Phase shedding based on phase efficiency determined by pulse-width modulator switching of voltage regulator field-effect transistors.
By identifying and disabling the least efficient phase in polyphase power supplies based on pulse width comparisons, the mechanism addresses efficiency loss from unequal phase efficiencies, improving overall performance and extending phase lifespan.
Patent Information
- Application Number
- JP2023518723
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Priority Date
- 2020-09-23
- Filing Date
- 2021-09-17
- Publication Date
- 2024-11-21
- Estimated Expiration
- 2041-09-17
AI Technical Summary
In polyphase power supplies, phase shedding based on numerical order reduces overall efficiency due to unequal phase efficiencies caused by manufacturing variations and degradation over time.
A mechanism that identifies the least efficient phase by comparing measured pulse widths to a nominal base pulse width, disabling it first during power demand reduction to balance phase wear and improve efficiency.
This approach maintains higher overall efficiency by prolonging the lifespan of more efficient phases and reducing energy loss, thereby enhancing the performance of polyphase power supplies.
Smart Images

Figure 0007588219000002 
Figure 0007588219000003 
Figure 0007588219000004
Abstract
Description
[Background technology]
[0001] A voltage regulator automatically maintains a constant output voltage level to provide a stable supply voltage to sensitive electronic devices given a potentially variable input voltage. Voltage regulators are often used in devices such as computer power supplies to stabilize the direct current (DC) voltage used by the processor, memory and other components. Voltage regulators are used to adjust DC and / or alternating current (AC) voltages using electromechanical mechanisms or electronic components.
[0002] High power DC-DC voltage regulators commonly used in computing systems can contain multiple phases, with each phase contributing to a stable output voltage. Phase shedding is the process of turning off a phase to improve power supply efficiency when less power is being consumed.
[0003] The present disclosure is illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings. [Brief description of the drawings]
[0004] [Figure 1] FIG. 1 illustrates a computing system, according to one embodiment. [Diagram 2] FIG. 2 illustrates a diagram showing phases in a multi-phase power supply, according to one embodiment. [Diagram 3] FIG. 2 illustrates voltage waveforms for each phase in a multi-phase power supply, according to one embodiment. [Figure 4] FIG. 2 illustrates components of a multi-phase power supply, according to one embodiment. [Diagram 5] FIG. 2 illustrates a process for performing phase shedding, according to one embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0005] In the following description, numerous specific details are described, such as examples of specific systems, components, methods, etc., to enhance understanding of the embodiments. However, it will be apparent to those skilled in the art that at least some of the embodiments may be practiced without these specific details. In other instances, well-known components or methods are not described in detail or are presented in simplified block diagram form to avoid unnecessarily obscuring the embodiments. Thus, the specific details described are merely exemplary. Particular embodiments may differ from these example details and still be considered to be within the scope of the embodiments.
[0006] A multi-phase power supply that supplies power to a load includes multiple phases that each contribute to the output voltage generated by the power supply. When the power supply is supplying a variable load, such as a computing system, the overall efficiency is increased by phase shedding, in which one or more phases are disabled when the power required by the load decreases. In an ideal multi-phase power supply, the power phases are equally efficient, so that they share the load equally and each provides an equal amount of power to the load. However, in practice, one or more of the phases may have higher losses due to variations in their component devices, even within acceptable manufacturing tolerances. For example, device geometry, composition, and other factors may affect the capabilities of the power phases. Thus, when phase shedding is performed, more efficient phases may be turned off (e.g., when phases are selected for shutoff in numerical order), leading to a decrease in the overall efficiency of the power supply.
[0007] An embodiment of a multi-phase power supply that performs phase shedding in response to a decrease in power demand identifies the least efficient of its power phases and disables these least efficient phases first when phase shedding is performed. For each phase, the power supply identifies the least efficient phase by comparing the measured pulse width of the phase to a nominal base pulse width calculated based on the input voltage, output voltage, and switching frequency of the power supply. A phase that utilizes a longer on-time in each cycle to deliver the same amount of power as one or more other phases is determined to be less efficient than the other phases. Thus, when the power demanded by the load decreases, the phase shedding mechanism disables the identified least efficient phase first. The least efficient phase is the phase that experiences the greatest energy loss (e.g., in the form of heat) relative to the amount of power delivered by the phase.
[0008] As power phases degrade over time due to operational stresses, they often become less efficient. The phase shedding mechanism reduces the amount of time that the least efficient phase is operational and subject to operational stresses. As a result, the least efficient phase tends to suffer less degradation before another phase becomes less efficient. Thus, the phase shedding mechanism automatically averages out phase wear among multiple phases.
[0009] 1 illustrates an embodiment of a computing system 100 in which the mechanism for phase shedding based on the phase efficiency detection mechanism described above is implemented. Generally, the computing system 100 may be embodied as any of a number of different types of devices, including, but not limited to, a laptop or desktop computer, a mobile terminal, a server, a network switch or router, and the like. The computing system 100 includes a number of hardware resources including components 102-108 that communicate with each other via a bus 101. In the computing system 100, each of the components 102-108 may communicate with any of the other components 102-108 directly via the bus 101 or through one or more of the other components 102-108. The components 101-108 in the computing system 100 are contained within a single physical enclosure, such as the chassis of a laptop or desktop computer, or the casing of a mobile phone. In alternative embodiments, some of the components of computing system 100 are embodied as external peripheral devices, such that the entire computing system 100 does not reside within a single physical enclosure.
[0010] Computing system 100 also includes user interface devices for receiving information from a user or providing information to a user. Specifically, computing system 100 includes input devices 102, such as a keyboard, mouse, touch screen, or other device for receiving information from a user. Computing system 100 displays information to the user via a display 105, such as a monitor, light-emitting diode (LED) display, liquid crystal display, or other output device.
[0011] Computing system 100 further includes a network adapter 107 for transmitting and receiving data over a wired or wireless network. Computing system 100 also includes one or more peripheral devices 108. Peripheral devices 108 may include mass storage devices, location sensing devices, sensors, input devices, or other types of devices used by computing system 100.
[0012] Computing system 100 includes one or more processing units 104, which in the case of multiple processing units 104 can operate in parallel. Processing units 104 receive and execute instructions 109 stored in memory system 106. In one embodiment, processing units 104 include multiple processing cores residing on a common integrated circuit substrate. Memory system 106 includes storage devices used by computing system 100, such as random-access memory (RAM) modules, read-only memory (ROM) modules, hard disks, and other non-transitory computer-readable media.
[0013] Some embodiments of computing system 100 may include fewer or more components than the embodiment shown in Figure 1. For example, certain embodiments are implemented without display 105 or input devices 102. Other embodiments have more than one of certain components, for example, an embodiment of computing system 100 may have multiple buses 101, network adapters 107, memory systems 106, etc.
[0014] Power is provided to components 101-108 in computing system 100 by power supply 110. In one embodiment, power supply 110 is a multi-phase power supply that disables its least efficient power phase in response to a decrease in power demand by its load (e.g., components 101-108).
[0015] 2 illustrates one embodiment of a multi-phase power supply 110 that includes a voltage regulator 201 and N power phases 1 through N. The voltage regulator 201 receives an input voltage V IN A lower stable output voltage V OUT , which controls the switching of field effect transistors (FETs) in each of phases 1-N. Each of a set of pulse width sensors 210(1)-210(N) in power supply 110 is used to generate pulse width measurements for each of phases 1-N. Each of pulse width sensors 210(1)-210(N) measures the on-time of an upper FET (e.g., one of 211(1)-211(N)) in its respective phase. For example, pulse width sensor 210(1) measures the on-time of upper FET 211(1) in phase 1 based on the voltage measured at the gate of upper FET 211(1). Phases 1-N in multi-phase power supply 110 are implemented using FETs. However, alternative embodiments utilize other types of switching devices, such as bipolar junction transistors (BJTs), insulated-gate bipolar transistors (IGBTs), etc., in place of the illustrated FETs.
[0016] FIG. 3 illustrates a graph of voltage V over time t according to one embodiment. G1 ~V GN 1 is a timing diagram showing V G1 ~V GN is measured at the gate of the upper FET 211(1) to 211(N) for each of the respective phases 1 to N. ON1 ~T ONN represents the length of time that each of the FETs 211(1) to 211(N) is in a conductive state, and T OFF1 ~T OFFN represents the length of time that each of the FETs 211(1) to 211(N) is in a non-conducting state. As shown in FIG. 3, the FET 211(2) in phase 2 has a pulse width of 100 s (the widest pulse T ON2 is on for the longest duration, as shown by TON1 and T ONN continues.
[0017] 4 illustrates one embodiment of a multi-phase power supply 110 that supplies power to a load 430, such as a component of computing system 100. The components of multi-phase power supply 110 are implemented with hardware circuits or a combination of hardware circuits and software and / or firmware components. In various embodiments, computations in power supply 110 are performed using digital logic circuits, or alternatively using analog circuits and signals.
[0018] The multi-phase power supply 110 includes a voltage regulator 201 that identifies the least efficient power phase among the phases 1-N by comparing the measured pulse widths of the phases 1-N. In one embodiment, the power phase with the widest measured pulse width (i.e., the longest on-time per cycle) is identified as the least efficient phase. In one embodiment, the phases are ranked by their efficiency according to their measured pulse widths, with wider pulse widths corresponding to lower efficiency. The voltage regulator 201 detects a decrease in power demanded by the load 430 and responds by performing phase shedding to disable one or more of the power phases 1-N. The voltage regulator 201 selects the least efficient phase to disable (i.e., disables the phase that is extreme in terms of lack of efficiency), thus reducing the amount of power delivered to the load 430 by the identified least efficient phase. As a result, the overall efficiency of the power supply is increased.
[0019] Voltage regulator 201 is V IN and V OUT The circuit includes a calculation unit 411 for calculating the base duty cycle and active pulse width (representing the on-time of the upper FET in an ideal phase) based on the measurements of the duty cycle D and the switching frequency f of phases 1 to N. ON is calculated according to Equation 1 below.
[0020]
number
[0021] As shown in Equation 1, the duty cycle depends on the input voltage V supplied to power phases 1 to N. IN and an output voltage V generated at the output of the power phases 1 to N and applied to the load 430. OUT If the power supply 110 is used, the input voltage V IN can change over time (e.g., V IN is provided by an unregulated power supply. Therefore, V IN The sensor 401 senses the input voltage V IN Also, measure the output voltage V OUT V varies over time (e.g., due to changes in the power demanded by the load 430 or due to changes within the operating tolerances). OUT The sensor 402 measures V for each cycle during operation of the power supply 110. OUT The switching frequency f is determined at design time and is provided by a counter, timer or other timing device 403.
[0022] V IN Sensor 401, V OUT The sensor 402 and the timing device 403 are V IN , V OUT The values of f and f are sent to a calculation unit 411 within the voltage regulator 201 via a set of measurement inputs 412. The calculation unit 411 calculates the base (i.e., nominal) duty cycle D and base active pulse width T for the current cycle according to Equation 1. ON That is, calculate the nominal upper FET on time.
[0023] In normal operation, individual phases may experience losses occurring outside the switching device area (e.g., due to factors such as DC resistance (DCR) in the printed circuit board) that cause the actual duty cycle and pulse width of each phase to differ from their calculated base duty cycle and pulse width. Therefore, the actual pulse width is measured for each of the phases so that it can be compared to the calculated base pulse width. Voltage regulator 201 receives pulse width measurements T from pulse width sensors 210(1)-210(N) of each phase 1-N via measurement input 412. ON1 ~T ONN The pulse width measurements include measured pulse widths indicating the on-time of the upper FET for each of phases 1 through N.
[0024] In the voltage regulator 201, the comparison logic 413 compares the pulse width measurement T ON1 ~T ONN and the calculated base pulse width T ON The comparison logic 413 performs a set of comparisons by determining the difference between phase 1-N. The comparison logic 413 identifies the phase associated with the largest difference as the least efficient phase of phases 1-N. In one embodiment, the comparison logic 413 determines the relative efficiencies of the phases (i.e., ranks the phases by their determined efficiencies) for use in prediction or other calculations.
[0025] Pulse width sensors 210(1)-210(N) do not generate valid pulse width measurements for phases that are not in operation, such as phases that were previously disabled due to phase shedding or out-of-tolerance shutdown. Thus, comparison logic 413 excludes phases that are not currently in operation from the calculation. As a result, the least efficient phase identified by comparison logic 413 is the least efficient of the currently operating phases.
[0026] The calculated difference for each phase is also used to determine if that phase is operating outside of an acceptable duty cycle tolerance. In an alternative embodiment, the comparison logic 413 compares the pulse width T ON1 ~TONN and identifies the phase with the widest measured pulse width as the least efficient phase. In one embodiment, comparison logic 413 performs the comparison using digital logic. Alternatively, logic 413 performs the comparison using analog signals and components.
[0027] In one embodiment, the pulse width T ON1 ~T ONN is measured every cycle, and the resulting pulse width of each successive pulse occurring in each phase is measured. Therefore, the base pulse width T ON is calculated and the comparison performed by the comparison logic 413 is also performed on a cycle-by-cycle basis, i.e., a comparison is performed by the comparison logic 413 to determine the least efficient phase for each successive phase cycle of the multi-phase power supply.
[0028] The memory 414 records the identity of the least efficient phase as determined by the comparison logic 413. In some embodiments, the memory 414 records the most recently measured pulse width T for each of phases 1-N. ON1 ~T ONN, as well as historical data such as the most recently calculated base duty cycle, base pulse width, and / or phase efficiency ranking. The information stored in memory 414 is used for predictive failure analysis or other advanced calculations or control. In various embodiments, memory 414 may be a register, a counter, a memory cell, or the like, and may be implemented using any of a number of volatile or non-volatile memory technologies, such as, for example, dynamic random access memory (DRAM), static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), flash memory, or another memory technology. Although FIG. 4 illustrates memory 414 as being included in voltage regulator 201, in alternative embodiments, memory 414 is external to voltage regulator 201.
[0029] The phase shedding logic unit 416 reduces the number of operating phases 1-N in response to a decrease in power required by the load 430. The phase shedding logic 416 also increases the number of operating phases if the power demand increases. When the phase shedding logic 416 receives a power demand indicator 431 (e.g., determined from load current and / or voltage measurements) indicating a decrease in power demand from the load 430, the phase shedding logic 416 reduces the amount of power provided to the load 430 by the least efficient currently operating phase by disabling the least efficient phase. The phase shedding logic 416 shuts down the least efficient phase by reading a recently recorded identifier 415 from the memory 414 that identifies this phase.
[0030] In various embodiments, disabling of the least efficient phase is accomplished by deasserting one or more switching control signals of the phase, such as pulse width modulated (PWM) switching signals or gate drive signals (e.g., periodic signals that control switching of upper and lower FETs in the phase), depending on the design of the phase. In one embodiment, one or more switching control signals (e.g., PWM signals or gate drive signals) of a phase are turned off (e.g., deasserted) and on when the enable signal of that phase is deasserted and asserted, respectively. As shown in FIG. 4, phase shedding logic 416 disables the phase corresponding to identifier 415 by deasserting the enable signal (i.e., one of signals 420(1)-420(N)) corresponding to the least efficient phase. In an alternative embodiment, phase shedding logic 416 stops sending one or more switching control signals from voltage regulator 201 to the least efficient phase. If the least efficient phase 415 cannot be determined, the phase shedding logic 416 reverts to alternative methods for selecting a phase to disable, such as selecting a phase in sequential order, or selecting the phase that is drawing the greatest amount of current.
[0031] When the phase shedding logic 416 receives an indication 431 that the power demand of the load 430 has increased, the phase shedding logic 416 enables one of the previously disabled phases. In one embodiment, the phase shedding logic 416 enables the most efficient of the disabled phases (i.e., the extreme of the phases in terms of efficiency) as identified in the memory 414. Alternatively, the phase shedding logic 416 enables the most recently disabled phase.
[0032] In one embodiment, the phase shedding logic 416 further includes shutdown logic to disable any phase that exceeds a duty cycle tolerance threshold. For each of phases 1-N, the measured pulse width or duty cycle for the phase is calculated based on the base duty cycle D or pulse width T ONIf it differs from the threshold amount or percentage, the comparison logic 413 identifies the out-of-tolerance phase 417 to the phase shedding and shutdown logic 416 and disables the phase.
[0033] 5 illustrates a phase shedding process 500 that determines the least efficient phase in multi-phase power supply 110 and disables the least efficient phase first when performing phase shedding in response to a decrease in power demand, according to one embodiment. When power demand increases, the most efficient phase that was previously disabled is re-enabled first. Process 500 is performed by circuitry within multi-phase power supply 110, according to one embodiment.
[0034] In block 501 of the phase shedding process 500, the input voltage sensor 401 and the output voltage sensor 402 each detect the input voltage V IN and the output voltage V OUT Measure the V IN and V OUT The measured value of is received by the calculation unit 411 together with the switching frequency f provided by the timing device 403, as provided in block 503, to obtain the base pulse width T ON is used by the calculation unit 411 to calculate
[0035] In block 505, pulse width sensors 210(1)-210(N) generate a set of N pulse width measurements by measuring the pulse width for each of phases 1-N in polyphase power supply 110. Comparison logic 413 compares the pulse width measurements T ON1 ~T ONN and the base pulse width T ONand the base pulse width by a maximum duration indicates the least efficient phase. In one embodiment, a phase having a wider pulse width measurement (i.e., a longer pulse) is identified as less efficient than a phase having a narrower pulse width measurement (i.e., a shorter pulse). The comparison logic 413 also identifies any phases operating at a duty cycle that is out of tolerance (e.g., exceeds the base duty cycle by more than a threshold percentage) based on the previously calculated differences. In block 511, the identity of the least efficient phase is recorded in memory 414. In some embodiments, memory 414 stores the calculated duty cycle, pulse width, and / or other values of the most recent cycle to perform predictive failure analysis or other calculations.
[0036] In block 513, the phase shedding logic 416 determines whether two or more of the phases are enabled and the power demand indicator 431 indicates a reduction in power required by the load 430. For example, a reduction may be indicated if the total power required by the load 430 falls below a preset threshold or decreases to a lower range of a set of predefined ranges. If both conditions are true (i.e., two or more phases are enabled and the power demand is sufficiently reduced), the phase shedding logic 416 reduces the amount of power provided to the load 430 by the least efficient power phase according to blocks 515 and 517. In block 515, the phase shedding logic 416 reads from the memory 414 a previously recorded identifier of the least efficient phase. In block 517, the phase shedding logic 416 disables the phase corresponding to the identifier (e.g., by deasserting the enable signal for this phase). In one embodiment, the phase shedding logic 416 disables the least efficient phase by terminating the delivery of one or more gate drive signals to the phase. From block 517, the process 500 returns to block 501 to identify the least efficient power phase for the next cycle.
[0037] At block 513, if only one phase is enabled, then no phase shedding is performed so that the last remaining phase is not disabled. Also, if the power demand indicator 431 does not indicate a sufficient reduction in the power required by the load 430, then no phase shedding is performed and the process 500 continues to block 519.
[0038] If, at block 519, the phase shedding logic 416 receives an indication 417 from the comparison logic 413 that one of the phases has an out-of-tolerance duty cycle (as determined at block 509), then the phase identified by the comparison logic 413 as being out-of-tolerance is disabled at block 517. In one embodiment, the out-of-tolerance phase is marked as being disabled due to an out-of-tolerance condition so that it is not re-enabled by the phase shedding mechanism or other mechanisms. From block 517, the process 500 returns to block 501 to identify the least efficient power phase for the next cycle.
[0039] If, at block 519, no indication 417 that a phase has an out-of-tolerance duty cycle is received, the process 500 continues to block 521. At block 521, the phase shedding logic determines whether at least one of the phases is currently disabled by a phase shedding mechanism and whether the power demand indicator 431 indicates that the power demanded by the load 430 has increased. For example, an increase may be indicated if the total power demanded by the load 430 exceeds a pre-set threshold or is determined to have increased within a particular range. If both conditions are true (i.e., at least one phase was previously disabled by phase shedding and the power demand has increased sufficiently), the phase shedding logic 416 increases the amount of power delivered to the load 430 by adding or enabling the most efficient of the one or more disabled power phases according to blocks 523 and 525. At block 523, the phase shedding logic 416 accesses the memory 414 to identify the most efficient phase currently disabled by a previous phase shedding. In one embodiment, the most efficient disabled phase is the phase most recently disabled by the phase shedding logic 416. Once identified, the most efficient disabled phase is re-enabled in block 525. Thus, the phases are re-enabled in order of their efficiency, with the more efficient phases being re-enabled first. In one embodiment, the phase shedding logic 416 re-enables the most efficient of the disabled phases (previously identified as the least efficient of the enabled phases) by restarting one or more switching control signals for the phase. From block 525, the process 500 returns to block 501 to identify the least efficient power phase for the next cycle.
[0040] If, at block 521, all of the phases have been enabled or power demand has not increased, process 500 returns to block 501. Thus, process 500 repeats as shown in blocks 501-525 to identify the least efficient power phase for each cycle of the multi-phase power source 110. Thus, the operations represented by blocks 501-513 to identify the least efficient phase are performed for each successive phase cycle of the multi-phase power source.
[0041] In subsequent iterations of the process 500 after the least efficient phase has been disabled by the phase shedding logic 416, the comparison logic 413 identifies the next least efficient phase in block 509 that is more efficient than the phase already disabled and less efficient than any of the phases still operating. In response to a further decrease in power demand, the phase shedding logic 416 disables the next least efficient phase. In this manner, the phases are disabled in an order determined by their efficiency, with the least efficient phase being disabled first. As a result, the more efficient phases are used for a longer period of time than the less efficient phases, increasing the overall efficiency of the multi-phase power supply.
[0042] A method of operating a multi-phase power supply includes generating a plurality of pulse width measurements by measuring a pulse width for each phase of a plurality of phases in the multi-phase power supply, identifying a least efficient phase of the plurality of phases based on performing a comparison of the plurality of pulse width measurements, and reducing an amount of power delivered to a load by the identified least efficient phase.
[0043] In the method, reducing the amount of power delivered to the load includes disabling the identified least efficient phase in response to detecting a first decrease in power demand of the load.
[0044] The method further includes identifying a next least efficient phase of the plurality of phases based on a comparison of the plurality of pulse width measurements, and disabling the next least efficient phase in response to detecting a second decrease in power demand of the load.
[0045] The method further includes measuring an input voltage of the multi-phase power supply, measuring an output voltage of the multi-phase power supply, the output voltage of which is applied to a load, and calculating a base pulse width based on a switching frequency of the multi-phase power supply, the measured input voltage, and the measured output voltage. Performing a comparison of the multiple pulse width measurements includes determining a difference between each of the pulse width measurements and the base pulse width.
[0046] In the method, reducing the amount of power delivered to the load by the identified least efficient phase includes deasserting one or more switching control signals of the identified least efficient phase.
[0047] The method also includes recording an identifier of the least efficient phase in a memory device. Reducing the amount of power delivered to the load by the identified least efficient phase includes reading the recorded identifier from the memory device and disabling the least efficient phase corresponding to the recorded identifier.
[0048] In the method, identifying the least efficient phase includes identifying the phase of the plurality of phases having the widest pulse width measurement as the least efficient phase.
[0049] In this method, the comparison of the plurality of pulse width measurements is one of a plurality of comparisons, each of the plurality of comparisons being performed for one of a plurality of successive phase cycles of the multi-phase power supply.
[0050] The multi-phase power supply includes an input for receiving a plurality of pulse width measurements including a pulse width measured for each phase of a plurality of phases in the multi-phase power supply, a comparison logic unit coupled to the input for identifying a least efficient phase of the plurality of phases based on performing a comparison of the plurality of pulse width measurements, and phase shedding logic coupled to the comparison logic unit for reducing an amount of power delivered to a load by the identified least efficient phase.
[0051] In a multi-phase power supply, phase shedding logic, in response to detecting a decrease in the load's power demand, reduces the amount of power delivered to the load by disabling an identified least efficient phase.
[0052] The multi-phase power supply further includes an input voltage sensor for measuring an input voltage of the multi-phase power supply, an output voltage sensor for measuring an output voltage of the multi-phase power supply, the output voltage of which is applied to a load, and a basic pulse width calculation unit for calculating a basic pulse width based on a switching frequency of the multi-phase power supply, the measured input voltages and the measured output voltages. The comparison logic unit performs a comparison of the multiple pulse width measurements by determining a difference between each of the pulse width measurements and a base pulse width.
[0053] In a multi-phase power supply, phase shedding logic reduces the amount of power delivered to a load by an identified least efficient phase by deasserting one or more switching control signals of the identified least efficient phase.
[0054] The multi-phase power supply further includes a memory coupled to the comparison logic unit for recording an identifier of the least efficient phase, the phase shedding logic reading the recorded identifier from the memory and disabling the least efficient phase corresponding to the recorded identifier to reduce the amount of power delivered to the load by the identified least efficient phase.
[0055] In a multi-phase power supply, the comparison logic unit identifies the phase of the plurality of phases having the widest pulse width measurement as the least efficient phase.
[0056] In a multi-phase power supply, the comparison logic unit performs a plurality of comparisons, including a comparison of a plurality of pulse width measurements, each of the plurality of comparisons being performed for one of a plurality of successive phase cycles of the multi-phase power supply.
[0057] The computing system includes a plurality of power phases, a set of one or more pulse width sensors for generating a plurality of pulse width measurements by measuring a pulse width for each power phase of the plurality of power phases, and a voltage regulator coupled to the plurality of power phases, for identifying a least efficient phase of the plurality of power phases based on performing a comparison of the plurality of pulse width measurements, and for reducing an amount of power delivered to a load by the identified least efficient phase.
[0058] In a computing system, a voltage regulator identifies a power phase having a widest pulse width measurement among a plurality of power phases as a least efficient phase, and in response to detecting a first decrease in power demand of the load, reduces an amount of power supplied to the load by disabling the identified least efficient phase.
[0059] The computing system further includes an input voltage sensor configured to measure an input voltage provided to the multiple power phases and an output voltage sensor configured to measure an output voltage generated by the multiple power phases. The voltage regulator calculates a base pulse width based on a switching frequency of the multiple power phases, the measured input voltage, and the measured output voltage, and performs a comparison of the multiple pulse width measurements based on determining a difference between each of the pulse width measurements and the base pulse width.
[0060] In the computing system, the voltage regulator further includes a memory for recording an identifier of the least efficient phase, the voltage regulator reading the recorded identifier from the memory and disabling the least efficient phase corresponding to the recorded identifier, thereby reducing the amount of power delivered to the load by the identified least efficient phase.
[0061] The computing system further includes a load, the load including one or more processing units, a memory system, and a set of one or more peripheral devices.
[0062] As used herein, the term "coupled to" may mean directly or indirectly coupled through one or more intervening components. Any of the signals provided via the various buses described herein may be time-shared with other signals and provided via one or more common buses. In addition, the interconnections between circuit components and blocks may be depicted as buses or single signal lines. Each of the buses may alternatively be one or more single signal lines, and each of the single signal lines may alternatively be a bus.
[0063] Certain embodiments may be implemented as a computer program product that may include instructions stored on a non-transitory computer-readable storage medium. These instructions may be used to program a general-purpose or special-purpose processor to perform the described operations. A computer-readable storage medium includes any mechanism for storing or transmitting information in a form (e.g., software, processing application) readable by a machine (e.g., a computer). Non-transitory computer-readable storage media may include, but are not limited to, magnetic storage media (e.g., floppy diskettes), optical storage media (e.g., CD-ROMs), magneto-optical storage media, read-only memory (ROM), random access memory (RAM), erasable programmable memory (e.g., EPROM and EEPROM), flash memory, or another type of medium suitable for storing electronic instructions.
[0064] Additionally, some embodiments may be implemented in a distributed computing environment in which a computer-readable storage medium is stored on and / or executed by two or more computer systems. Additionally, information transferred between computer systems may be pulled or pushed through a transmission medium connecting the computer systems.
[0065] In general, the data structures representing computing system 100 and / or portions thereof embodied in a computer-readable storage medium may be databases or other data structures that may be read by a program and used, directly or indirectly, to manufacture hardware including computing system 100. For example, the data structures may be behavioral or register-transfer level (RTL) descriptions of hardware functionality in a high-level design language (HDL) such as Verilog or VHDL. The descriptions may be read by a synthesis tool that may synthesize the descriptions to generate a netlist that includes a list of gates from a synthesis library. The netlist includes a set of gates that also represent the functionality of the hardware including computing system 100. The netlist may then be placed and routed to generate a data set that describes the geometric shapes that are applied to a mask. The mask may then be used in various semiconductor manufacturing processes to manufacture semiconductor circuits corresponding to computing system 100. Alternatively, the database on the computer readable storage medium may be a netlist (with or without a synthesis library) or a data set, or Graphic Data System (GDS) II data, as desired.
[0066] Although the operations of the methods herein are shown and described in a particular order, the order of operations of each method may be changed such that certain operations may be performed in the reverse order, or such that certain operations may be performed at least in part concurrently with other operations. In alternative embodiments, instructions or sub-operations of individual operations may be performed intermittently and / or alternatingly.
[0067] In the foregoing specification, the embodiments have been described with reference to certain exemplary embodiments thereof. It will be apparent, however, that various modifications and changes can be made thereto without departing from the broader scope of the embodiments as set forth in the appended claims. The specification and drawings are therefore to be regarded in an illustrative rather than a restrictive sense.
Claims
1. 1. A method of operating a multi-phase power supply, comprising: comparing pulse widths of phases in the multi-phase power source to one another and identifying a least efficient phase of the multiple phases based on a difference between each pulse width and a base pulse width; and reducing an amount of power delivered to a load by the identified least efficient phase. method.
2. generating a plurality of pulse width measurements by measuring a pulse width of each phase of the plurality of phases in the multi-phase power source; 2. The method of claim 1.
3. reducing the amount of power delivered to the load includes disabling the identified least efficient phase in response to detecting a first decrease in power demand of the load.
2. The method of claim 1.
4. identifying a next least efficient phase of the plurality of phases based on a comparison of pulse widths of the plurality of phases; and disabling the next least efficient phase in response to detecting a second decrease in power demand of the load. The method of claim 3.
5. Measuring an input voltage of the multi-phase power supply; Measuring an output voltage of the multi-phase power supply applied to the load; and calculating the base pulse width based on a switching frequency of the multi-phase power supply, the measured input voltage, and the measured output voltage.
2. The method of claim 1.
6. Reducing an amount of power delivered to the load by the identified least efficient phase includes deasserting one or more switching control signals of the identified least efficient phase.
2. The method of claim 1.
7. recording an identifier of the least efficient phase in a memory device; Reducing an amount of power delivered to the load by the identified least efficient phase comprises: reading the recorded identifier from the memory device; and disabling the least efficient phase corresponding to the recorded identifier.
2. The method of claim 1.
8. the comparison of the pulse widths of each of the plurality of phases is one comparison of a plurality of comparisons; each of the plurality of comparisons is performed for one of a plurality of successive phase cycles of the multi-phase power supply; 2. The method of claim 1.
9. 1. A multi-phase power supply, comprising: an input configured to receive a plurality of pulse width measurements, the pulse width measured for each phase of a plurality of phases in the multi-phase power source; a comparison logic unit coupled to the input and configured to compare the plurality of pulse width measurements with one another and identify a least efficient phase of the plurality of phases based on a difference between each pulse width measurement and a base pulse width; phase shedding logic coupled to the comparison logic unit and configured to reduce an amount of power delivered to a load by the identified least efficient phase. Polyphase power supply.
10. the phase shedding logic is configured to, in response to detecting a decrease in power demand of the load, reduce an amount of power delivered to the load by the identified least efficient phase by disabling the identified least efficient phase.
10. The multiphase power supply of claim 9.
11. an input voltage sensor configured to measure an input voltage of the multi-phase power supply; an output voltage sensor configured to measure an output voltage of the multi-phase power supply applied to the load; a base pulse width calculation unit configured to calculate the base pulse width based on a switching frequency of the multi-phase power supply, the measured input voltage, and the measured output voltage.
10. The multiphase power supply of claim 9.
12. the phase shedding logic is further configured to reduce an amount of power delivered to the load by the identified least efficient phase by deasserting one or more switching control signals of the identified least efficient phase.
10. The multiphase power supply of claim 9.
13. a memory coupled to the comparison logic unit and configured to record an identifier of the least efficient phase; The phase shedding logic comprises: reading the recorded identifier from the memory; and Disabling the least efficient phase corresponding to the recorded identifier; and and reducing an amount of power delivered to the load by the identified least efficient phase by 10. The multiphase power supply of claim 9.
14. the comparison logic unit is configured to perform a plurality of comparisons including a comparison of the plurality of pulse width measurements; each of the plurality of comparisons is performed for one of a plurality of successive phase cycles of the multi-phase power supply; 10. The multiphase power supply of claim 9.
15. 1. A computing system comprising: A plurality of power phases; a set of one or more pulse width sensors configured to measure a pulse width for each power phase of the plurality of power phases to generate a plurality of pulse width measurements; a voltage regulator coupled to the plurality of power phases; The voltage regulator comprises: comparing the plurality of pulse width measurements to one another and identifying a least efficient one of the plurality of power phases based on a difference between each pulse width measurement and a base pulse width; reducing an amount of power delivered to a load by the identified least efficient phase; and [0023] Computing system.
16. The voltage regulator comprises: and configured to: in response to detecting a first decrease in power demand of the load, reduce an amount of power delivered to the load by disabling the identified least efficient phase.
16. The computing system of claim 15.
17. an input voltage sensor configured to measure an input voltage provided to the plurality of power phases; an output voltage sensor configured to measure an output voltage generated by the plurality of power phases; The voltage regulator comprises: and calculating the base pulse width based on a switching frequency of the plurality of power phases, the measured input voltage, and the measured output voltage.
16. The computing system of claim 15.
18. the voltage regulator further comprising a memory configured to record an identity of the least efficient phase; The voltage regulator comprises: reading the recorded identifier from the memory; and Disabling the least efficient phase corresponding to the recorded identifier; and and reducing an amount of power delivered to the load by the identified least efficient phase by 16. The computing system of claim 15.
19. The load further comprises: The load is a set of one or more processing units; A memory system; one or more peripheral devices; 16. The computing system of claim 15.
20. 1. A method of operating a multi-phase power supply, comprising: comparing pulse widths of phases in the multi-phase power source to one another and identifying extrema for the phases based on a difference between each pulse width and a base pulse width; decreasing an amount of power delivered to a load by an extreme least efficient phase of the plurality of phases, or increasing an amount of power delivered to the load by an extreme most efficient phase of the plurality of phases. method.
21. The method of claim 20, wherein identifying the least efficient phase includes identifying a phase of the plurality of phases having a widest pulse width as the least efficient phase.
2. The method of claim 1.
22. The comparison logic unit configured to identify a phase of the plurality of phases having a widest pulse width measurement as the least efficient phase.
10. The multiphase power supply of claim 9.
Citation Information
Patent Citations
Power supplying device
JP1997204240A
Power supply system
JP2009159691A
Control power supply program and power supply device
JP2020022315A
Power supply system for optimizing overall efficiency using artificial intelligence
US20200313570A1
