast dynamic response and it is widely used to meet a stringent dynamic response. The block diagram of a classical current-mode DC-DC buck converter topology is shown in Figure 1. To meet the high power demands for these APs while maintaining a high power efficiency over a wide load range, active phase count (APC) in the current-mode DC-DC buck converters is becoming more prevalent. The current-mode DC-DC buck converter with conventional APC has a parallel set of power stages with its individual inductor and power MOSFET switches of appropriate size as shown in the Figure 2.As mentioned in the article, "Benefits of a multiphase buck converter" in Texas Instruments, Tech article no SLYT449 by D. Baba and the article, "Multiphase Buck Design from Start to Finish (part1)" in Texas Instruments, Tech by C. Parisi; there are various advantages with the APC, such as reduction of the size of the input and the output capacitors, improvement in thermal performance and the efficiency at high load current. Unlike single phase, the active phases handle only a portion of the total output current, and hence the inductors and the MOSFETs selected, are having less thermal stress. However, to maintain the high power efficiency over the wide dynamic power range, the switch in the APC are having unequal size. This leads to an uneven distribution of the current in all the active phases. What is more, if there is no control mechanism to get a predictable current distribution, the active phases carry current arbitrarily.US 6,278,263 B1 reports a technique to avoid the arbitrary distribution of the phase currents, wherein the current sensed from each of the phases are summed up and averaged. An error signal is generated by comparing the averaged current with the individual phase currents. The error signal is further added with the voltage control loop to generate a PWM control signal for the respective phase. However, the addition of the current loop with the voltage control loop makes the control susceptible to the load variations. US 6,670,794 B1 and US 2009 / 0051335 proposes a technique in which the voltage reference signal is adjusted and controlled with the summed-up phase current. However, this also requires multiple loops with other transient degrading blocks such as summers and scalers.US 2009 / 0058379 uses a digital approach towards the present problem that has a tradeoff between the accuracy and the no of bits.In US 10,338,615, the average current information of each of the phases are sensed using the simplified R-C based filter with low power dissipation. This is further compared with the average phase current in each of the phases to generate the error-correcting signal. However, the technique requires a time constant matching of the R-C filter and that of the inductor with the direct current resistor. Therefore, in all the above-mentioned works, the current sensing and the processing time are so large that it degrades the transient response. Several articles in the literature also have targeted improving the transient response. For example in the article, "An 82.4% efficiency package-bond wire-based four-phase fully integrated buck converter with flying capacitor for area reduction," in IEEE Int. Solid-State Circuits Conf. Dig. Tech. Papers, Feb. 2013, pp. 362-363 by C. Huang and P. K. T. Mok and the article, "Capacitor-current-sensor calibration technique and application in a 4-phase buck converter with load-transient optimization," in IEEE Int. Solid-State Circuits Conf. Dig. Tech. Papers, Jan. / Feb. 2016, pp. 228-229 have proposed the APC technique to adjust the number of the active phases according to the instantaneous load condition. However, they have used the conventional PWM control, which unfortunately suffers from inherent bandwidth limitation and thus limits the transient response making them unsuitable for present demand. To improve this, the article titled, "A 6A 40MHz Four-Phase ZDS Hysteretic DC-DC Converter with 118mV Droop and 230ns Response Time for a 5A / 5ns Load Transient" in ISSCC, pp. 80-81, Feb. 2014 by M. K. Song, J. Sankman and D. Ma uses the fast responding current mode control and proposes the technique therein for delay less response during the load step-up. Nevertheless, the load step-down still suffer the delay.The article titled "A 25-MHz Four-Phase SAW Hysteretic Control DC-DC Converter With 1-Cycle Active Phase Count" in IEEE Journal of Solid-State Circuits, vol. 54, no. 6, pp. 1755-1763, June 2019 by Bumkil Lee, Min Kyu Song, Ashis Maity, and D. Brian Ma implements a hysteretic control with adaptive adjustment of the hysteretic band in order to improve the load transients and establish the APC in one switching cycle by sensing the current internally. However, in the presented scheme adding the ripple in the hysteretic band deteriorates the DC load regulation performance.Therefore, based on the review of the presently available state of art solutions, there is a need of distributing the phase currents without degrading the transient performance of a current mode DC-DC buck converter with APC.Objects of the invention:It is thus the basic object of the present invention is to provide an improved current mode DC-DC converter which can meet the higher power demands and stringent load current slew-rate of modern Application ProcessorsAnother object of the present invention is to provide a system for efficiently extending the current-mode DC-DC converter from the single-phase to APC configuration and fixing distribution of load current in the different active phases including prohibiting any arbitrary distribution of the current.Another object of the present invention is to provide a system to improve transient responses of the current-mode DC-DC converter while maintaining high efficiency throughout entire load current range and achieve almost the theoretical limit of the undershoot and the overshoot in the current-mode DC-DC converter with APC including forcefully and selectively turning ON-OFF the phases during the extreme load transient.Summary of the invention:Thus according to the basic aspect of the present invention there is provided asystem for improving current-mode DC-DC converter operation involving high load current slew-rate and fast dynamic load transient response comprising said current-mode DC-DC converter; aphase current distributionunit operable with the current-mode DC-DC converterfor extending the current-mode DC-DC converter from single-phase to APC configuration; said phase current distribution unit is further operable with equal sized APC and multiphase DC-DC converter; said phase current distribution unit is operatively connected with plurality of power stages, whereby the phase current distributionunit includes a summing amplifier followed by a flash ADC for selectively turning on active phases in the power stages in accordance with load current range for sharing the load current among the turned on active phases and maintain overall power efficiency in entire load range.In a preferred embodiment of the present current-mode DC-DC converter, the power stage comprises atleast one power transistor capable of delivering a large amount of current to the load, whereby gate terminal of said power transistor is driven by a driver block.In a preferred embodiment of the present current-mode DC-DC converter, the power stages including their individual inductors are connected in parallel between input and output voltage nodes with sense filter network, whereby said sense filter network sense average inductor current information in each active phase.In a preferred embodiment of the present current-mode DC-DC converter, thesense filter network includes RF - CF network to sense the average inductor current information in form of DC voltage drop across the CF. In a preferred embodiment of the present current-mode DC-DC converter, the sense filter network is operatively connected to the summing amplifier, whereby the summing amplifier is configured to sum up the DC voltage drops in the different phases and amplify the same to produce an amplified voltage VA, which is fed to the flash ADC for comparing with reference voltages fordeciding the number of active phases to be turned on.In a preferred embodiment of the present current-mode DC-DC converter, the reference voltage levels are decided based on the load current range for which the active phases should be turned on in the load current range, whereby number and the size of the active phases are decided from overall power efficiency in entire load range.In a preferred embodiment of the present current-mode DC-DC converter, the output of the flash ADC is connected to enable signals EN of the power stage drivers in order to activate the phases including always turning on first phase and adding other phases by selectively enabling the power stage of a particular phase, when the load current exceeds a current limit corresponding to that particular phase, to share the load current. In a preferred embodiment of the present current-mode DC-DC converter, the flash ADC includes PVT independent hysteretic comparator topology with hysteretic band equivalent to a current change of ΔIH to avoid spurious triggering near the comparison point.In a preferred embodiment of the present current-mode DC-DC converter, the summing amplifier operates at a bandwidth having a cutoff frequency lower than switching frequency of the DC-DC buck converter to filter out ripple current information in the amplified voltage VAIn a preferred embodiment of the present current-mode DC-DC converter, power MOSFETs in the individual phases are unequal from each other to carry different amount of current, wherein the current gets distributed in the phases in proportion to the average equivalent resistance of the power MOSFET switches per phase.In a preferred embodiment, the present current-mode DC-DC converter comprises a slew rate enhancement block to selectively and forcefully turns ON / OFF the phases during load step-up / -down transient and when under- / over-shoot of the output voltage VOis larger than a predefined limit for improving slew-rate and transient response, whereby on reaching of steady state, the flash ADC finalizes the phase selection based on the average load current. In a preferred embodiment of the present current-mode DC-DC converter, the slew rate enhancement block includes a VOslew-sensing sub-block; a phase selector; and a forced set-reset block.In a preferred embodiment of the present current-mode DC-DC converter, the VO slew-sensing sub-block generates pulses for high load current step to control the phases, said VO slew-sensing sub-block includes atleast one RL-CLbased low-pass filter (LPF); atleast one hysteretic comparator; and atleast one pulse generating circuit.In a preferred embodiment of the present current-mode DC-DC converter, the RL-CL introduces an additional delay to inverting terminal of the hysteretic comparator, whereby the VO undershoots, which propagates directly to non-inverting terminal of the hysteretic comparator during large load step-up transient, creates a momentary voltage difference between the input terminals of the hysteretic comparator resulting a change of state of the hysteretic comparator from high to low, whereby the VO overshoots, which propagates directly to non-inverting terminal of the hysteretic comparator during large load step-down transient, creates a momentary voltage difference between the input terminals of the hysteretic comparator resulting a change of state of the hysteretic comparator from low to high, and whereby during the steady-state, the non-inverting terminal of the comparator is equal to the inverting terminal, enabling the hysteretic comparator to memorizes its previous state; wherein the pulse generating circuit senses edges (positive or negative) in state change and generate pulses for the phase selector block; and wherein, during small load step, the undershoot or the overshoot remains undetected as it fail to crosses hysteretic band of the hysteretic comparator thus avoiding spurious triggering.In a preferred embodiment of the present current-mode DC-DC converter, the phase selector select the phases from the flash ADC to slew enhanced APC block (and vice-versa), said phase selector includes atleast one MUX; and atleast one S-R latch. In a preferred embodiment of the present current-mode DC-DC converter, the pulse generating circuit generated pluses fed to inputs of the S-R latch and output Q of the S-R latch is connected to select lines of the MUX to select the phase, wherein, the pulse generated from the VO slew-sensing block at the S-input of the S-R latch selects the phase from flash ADC to slew enhanced APC block and the pulse at the R- input of the S-R latch select the phase back to the flash ADC. In a preferred embodiment of the present current-mode DC-DC converter, the pulse generating circuit includes atleast one R-C based delay generator and atleast one XOR gate. In a preferred embodiment of the present current-mode DC-DC converter, thepulse generating circuit is configure to sense positive or negative edge of signal HL and accordingly generates a pulse at the reset input R of the S-R latch to decide the number of active phases during a steady state, whereby the signal HL indicates the load condition of the converter (i.e. HL=0 for low load and HL=1 for high load), and said signal HL includes one of the outputs of the flash ADC (PH1,2,3,4 ) or a separate load-dependent logic signal to determine the load condition as low or high without any loss in generality.In a preferred embodiment of the present current-mode DC-DC converter, the forced set-reset block initializes the hysteretic comparator output based on the state of the converter (low load or high load state) including forcefully sets the output of the hysteretic comparator when the DC-DC buck converter enters into the low load current state and / or resets the hysteretic comparator during the transition to the high load current state. In a preferred embodiment of the current mode DC-DC converter, the forced set-reset block includes a positive edge detector sub-block; a negative edge detector sub-block; and a AND gate.In a preferred embodiment of the current mode DC-DC converter, wherein the positive edge detector or negative edge detector generates forcefully set or forcefully reset signal by detecting the positive or negative transition of the load indicator signal; wherein the AND of and is connected to select lines of MUX1 to control the transmission of undesirable glitch pulses from pulse generating circuit while forcibly setting or resetting the hysteretic comparator; wherein to suppress undesirable glitch pulses completely, the forcefully set and forcefully reset signal pulse width is greater than pulse width of OPTPLS.In a preferred embodiment of the current mode DC-DC converter, the positive edge detector and negative edge detector includes atleast one R-C based delay generator block; atleast one NOT gate; and atleast one NAND gate.Brief description of the accompanying drawings:Figure 1 shows the schematic illustration of a classical current-mode DC-DC buck converter.Figure 2 shows the schematic illustration of a classical current-mode DC-DC buck converter with the N-number of APCs.Figure 3 shows the schematic illustration of the current-mode DC-DC buck converter where the APC with the proposed phase current distribution block followed by the proposed slew enhancement blocks in accordance with the present invention for the N-number of phases.Figure 4 shows the schematic illustration of the current-mode DC-DC buck converter where the APC with the proposed phase current distribution block followed by the proposed slew enhancement block with filter-based current sensing mechanism in accordance with the present invention for the 4-number of phases as an example embodiment.Figure 5 shows the schematic illustration of the various components inside the APC with the proposed phase current distribution block in accordance with the present invention for the 4-number of phases as an example embodiment.Figure 6 shows representation of the power MOSFET switches with their equivalent resistances in each of the phases in accordance with the present invention.Figure 7 illustrates a difference between the current distribution in the conventional APC scheme and the proposed APC scheme while two phases are active as an example embodiment. Figure 7(a) illustrates the conventional scheme as VX1= VX2 and hence VSW1= VSW2. Figure 7(b) illustrates the proposed APC scheme to prevent the arbitrary current distribution through the equivalent resistance of the power MOSFETs as VX1= VX2 and hence VSW1= VSW2. Figure 8 shows the schematic illustration of the current distribution with the proposed APC scheme while four phases are active as an example embodiment. It fixes the current distribution through the equivalent resistance of the power MOSFETs as VX1= VX2= VX3 = VX4and hence VSW1= VSW2 =VSW3 = VSW4.Figure 9 shows the schematic illustration of the various components in the proposed slew enhancement block for improving dynamic response in accordance with the present invention.Figure 10, 11 shows the plot of the load current distribution in each of the active-phases for a load current of 5 A, 3 A respectively with the implementation of the proposed technique in accordance with the present invention as an example illustration.Figure 12(a) shows plot for the load transient response at a slew-rate of 1 A / ns for a load step-up of 100 mA to 5 A and step-down of 5 A to 100 mA without the activation of slew enhancement block in accordance with the present invention as an example illustration.Figure 12(b) shows plot for the load transient response at a slew-rate of 1 A / ns for a load step-up of 100 mA to 5 A and step-down of 5 A to 100 mA with the activation slew enhancement block in accordance with the present invention as an example illustration.Figure 13 shows the efficiency plot comparison of the theoretical V / S the simulation of the proposed DC-DC buck converter with the slew enhanced APC block in accordance with the present invention as an example illustration.DETAILED DESCRIPTION OF THE INVENTION WITH REFERENCE TO THE ACCOMPANYING FIGURES:The detailed schematic illustrations of various possible embodiments of the present invention are described in the accompanying Figure 3-9. However, it should be noted that the description of the embodiments are illustrative of the invention and are not intended to be limiting. Various modifications and applications may occur to those skilled in the art without departing from the true spirit and scope of the invention. As discussed earlier, higher power demands and stringent load current slew-rate in modern APs necessitate extending the current-mode DC-DC converter from the single-phase to APC configuration. Therefore, the extension of single-phase to APC requires an additional phase selection mechanism. As an example embodiment, four parallel set of the power stages with its individual inductor in parallel with the sense filter network are connected in between input voltage node VIN and output voltage node VO as shown in Figure 4. Average inductor current information in each active phase is sensed through the RF - CF network in the form of DC voltage drop across the CF. In the proposed system, a summing amplifier followed by a flash ADC has been used for deciding the number of active phases as shown in Figure 5. The DC voltage drop in the different phases are then summed up and amplified to produce the amplified voltage VA, which is then fed to the flash ADC for comparing with reference voltages VLi where i = 2, 3 and 4. For i=1, the corresponding phase is always ON. These reference voltage levels are decided based on the load current range for which the active phases should be turned on in the load current range. The number and the size of the active phases are mainly decided from the overall power efficiency in the entire load range. Table 1 shows the number of active phases and the corresponding outputs of the ADC.Table 1: the outputs of the flash ADC(PH2, PH3 and PH4) with the number of active phases in different load ranges for the 4-number of phases as an example embodiment.In the dynamic operation, when the load current exceeds the maximum limit of the particular phase, it adds another phase to share the load current. The addition of the phases are done by enabling the power stage of that particular phase. The output of the flash ADC PHi(where i = 2,3 and 4) are connected to the enable signals ENi (where i = 2,3 and 4) of the power stage drivers in order to activate the phases. The first phase is by default on. A PVT independent hysteretic comparator topology with hysteretic band equivalent to a current change of ΔHI is used in the flash ADC. The hysteretic band is useful to avoid spurious triggering near the comparison point.The voltage across the capacitorCFin each phase can be computed asEquation (1)Equation (2)Equation (3)Equation (4)Where, indicates the average and the ripple inductor current respectively in the ithphase. The resistance RDC is the series resistance of the inductor. The voltage at the output of the amplifier can be computed asEquation (5)Equation (5) contains both the average and the ripple inductor current information. However, the average inductor current information is only required for the APC. Therefore, the ripple current information is filtered out by reducing the bandwidth of the summing amplifier targeting a cutoff frequency lower than the switching frequency of the DC-DC buck converter. The first phase remains permanently ON; so, EN1 is always connected to the high logic. Table 1 shows the output of the flash ADC with the number of active phases in different load ranges to maximize the efficiency over a large load range. To achieve this, it is quite likely that the size of the power MOSFET in the individual phases appears as unequal from each other. Hence, it is obvious that, each of the active phases are designed to carry different amount of current. However, the current should be distributed among the active phases according to the theoretical value. Otherwise, if the current is getting distributed arbitrarily, it would affect the thermal stress management of the IC and its lifetime. Therefore, the present invention ensures a theoretically predicted distribution of current in each of the phases by forcefully shorting the node VXi as in Figure 5. Hence, the average value of the switch nodes VSWi are made equal. As a result, the current gets distributed in proportion to the average equivalent resistance of the power MOSFET switches per phase. The analytical justification and the verification using the simulation results of the present invention are as follows. To get the theoretical estimate of the current in each of the phase, a steady state mathematical modelling of the converter is shown in Figure 6. The power MOSFETs switches are replaced with the average equivalent resistances that are calculated using the equation as:Equation (6)Where, is for ithphase. The and are the total ON resistances of the PMOS and NMOS switches in the ithphase respectively which is represented by the equation. Equation (7)Equation (8)Here, , and are the process parameters. and are the widths and lengths of the PMOS and NMOS respectively in the ith phase.Figure 7 illustrates the idea considering the two active-phases, before and after the VX1 and VX2 are shorted. With no loss of generality, the idea can be extended for more no of active-phases. For example, let us assume that the load current is flowing such that the number of active-phases are two, as decided by the phase selection mechanism and the VX1 and VX2 are not shorted as shown in Figure 7(a). Therefore, in this condition, if the average value of VX1 and VX2 as well as the VSW1 and VSW2 are unequal and the amount of the current gets distributed arbitrarily through the power MOSFETs. However, with the direct shorting in the proposed embodiment in Figure 7(b), the average value of VX1 and VX2 become equal by mutually sharing the charges across CF of the two phases. Thus, the average value of VSW1 and VSW2 becomes equal. Therefore, the average value of IL1 and IL2 currents that flows through each of the active-phases is based on the effective average equivalent resistances of the power MOSFETs switches in the particular phase. So, the theoretical estimation of the average current in each of the phases is done using the equation:Equation (9)Equation (10)Where, is the total resistance in the phase at the steady state.Similarly, in case of the four number of active phases, the node voltage VX1 = VX2 = VX3 = VX4 are shorted so VSW1 = VSW2 = VSW3 = VSW4, as shown in Figure 8. Therefore, the corresponding individual phase current can be calculated as:Equation (11)Equation (12)Equation (13)Equation (14)With no loss in generality, the above equations can be extended to calculate the individual phase currents for the N-no of active phases, as follows: Equation (15)Equation (16)Equation (17)Therefore, in the proposed embodiment, the individual phase current are fixed and it is mainly decided by Reqi. In the cases of equal-sized APC and the multiphase DC-DC converter, the power MOSFET switch sizes and in all the segments are equal. Therefore, from equation (7) and (8), Equation (18)Equation (19)Substituting, and in equation (6),Equation (20)Equation (21) Substituting in equation of the individual phase current for the N-no of active phases.Equation (22)Equation (23)Here, each of the active phases carries equal amount of current. Hence, the proposed scheme equally works well and balances the individual phase currents in the equal-sized APC and the multiphase DCDC converters, where all the phases are having equal power MOSFET switch sizes.The other major issue associated with the current-mode DC-DC buck converter with APC is the degradation of the transient response in order to maintain high efficiency over a wide current load range. To mitigate this issue, the present invention also proposes the slew rate enhancement block as shown in Figure 9. Whenever the load step-up / -down transient occurs and the under- / over-shoot of the output voltage VOis larger than a predefined limit, the slew enhancement block forcefully turns ON / OFF rest of the three phases momentarily during the transient. The immediate turning ON / OFF of all the phases improves the slew-rate and hence the transient response. Once the steady state is reached, the flash ADC finalizes the phase selection based on the average load current. Therefore, the slew-rate enhancement block has the following functions. Firstly, it detects the load transient edgesat which thenumber of active phase is being turned ON / OFF in transients. Secondly, it decides number of active phases in the steady state. To achieve the above purposes, the slew enhancement block contains various sub-blocks described in detail in the following subsection.VOSlew-Sensing:The VOslew-sensing sub-block generates the pulses for the high load current step to control the phases. It consists of RL-CLbased low-pass filter (LPF) and a hysteretic comparator, as shown in Figure 9. During the large load step-up, VOundershoots and propagates directly to the non-inverting terminal. At the same time, RL-CLintroduces an additional delay to the inverting terminal, thus creating a momentary voltage difference between the input terminals of the hysteretic comparator, which results in the change of its state from high to low. Similarly, during the load current step-down transient, the state of the hysteretic comparator changes from low to high. The following circuit senses these edges (positive or negative) to generate the pulses for the phase selector block. In the steady-state, the non-inverting terminal of the comparator is equal to the inverting terminal, so the hysteretic comparator memorizes its previous state. When the small load step occurs, the undershoot or the overshoot remains undetected as it never crosses the hysteretic band. In this condition, the transient specification are any way within the limit. Here, hysteretic comparator is distinguishing between the small signal (within hysteretic band) and large signal (out of the hysteretic band). In addition to that, hysteretic band also avoids spurious triggering.Phase Selector:The phase selector, comprising of MUX and S-R latch, is used to select the phases from the flash ADC to slew enhanced APC block (and vice-versa). According to the pulses generated at the input of the S-R latch, the output Q connected to the select lines of the MUX selects the phase. The pulse generated from the VO slew-sensing block at the S-input selects the phase from flash ADC to slew enhanced APC block. Whereas, the pulse at the R- input is required to select the phase back to the flash ADC. In order to generate this pulse, positive negative edge detector is fed with the signal HL which indicates the load condition of the converter (i.e. HL=0 for low load and HL=1 for high load). The decision to turn ON or OFF all the phasessimultaneously is based on the previous state of the multiphase DC-DC buck converter. The number of active phases decides the state of the converter. During the low load current state, with the detectable load current step-up, it is required to turn on all the phases simultaneously in order to have high charging current rate for the output capacitor. Similarly, in the high load state and with the detectable load current step-down it is to be turned off simultaneously. Therefore, signal is fed to the high input of the MUX, which is high( =1) during the low load current state and low ( =0) in the high load current state.Forced Set-Reset Block:VOslew-sensing block has a limitation in terms of the detecting the overshoot andthe undershoot. As per the configuration, the hysteretic comparator indicates the overshoot by changing its state from low to high and the undershoot by changing from high to low. Therefore, the initial state of the output of the hysteretic comparator has to be low to detect the overshoot, and it has to be high for detecting the undershoot. So, the forced set-reset sub-block is proposed to initialize the hysteretic comparator output based on the state of the converter (low load or high load state), as shown in Figure 9. The sub-block forcefully sets the output of the hysteretic comparator when the DC-DC buck converter enters into the low load current state (i.e. when HL = 0). It resets the hysteretic comparator during the transition to the high load current state (i.e. when HL = 1). The load indicator signal HL has the rising edge as the state of the converter changes from the low load to the high load current, whereas it has the falling edge for the change from the high load to the low load current. Therefore, the rising edge is sensed by the positive edge detector block to generate the negative pulse of , which forcefully reset the hysteretic comparator output, making it ready for possible detection of the next overshoot. Similarly, the falling edge is sensed by a negative edge detector block to generate a negative pulse of , which forcefully sets the hysteretic comparator output making it ready for possible detection of the next undershoot. A positive or negative edge detector block may generate undesirable pulses while forcibly setting or resetting the hysteretic comparator. To control the transmission of pulses to the S input of SR latch, MUX1 is placed between positive-negative edge detector and S input of SR latch. The MUX1 only allows the pulse generated due to the VOslew-sensing block. Therefore, its select line connected to the X signal, which is the AND of and . The high input of the MUX1 input is fed with the output of the positive-negative edge detector block, and low input fed with low signal. To block the undesirable pulses completely, the signal and pulse width has to be greater than the pulse width of OPTPLS.RESULTS:The different embodiments of the proposed converter have been designed and tested in 0.18 μm CMOS technology. The reference clock frequency is set to 25 MHz. The input voltage range is targeted as 2.5 V to 3.3 V with the output voltage fixed at 1.2 V. The value of the inductor in each phase is Li = 100 nH and the direct current resistor is RDCi = 20 mΩ (where i is the ithphase). The value of the output capacitor CO is 4.4 μF and RESR = 10 mΩ. Table 2 contains the lists of the theoretically calculated phase current values along with its actual values in each of the active phases for the load current 5 A and 3 A. With the proposed technique, the theoretically predicted current distribution almost matches with the actual current distribution. Table 2comparison of the theoretically calculated current distribution with the actual current distribution at the 5 and 3 A of the load currentin accordance with the present invention as an example illustration.Figure 10 and 11 shows the plot of the load current distribution in each of the four phases corresponding to the load 5 A and 3 A respectively. Here, IO corresponds to the load current and ILi is the phase current in the ith phase. With the proposed technique in the present invention, fixing of the actual phase currents values close to the theoretical value is validated from the plot.Figure 12(a) show the plots of the load transient response at a slew-rate of 1 A / ns for a load (IO) step-up of 100 mA to 5 A and step-down of 5 A to 100 mA without activation of the slew enhancement block. The undershoot and the overshoot of 200 mV and 81 mV, respectively, which is too large. The step-up and step-down recovery time is 420 ns and 260 ns.Figure 12(b) show the plots of the load transient response at a slew-rate of 1 A / ns for a load (IO) step-up of 100 mA to 5 A and step-down of 5 A to 100 mA with activation of the slew enhancement block. The proposed invention achieves the undershoot and the overshoot of 60 mV and 52 mV, respectively, which is close to the theoretical limit (i.e., IO.RESR = 50 mV). The step-up and step-down recovery time of the proposed converter is 298 ns and 192 ns. Figure 13 shows the comparative analysis of the theoretical and simulated efficiency plot for the proposed DC-DC converter over the wide load current range. As can be seen theoretically, with the proposed APC technique, efficiency stays around 85 % across the total power range. This is further verified and validated with the simulation results.Thus, it is apparent from the above-discussed result, that the issue of the arbitrary distribution of the load current is resolved in the proposed method. It also fixes the phase current in the different active phases close to the theoretical values. The present invention also improves the transient response and achieves almost the theoretical limit of the undershoot and the overshoot by forcefully turning ON-OFF all the three phases simultaneously during the extreme load transient.

Claims

1. A system for improving current-mode DC-DC converter operation involving high load current slew-rate and fast dynamic load transient response comprising   A phase current distribution unit operable with the current-mode DC-DC converter for extending the current-mode DC-DC converter from single-phase to APC configuration, said phase current distribution unit is further operable with equal sized APC and multiphase DC-DC converter. said phase current distribution unit is operatively connected with plurality of power stages, whereby the phase current distribution unit includes a summing amplifier followed by a flash ADC for selectively turning on active phases in the power stages in accordance with load current range for sharing the load current among the turned on active phases and maintain overall power efficiency in entire load range.

2. The current-mode DC-DC converter as claimed in claim 1, wherein each of the power stages comprises atleast one power transistor capable of delivering a large amount of current to the load, whereby gate terminal of said power transistor is driven by a driver block.

3. The current-mode DC-DC converter as claimed in claim 1 or 2, wherein the power stages including their individual inductors are connected in parallel between input and output voltage nodes with sense filter network, whereby said sense filter network sense average inductor current information in each active phase.

4. The current-mode DC-DC converter as claimed in claims 1 to 3, wherein the sense filter network includes RF - CF network to sense the average inductor current information in form of DC voltage drop across the CF.

5. The current-mode DC-DC converter as claimed in claims 1 to4, wherein the sense filter network is operatively connected to the summing amplifier, whereby the summing amplifier is configured to sum up the DC voltage drops in the different phases and amplify the sameto produce an amplified voltage VA, which is fed to the flash ADC for comparing with reference voltages fordeciding the number of active phases to be turned on.

6. The current-mode DC-DC converter as claimed in claims 1 to 5, wherein the reference voltage levels are decided based on the load current range for which theactive phases should be turned on in the load current range, whereby number and the size of the active phases are decided from overall power efficiency in entire load range.

7. The current-mode DC-DC converter as claimed in claims 1 to 6, wherein the output of the flash ADC is connected to enable signals EN of the power stage drivers in order to activate the phases including always turning on first phase and adding other phases by selectively enabling the power stage of a particular phase, when the load current exceeds a current limit corresponding to that particular phase,to share the load current.

8. The current-mode DC-DC converter as claimed in claims 1 to 7, wherein the flash ADC includes PVT independent hysteretic comparator topology with hysteretic band equivalent to a current change of ΔIH to avoid spurious triggering near the comparison point.

9. The current-mode DC-DC converter as claimed in claims 1 to 7, wherein the summing amplifier operates at a bandwidth having a cutoff frequency lower than switching frequency of the DC-DC buck converter to filter out ripple current information in the amplified voltage VA.

10. The current-mode DC-DC converter as claimed in claims 1 to 9, wherein power MOSFETs in the individual phases are unequal from each other to carry different amount of current, wherein the current gets distributed in the phasesin proportion to the average equivalent resistance of the power MOSFET switches per phase.

11. The current mode DC-DC converter as claimed in claims 1 to 10, comprises a slew rate enhancement block to selectively and forcefully turns ON / OFF the phases during load step-up / -down transient and when under- / over-shoot of the output voltage VOis larger than a predefined limit for improving slew-rate and transient response, whereby on reaching of steady state, the flash ADC finalizes the phase selection based on the average load current.

12. The current mode DC-DC converter as claimed in claim 11, wherein the slew rate enhancement block includes a VOslew-sensing sub-block; a phase selector; and a forced set-reset block.

13. The current mode DC-DC converter as claimed in claim 12, wherein the VO slew-sensing sub-block generates pulses for high load current step to control the phases, said VO slew-sensing sub-block includes atleast one RL-CLbased low-pass filter (LPF); atleast one hysteretic comparator; and atleast one pulse generating circuit.

14. The current mode DC-DC converter as claimed in claim 13, wherein the RL-CL introduces an additional delay to inverting terminal of the hysteretic comparator, whereby the VO undershoots, which propagates directly to non-inverting terminal of the hysteretic comparator during large load step-up transient, creates a momentary voltage difference between the input terminals of the hysteretic comparator resulting a change of state of the hysteretic comparator from high to low, whereby the VO overshoots, which propagates directly to non-inverting terminal of the hysteretic comparator during large load step-down transient, creates a momentary voltage difference between the input terminals of the hysteretic comparator resulting a change of state of the hysteretic comparator from low to high, and whereby during the steady-state, the non-inverting terminal of the comparator is equal to the inverting terminal, enabling the hysteretic comparator to memorizes its previous state; wherein, the pulse generating circuit senses edges (positive or negative) in state change and generate pulses for the phase selector block; and wherein, during small load step, the undershoot or the overshoot remains undetected as it fail to crosses hysteretic band of the hysteretic comparator thus avoiding spurious triggering.

15. The current mode DC-DC converter as claimed in claim 12, wherein the phase selector select the phases from the flash ADC to slew enhanced APC block (and vice-versa), said phase selector includes atleast one MUX; and atleast one S-R latch.

16. The current mode DC-DC converter as claimed in claim 15, wherein the pulse generating circuit generated pluses fed to inputs of the S-R latch and output Q of the S-R latch is connected to select lines of the MUX to selects the phase, wherein, the pulse generated from the VOslew-sensing block at the S-input of the S-R latch selects the phase from flash ADC to slew enhanced APC block and the pulse at the R- input of the S-R latch select the phase back to the flash ADC.

17. The current mode DC-DC converter as claimed in claim 15, wherein the pulse generating circuit includes atleast one R-C based delay generator and atleast one XOR gate.

18. The current mode DC-DC converter as claimed in claim 15, wherein the pulse generating circuit is configure to sense positive or negative edge of signal HL and accordingly generates a pulse at the reset input R of the S-R latch to decide the number of active phases during a steady state, whereby the signal HL indicates the load condition of the converter (i.e. HL=0for low load and HL=1 for high load), and said signal HL includes one of the outputs of the flash ADC (PH1,2,3,4 ) or a separate load-dependent logic signal to determine the loadcondition as low or high without any loss in generality.

18. The current mode DC-DC converter as claimed in claim 12, wherein the forced set-reset block initializes the hysteretic comparator output based on the state of the converter (low load or high load state) including forcefully sets the output of the hysteretic comparator when the DC-DC buck converter enters into the low load current state and / or resets the hysteretic comparator during the transition to the high load current state.

19. The current mode DC-DC converter as claimed in claim 18, wherein the forced set-reset block includes a positive edge detector; a negative edge detector; and a AND gate.

20. The current mode DC-DC converter as claimed in claim 18, wherein the wherein the positive edge detector or negative edge detector generates forcefully set or forcefully reset signal by detecting the positive or negative transition of the load indicator signal; wherein the AND of and is connected to select lines of MUX1 to control the transmission of undesirable glitch pulses from pulse generating circuit while forcibly setting or resetting the hysteretic comparator; wherein to suppress undesirable glitch pulses completely, the forcefully set and forcefully reset signal pulse width is greater than pulse width of OPTPLS.

21. The current mode DC-DC converter as claimed in claim 19, wherein said positive edge detector and negative edge detector includes atleast one R-C based delay generator block; atleast one NOT gate; and atleast one NAND gate.