Improving Light Load Recovery in Multilevel Converters

By introducing a ripple signal or additional switching cycles, the multilevel converter addresses the issue of fly capacitor voltage rebalancing at low loads, enhancing efficiency and reducing switch stress.

JP2024542147AInactive Publication Date: 2024-11-13MURATA MFG CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
JP2024526931
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Priority Date
2021-12-23
Filing Date
2022-10-28
Publication Date
2024-11-13
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Multilevel converters face challenges in light load conditions where fly capacitors cannot rebalance their voltages efficiently, leading to stress on switches and inefficiencies due to the reliance on output load current for charge balancing.

Method used

Generating a ripple signal in the control loop of the multilevel converter to induce temporary positive and negative currents for fly capacitor balancing, or adding additional switching cycles at zero-current crossings to rebalance capacitor voltages during low load conditions.

Benefits of technology

This approach enables efficient charge balancing of fly capacitors even at low loads, reducing stress on switches and improving converter performance under varying input conditions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024542147000001_ABST
    Figure 2024542147000001_ABST
Patent Text Reader

Abstract

A circuit and method for solving the light load problem in a multilevel converter by generating a ripple signal in the control loop of the multilevel converter that generates a large output current ripple during light load conditions. This added current ripple does not change the average output current, but creates temporary positive and negative currents that can be used to balance and charge / discharge the fly capacitors of the multilevel converter. An alternative approach is to add an additional switching cycle for the fly capacitor when the output ripple current crosses zero.
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] [CROSS REFERENCE TO RELATED APPLICATIONS - PRIORITY CLAIM] This application claims priority to U.S. Provisional Application No. 63 / 276,962, filed November 8, 2021, relating to "Light-Load Recovery in a Multi-Level Converter," and U.S. Patent Application No. 17 / 560,683, filed December 23, 2021, relating to "Light-Load Recovery in a Multi-Level Converter," both of which are incorporated by reference in their entireties herein.

[0002] The present invention relates to electronic circuits, and more particularly to power converter circuits, including DC-DC power converter circuits. [Background technology]

[0003] Many electronic products, particularly mobile computing and / or communication products and components (e.g., mobile phones, laptops, ultrabook computers, tablet devices, LCD and LED displays), require multiple voltage levels. For example, radio frequency (RF) transmitter power amplifiers may require relatively high voltages (e.g., 12V or more), while logic circuits may require lower voltage levels (e.g., 1-3V). Still other circuits may require intermediate voltage levels (e.g., 5-10V).

[0004] DC power converters are often used to generate low or high voltages from a common power source such as a battery, solar cell, fuel cell, or rectified AC power. A power converter that produces a lower output voltage level from a higher input voltage source is commonly known as a buck converter, which produces an output voltage V OUT is the input voltage V INA power converter that produces a higher output voltage level from a lower input voltage source has a lower input voltage than V OUT V IN Some power converters are commonly known as boost converters because their input is greater than 100V. Some power converters may be either buck or boost converters, depending on which terminals are used for the input and output. Some power converters may provide an inverting output. Some power converters, known as multilevel converters, use charge transfer capacitors, commonly called "fly capacitors" or "pump capacitors."

[0005] One drawback of multilevel converters is the flyback capacitor C Fx The problem with this converter is that it can only change voltage by using the output load current to pump energy into or out of the converter. When the load is light (zero or very low, e.g., less than about 10% of the absolute maximum current specification of a typical converter), the fly capacitor C Fx Even if the fly capacitor C has charge balancing circuitry, it cannot rebalance or reregulate the voltage. Fx To maintain the average voltage of and avoid switch stress, a fly capacitor C Fx This characteristic is evident, for example, when switching power sources (e.g., from one battery to another, or from a battery to a rectified AC mains supply) at the input V of a multilevel converter. IN This is especially problematic when there are line steps in V. IN If there is a sudden change in load current, the fly-out capacitor C Fx cannot readjust their respective voltages.

[0006] SUMMARY OF THE DISCLOSURE Accordingly, there is a need for circuits and methods for improving light load recovery in multilevel converters. The present invention addresses this need. Summary of the Invention

[0007] The present invention includes a circuit and method that solves the light load problem of a multilevel converter by generating a ripple signal in the control loop of the multilevel converter that causes a large output current ripple during light load conditions. This added current ripple does not change the average output current, but creates temporary positive and negative currents that can be used to balance and charge / discharge the fly capacitor of the multilevel converter. Another approach is to add extra switching cycles to the fly capacitor when the output ripple current is close to or crosses zero.

[0008] A circuit embodiment includes a multi-level converter including an M-level converter cell configured to convert an input voltage applied to an input terminal to an output voltage at an output terminal in response to a switch control input; a controller coupled to the M-level converter cell, the controller configured to monitor an output voltage of the M-level converter cell and dynamically generate a switch control input to the M-level converter cell; and a sub-harmonic signal generator coupled to the controller and configured to selectively inject a sub-harmonic signal into a signal path of the controller to induce a sub-harmonic ripple at an output terminal of the M-level converter cell.

[0009] The methods include improving light load recovery in an M-level multilevel converter configured to convert an input voltage applied to an input terminal to an output voltage at an output terminal in response to an input signal from a controller that defines a charge cycle and a discharge cycle of the multilevel converter. A first method injects a subharmonic signal into a signal path of the controller to induce a subharmonic ripple at an output terminal of the M-level multilevel converter. A second method adds an additional charge cycle at a zero-current crossing point after the discharge cycle and adds an additional discharge cycle at the zero-current crossing point after the charge cycle.

[0010] The details of one or more embodiments of the invention are set forth in the accompanying drawings and the description below. Other features, objects, and advantages of the invention will be apparent from the description and drawings, and from the claims. [Brief description of the drawings]

[0011] [Figure 1] FIG. 1 is a block diagram of a circuit including a prior art power converter.

[0012] [Figure 2A] FIG. 1 is a schematic diagram of a portion of a prior art two-level DC-DC buck converter circuit including a particular converter cell.

[0013] [Figure 2B] 2B is a graph showing the voltage level at node LX as a function of time in the circuit of FIG. 2A.

[0014] [Figure 3A] FIG. 1 is a schematic diagram of a portion of a prior art three-level DC-DC buck converter circuit including a particular converter cell.

[0015] [Figure 3B] 3B is a graph showing the voltage level at node LX as a function of time in the circuit of FIG. 3A.

[0016] [Figure 4] FIG. 2 is a block diagram of one embodiment of a control circuit for an M-level converter cell coupled to an output block including an inductor L and an output capacitor COUT.

[0017] [Diagram 5] FIG. 13 is a block diagram of one embodiment of a control circuit for an M-level converter cell illustrating possible modifications.

[0018] [Figure 6] 1 is a graph of current through an inductor of a three-level converter as a function of time under a first set of load conditions.

[0019] [Figure 7] 11 is a graph of current through an inductor of a three-level converter as a function of time under a second set of load conditions.

[0020] [Figure 8] 1 is a graph of the current through an inductor of a multilevel converter as a function of time under conditions that may result in a negative current flow through the inductor.

[0021] [Figure 9A] 1 is a graph of the current through the inductor L of the multilevel converter as a function of time, illustrating the addition of switching cycles to improve light load recovery.

[0022] [Figure 9B] 2 is a graph of the voltage of a particular fly capacitor C of a multilevel converter as a function of time.

[0023] [Figure 10] FIG. 2 is a schematic diagram of one embodiment of a subharmonic signal generating circuit.

[0024] [Figure 11]FIG. 2 is a schematic diagram of one embodiment of a subharmonic PWM edge delay circuit.

[0025] [Figure 12] 1 is a process flow diagram illustrating a first method for improving light load recovery in an M-level multilevel converter.

[0026] [Figure 13] 1 is a process flow diagram illustrating a second method for improving light load recovery in an M-level multilevel converter.

[0027] Like reference numbers and designations in the various drawings indicate like elements. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0028] 1 is a block diagram of a circuit including a prior art power converter 100. In the illustrated example, the power converter 100 includes a converter cell 102 and a controller 104. The converter cell 102 receives an input voltage V from a voltage source 106 (e.g., a battery) between terminals V1+ and V1− (common). IN Receives the input voltage V IN The output voltage V between terminals V2+ and V2- (common) OUT The output voltage V OUT is generally the output capacitor C OUT , across which may be connected a load represented as an equivalent resistance R. In some embodiments of power converter 100, auxiliary circuits (not shown) such as bias voltage generator(s), clock generators, voltage control circuits, etc. may be present and coupled to converter cells 102 and controller 104.

[0029] The controller 104 receives a set of input signals and generates a set of output signals. Some of these input signals arrive along signal paths 110 connected to the converter cells 102. Some of the input signals carry information indicative of the operating state of the converter cells 102. The controller 104 also generally receives at least a clock / timing signal CLK (although some controllers may be configured for asynchronous operation, e.g., hysteretic controllers or constant-on-time controllers), and one or more external input / output signals I / O, which may be analog, digital (coded or direct signal line), or a combination of both. Based on the received input signals, the controller 104 controls the converter cells 102 to generate V IN V OUT In some embodiments, auxiliary circuitry (not shown) generates a set of control signals that are returned to the converter cells 102 on signal paths 110 that control internal components of the converter cells 102 (e.g., internal switches such as FETs, particularly MOSFETs) to convert the clock signal CLK, the input / output signals I / O, and the general power supply voltage V DD and the transistor bias voltage V BIAS Various signals, such as various voltages, may be provided to the controller 104 (and optionally directly to the converter cells 102).

[0030] In some power converter designs, the converter cells 102 use inductors as energy storage elements. For example, FIG. 2A is a schematic diagram of a portion of a prior art two-level DC-DC buck converter circuit 200 that includes a particular converter cell 102a. Within the converter cell 102a, a set of two switches QH, QL are connected to the V IN An energy storage inductor L is coupled in series between a node L between the set of switches QH, QL (applied to terminal V1+) and a common reference voltage (e.g., circuit ground GND coupled to terminal V1−). X output capacitor C, which provides smoothing and energy storage OUT The output capacitor C OUTThe voltage across the OUT and is coupled to a load R.

[0031] Inductor L and output capacitor C OUT One function of the switches QH, QL is the transfer and storage of energy. Part of the controller circuitry of the converter cell 102a typically includes a pulse width modulation (PWM) duty cycle controller (not shown) that is coupled to the control inputs (e.g., gates of MOSFETs) of the switches QH, QL to alternately enable (close or "ON") and disable (open or "OFF") the switches QH, QL to control the flow of energy to the load R. The PWM duty cycle controller typically operates in response to a clock signal and a feedback voltage V OUT The feedback voltage allows the PWM duty cycle controller to vary the duty cycle of the PWM control signals to the switches QH, QL to offset changes in the load R, thereby reducing V OUT The PWM duty cycle controller may be part of controller 104 in FIG. 1, but may also be a stand-alone circuit.

[0032] In the example shown in FIG. 2A, the converter cell 102a has two states: QH closed and QL open (node ​​L X Voltage level at = V IN ), or QH open and QL closed (node ​​L X FIG. 2B shows the switching between only the voltage level at node L as a function of time in the circuit of FIG. 2A. X Graph line 202 is a graph showing the voltage levels at node L when switches QH and QL toggle between only two available switch states (i.e., QH closed and QL open, or QH open and QL closed). X is the average voltage level at node L. XThe time in each switch state determines the amplitude of the average voltage at A power converter 100 based on such a converter cell 102a is also known as a switched mode power supply (SMPS) or DC-DC converter.

[0033] As can be seen by considering FIG. 2B, the inductor L is connected to the node L X At GND to V IN You will notice a large jump in voltage level from V to GND and back to GND. The resulting voltage ripple across inductor L is OUT requires a significant amount of filtering to smooth V, which generally means that the inductor L is large; the higher the voltage across inductor L, the larger the inductance required. Additionally, the switches QH and QL are IN In some cases, the switches QH, QL may need to withstand the full voltage range from I to GND, which generally means that the switches QH, QL are physically large if implemented as FETs.

[0034] The voltage ripple across the inductor L and the voltage swing across any one switch are V IN From V OUT This can be reduced by adding more series switches and charge transfer capacitors as energy storage elements to transfer charge to the converter. The charge transfer capacitors are commonly known as "fly capacitors" or "pump capacitors" and may be internal components to an integrated circuit embodiment of the converter circuit, or external components coupled to the remaining integrated circuit embodiment of the converter circuit.

[0035] For example, Figure 3A is a schematic diagram of a portion of a prior art three-level DC-DC buck converter circuit 300 that includes a particular converter cell 102b. A set of four switches, QH2, QH1, QL1, and QL2, are connected to V INA fly capacitor C1 is coupled in series between the “high-side” node N between switches QH2 and QH1 (applied at V1+) and a common reference voltage (e.g., circuit ground GND coupled to terminal V1−). H From the "low-side" node N between switches QL1 and QL2, L An energy storage inductor L is coupled to node L between the innermost set of switches QH1 and QL1. X to the output capacitor C OUT Again, the output capacitor C OUT The voltage across the OUT In some embodiments, the inductor L may be external to the integrated circuit embodiment of the converter circuit 300, in which case the node L X may be considered to be at the output terminal of the converter circuit 300.

[0036] In the illustrated example, the presence of fly capacitor C1 in converter circuit 300 results in a 100 volt supply to node L, respectively, as shown in Table 1 below. X Four switch states are possible which produce one of three "node" voltage levels at [Table 1]

[0037] FIG. 3B shows the output of node L as a function of time in the circuit of FIG. 3A. X Graph line 302 shows the voltage levels at the switch between GND and level 2 (i.e., V IN / 2) Node L when cycling between switch states X 3B, the inductor L is the average voltage level at V IN / 2 (Level 2) and back to GND, Node L XThe resulting reduced voltage ripple across the inductor L is OUT This requires much less filtering to smooth the input, and allows for the use of smaller switches when implemented as FETs.

[0038] The topology of converter cell 102a in FIG. 2A is commonly known as a buck converter. The topology of converter cell 102b in FIG. 3A is commonly known as a "multilevel buck converter," or more specifically, a three-level buck converter, which includes an additional pair of series switches and corresponding fly capacitors C2 ... Cn (commonly referred to as "C Fx ") to at least five levels.

[0039] As with voltage levels 2a and 2b in Table 1 of converter circuit 300, some levels are X This has multiple ways of generating a selected voltage level at the fly capacitor C based on the particular state selected. Fx It allows charging or discharging all fly capacitors C Fx has a target average voltage to maintain the proper output level. For an M-level converter, where M ≥ 3 and x = 1, 2, ... [M-2], the fly capacitor C Fx The target voltage for is:

number

[0040] A voltage detector (not shown), which can be a simple comparator type circuit, is connected to the fly capacitor C Fx A reference voltage V represents the desired target voltage of REF The corresponding fly capacitor C Fx may be used to sense the voltage across

[0041] Essentially, a multilevel converter combines a charge pump capacitive voltage converter and an inductor-based power converter in one structure. In a multilevel converter, the inductor L is coupled to the flyback capacitor C Fx The capacitors act as a virtual current source moving all the charge between them. This creates a highly efficient form of charge transfer, substantially reducing or eliminating the charge redistribution losses that inherently occur when capacitors with different voltages are connected in parallel.

[0042] As mentioned above, one of the drawbacks of multilevel converters is the flyback capacitor C Fx However, the only thing that can change the voltage is by using the output load current to move energy in and out. When the load is light (zero or very low, e.g., less than about 10% of the absolute value of the maximum current specification of a typical converter), even with a charge-balancing circuit, the fly capacitor C Fx cannot rebalance or reregulate the voltage. To maintain the average voltage and avoid stress on the switches, a fly capacitor C Fx Balancing charge is required at the input V of a multilevel converter, as can occur when switching power sources (for example, from one battery to another, or from a battery to a rectified AC mains supply). IN This is especially problematic when there are line steps in V. IN If there is a sudden change in the load current, the fly-out capacitor C Fx cannot readjust their respective voltages.

[0043] For example, V IN When V rises rapidly, IN All states referenced to V are generally raised by the same amount (for example, in Table 1, voltage levels 2a and 3 are both V IN For all intermediate voltage levels (all levels other than voltage level 1 and the highest voltage level), all ground reference states are V INThe voltage remains the same as before the change, V IN All states referenced to V IN The steady-state solution is to add a fly capacitor C Fx The ground reference state is increased so that V IN The reference state needs to fall. In the example of Table 1, for voltage level 2a, the voltage across the fly capacitor C1 is V IN Based on the increase in the new V IN To allow this settling (charge balancing), the converter output current must be adjusted to the fly capacitor C so that the respective target voltage is reached. Fx However, when the output current is zero or very low, charge balancing occurs very slowly, if at all, and the converter's target output voltage V OUT This makes adjustments difficult or impossible.

[0044] Conventional balance circuits use fly capacitors C Fx It is known that the fly capacitors can be rebalanced, but only by throwing away energy: they add additional current paths (current sources, resistors, etc.) to add or remove energy from the fly capacitors, which, while accomplishing the balancing task, is very energy-wasting. Moreover, the additional current paths need to supply large amounts of current to quickly rebalance the fly capacitors.

[0045] The present invention encompasses circuits and methods that solve the light load problem of multilevel converters in an energy efficient manner by generating a ripple signal in the control loop of the multilevel converter that can cause large output current ripples under light load conditions. This added current ripple does not change the average output current, but creates temporary positive and negative currents that can be used to balance and charge / discharge the fly capacitors of the multilevel converter. An alternative energy efficient approach is to add additional switching cycles for the fly capacitors when the output ripple current is close to or crosses zero.

[0046] Example Control Circuit of an M-Level Converter Cell

[0047] Figure 4 shows the inductor L and the output capacitor C OUT 1 is a block diagram of an embodiment of a control circuit 400 for an M-level converter cell 401 coupled to an output block 403 having an inductor L (conceptually, the inductor L can be considered to be included within the M-level converter cell 401). This exemplary control circuit 400 follows the teachings described in commonly assigned U.S. Patent Application No. 63276923, entitled “Controlling Charge-Balance and Transients in a Multi-Level Power Converter” [Attorney Docket No. PER-370-PROV], filed November 8, 2021, the contents of which are incorporated by reference. However, the present invention may be used in combination with other types of control circuits for the M-level converter cell 401.

[0048] The control circuit 400 functions as a control loop coupled to the output of the M-level converter cell 401 and the switch control input of the M-level converter cell 401. In general, the control circuit 400 monitors the output (e.g., voltage and / or current) of the M-level converter cell 401 and controls V INThe control circuit 400 is configured to dynamically generate a set of switch control inputs to the M-level converter cell 401 that attempt to stabilize the output voltage and / or current at a particular value, taking into account variations in the input and output load. In alternative embodiments, the control circuit 400 may be configured to monitor the inputs (e.g., voltages and / or currents) of the M-level converter cell 401 and / or internal nodes (e.g., voltages across one or more fly capacitors or currents through one or more power switches) of the M-level converter cell 401. Thus, most generally, the control circuit 400 may be configured to monitor voltages and / or currents at nodes (e.g., input terminals, internal nodes, output terminals) of the M-level converter cell 401. The control circuit 400 may be incorporated into or configured separately from the overall controller 104 of the power converter 100 embodying the M-level converter cell 401.

[0049] The first block includes a feedback controller 402, which can be a conventional controller such as a fixed frequency voltage mode or current mode controller, a constant on-time controller, a hysteretic controller, or any other variation. The feedback controller 402 outputs the V OUT In alternative embodiments, the feedback controller 402 may be configured to monitor the input of the M-level converter cell 401 and / or an internal node of the M-level converter cell 401. The feedback controller 402 is shown coupled to V OUT A signal is generated that directly or indirectly indicates the voltage at OUT Determine what needs to be done in the M-level converter cell 401 to maintain the desired value of: charge, discharge, or tri-state (i.e., no current flows, open).

[0050] In the illustrated example, the feedback controller 402 includes a feedback circuit 404, a compensation circuit 406, and a PWM generator 408. The feedback circuit 404 is, for example, a PWM circuit for OUT (or VOUT attenuated version of the desired V OUT Compare it to a reference voltage that represents the target voltage (which can be dynamic) and V OUT A control signal V is applied to indicate whether V is above or below the target voltage. COMP The feedback loop voltage detector may include a feedback loop voltage detector that outputs: The feedback loop voltage detector may be implemented using a comparison device 405, such as an operational amplifier (opamp) or a transconductance amplifier (gmamp).

[0051] The compensation circuit 406 is configured to stabilize the closed loop response of the feedback controller 402 by avoiding unintentional generation of positive feedback that may cause oscillations and by controlling overshoot and ringing in the step response of the feedback controller 402. The compensation circuit 406 may be implemented in a known manner and may include LC and / or RC circuits. The PWM generator 408 generates the actual PWM control signal that ultimately sets the duty cycle of the switches of the M-level converter cell 401.

[0052] In some embodiments, the PWM generator 408 may be, for example, V OUT The magnitude of the difference between the reference voltage and 401 (e.g., indicating that some levels of the M-level converter cell 401 should be bypassed to reach a higher or lower level), and the direction of that difference (e.g., V OUT In other embodiments, the optional control signal CTRL may be passed from the output of the compensation circuit 406, or from the output of the feedback circuit 404, or from, for example, V OUT One purpose of the optional control signal CTRL is to control the V OUT This is for advanced control algorithms where it can be useful to know how far away V is from the target output voltage, and therefore V OUT is significantly under-regulated, allowing faster charging of inductor L.

[0053] The second block contains the M-level controller 410, whose main function is to generate the desired V while maintaining the charge-balance state of the fly capacitors in the M-level converter cell 401 whenever an output voltage level is selected, regardless of what switch states were used in the past. OUT The aim is to select the switch state that produces

[0054] The M-level controller 410 includes a voltage level selector 412 that receives the PWM control signal and possibly an additional control signal CTRL. OUT and / or V IN , and in some embodiments, a corresponding fly capacitor C Fx HIGH / LOW state signal from a voltage detector coupled to C Fx_H / L The function of the voltage level selector 412 is to convert the received signal to a target output voltage level (e.g., cycle by cycle). The voltage level selector 412 typically includes at least OUT and V IN Taking into account the above, it is determined which target level the output of the M-level converter cell 401 should be charged or discharged at the desired rate.

[0055] The output of the voltage level selector 412 is coupled to an M-level switch state selector 414, which generally includes a fly capacitor C Fx Status signal C from the capacitor voltage detector Fx_H / L. Given the target levels generated by the voltage level selector 412, the M-level switch state selector 414 determines which switch state for the desired output level should be optimal for capacitor charge balance. The M-level switch state selector 414 may be implemented, for example, as a look-up table (LUT), or as a comparison circuit and combinational logic or a more generalized processor circuit. The output of the M-level switch state selector 414 is coupled to the switches of the M-level converter cell 401 (through appropriate level shifter and driver circuits, if required for the particular converter cell) and contains the switch state settings determined by the M-level switch state selector 414 (which selects the configuration of the switches in the M-level converter cell 401 that corresponds to the selected target level).

[0056] Typically (but not always), the voltage level selector 412 and the M-level switch state selector 414 change their state only when the PWM signal changes. For example, when the PWM signal goes high, the voltage level selector 412 selects which level will result in charging the inductor L, and the M-level switch state selector 414 sets which version of that level to use. Then, when the PWM signal goes low, the voltage level selector 412 selects which level should discharge the inductor L, and the M-level switch state selector 414 sets which version of that level to use. Thus, the voltage level selector 412 and the M-level switch state selector 414 generally change state only when the PWM signal changes (the PWM signal is effectively a clock signal). However, there may be situations or events where it is desirable for the CTRL signal to change the state of the voltage level selector 412. Additionally, when a fly capacitor C in the M-level converter cell 401 changes, such as when a significant mid-cycle imbalance occurs, the voltage level selector 412 selects which level will result in charging the inductor L, and the M-level switch state selector 414 sets which version of that level to use. Fx C from a voltage detector coupled to Fx_H / LThere may be situations or events in which it is desirable for a state signal(s) to cause the M-level switch state selector 414 to select a particular configuration of power switch settings. In some embodiments, it may be useful to include a timing function that causes the M-level switch state selector 414 to periodically re-evaluate the optimal version of the state, for example, to avoid getting "stuck" at one level for too long, potentially causing a charge imbalance.

[0057] In an embodiment utilizing the teachings described in the above-referenced patent application entitled “Controlling Charge-Balance and Transients in a Multi-Level Power Converter,” the M-level controller 410 controls the L X Each time a voltage level at a node is selected, the fly capacitor C Fx The control method for the M-level converter cell 401 selects essentially the optimal switch state that moves the L toward a charge-balanced state. Thus, such a multi-level converter circuit can select different switch states or L every switching cycle without having to keep track of previous switch states or sequences of switch states. X The voltage level can be freely selected.

[0058] One notable advantage of the control circuit shown in FIG. 4 is that it enables the generation of voltages in the boundary zones between voltage levels that represent output voltages unattainable with conventional multilevel DC-DC converter circuits.

[0059] In an alternative embodiment of an uncontrolled charge pump, the feedback controller 402 and voltage level selector 412 may be omitted and instead the clock signal CLK may be applied to an M-level switch state selector 414. The M-level switch state selector 414 selects the fly capacitor C regardless of what switch state was previously used. FxGenerate a pattern of switch state settings that periodically charge-balances the V IN In the event of a change in or an abnormal event, the system generally always Fx Ensure that the charge balance is determined.

[0060] In some embodiments, the M-level switch state selector 414 measures the current I through the inductor L via an optional current measurement input 416, which may be implemented in a conventional manner. L can be taken into consideration.

[0061] Although FIG. 4 shows a particular embodiment of a control circuit for an M-level converter cell modified in accordance with the present invention, it should be understood that other control circuits can be adapted or devised to provide appropriate switching signals to the switches in the converter cell.

[0062] First embodiment

[0063] In accordance with the present invention, a control circuit the same as or similar to that of FIG. 4 may be modified to solve the light load problem of a multilevel converter. FIG. 5 is a block diagram 500 of one embodiment of a control circuit for an M-level converter cell 401 illustrating possible modifications. In particular, a "ripple" generator circuit may be inserted into the control feedback loop at one of several insertion nodes and configured to inject a sub-harmonic signal. The details of the ripple generator circuit may vary slightly from one insertion node to another, but the function in each case is to induce sub-harmonic ripple in the inductor L of the multilevel converter when a light load condition occurs. The sub-harmonic ripple does not change the average output current through the inductor L, but does increase the fly-capacitor current through the inductor L. Fx This produces momentary positive and negative currents that can be used to balance and charge / discharge the

[0064] The injection of the subharmonic signal can occur at any of several insertion nodes, with each insertion node offering different advantages and disadvantages. The subharmonic signal can be an AC waveform, such as, for example, a square wave, a sine wave, a triangular wave (sawtooth wave), etc., so long as the average voltage of the applied signal is zero. In some embodiments, the subharmonic signal can be injected by modifying the timing of a digital waveform (e.g., a PWM signal).

[0065] As an example, FIG. 5 shows four insertion nodes A, B, C, and D where injection of subharmonic signals may be beneficial.

[0066] Insertion node A1: Injecting a subharmonic signal into the insertion node A1, thus V OUT Modulating the feedback signal from to the input of the comparator 405 is relatively simple to implement. A drawback of injecting a subharmonic signal into the insertion node A1 is that the injected signal may be affected by the compensation circuit 406, as the injected subharmonic signal is also filtered, making the resulting amplitude more difficult to control. For example, if the injected signal is a 1V AC sine wave, the filtering action of the compensation circuit 406 will reduce the effective amplitude of the injected signal. As another example, if the injected signal is a square wave, the low-pass filtering action of the compensation circuit 406 may make the injected signal appear more sinusoidal.

[0067] Insertion Node A2: Injecting the subharmonic signal into insertion node A2, which is the reference side of comparator 405, has the same advantages and disadvantages as injecting into insertion node A1.

[0068] Insertion Node B: Injecting a subharmonic signal into insertion node B, which is the output of comparison device 405, has the same advantages and disadvantages as injecting into insertion nodes A1 or A2, since injecting into the output is the same as injecting into the input, only divided by the gain of comparison device 405.

[0069] Insertion Node C: Injecting a sub-harmonic signal into insertion node C after compensation circuit 406 can be very effective. In current mode control embodiments, the sub-harmonic signal can instead be injected into the current sense circuit (usually part of PWM generator 408) to achieve the same effect.

[0070] Insertion Node D: Once the PWM signal is generated, a sub-harmonic signal can be injected into the insertion node D by modulating the PWM duty cycle to add a variable delay to the rising and / or falling edges of the PWM waveform. For example, referring to Table 1, if for a particular converter cell cycle the duty cycle is 60% for level 2a and 40% for level 1a, the sub-harmonic generator can delay the falling edge of the duty cycle for level 2a so that the ratio goes from 65% to 35% in one cycle and 55% to 45% in the next cycle, thereby maintaining the selected 60% to 40% ratio on average.

[0071] For each of the insertion nodes A1, A2, B, C, and D, the injected sub-harmonic signal causes a change in the PWM signal that controls the duration (duty cycle) of the switching state of the M-level converter cell 401. As will be appreciated, there may be other insertion nodes and methods of injecting sub-harmonic signal waveforms, so long as the average voltage of the applied signal is zero, thereby inducing sub-harmonic ripple in the inductor L when light load conditions occur.

[0072] The injected subharmonic signal can be approximately half the switching frequency of the switches of the M-level converter cell 401 (e.g., switches QH2, QH1, QL1, and QL2 in FIG. 3A). This results in minimal loop interference and minimal output ripple degradation. The lower subharmonic frequency can be achieved by using the fly capacitor C FxBy sustaining multiple subsequent cycles of charging and discharging the MOSFET, it becomes easier to increase the charge / discharge current. However, such low subharmonic frequencies can cause side effects such as increased output current and voltage ripple in more systems.

[0073] 6 is a graph 600 of current through an inductor of a three-level converter as a function of time under a first set of load conditions. The three-level converter may, for example, be similar to the converter shown in FIG. 3A, but with a control circuit as shown in FIG. 5, having a subharmonic signal selectively injectable at the appropriate insertion node.

[0074] In the illustrated example, the average current through inductor L of the three-level converter is about 8 amps until time T1. When the inductor current is high, the fly capacitor C Fx However, at time T1, the average inductor current decreases due to the reduced load (shown as the time between T1 and T2). The average inductor current is proportional to the fly capacitor C Fx If there is not enough current to rebalance, it may approach or reach approximately 0 amps (shown as the time between T2 and T3).

[0075] Between times T2 and T3, the fly capacitor C Fx If there are sufficient residual voltages on the MOSFETs 1 and 2 that are equal to or greater than their respective target voltages, there is no need to inject a subharmonic signal into the converter control circuitry. However, once the residual voltages fall below their respective target voltages, such as under light load conditions at time T3, the fly capacitor C Fx The converters are unable to readjust their respective voltages to respond to the line step until the load current is present for a long enough period of time that the average current through the converter inductor L approaches or reaches approximately 0 amps and the fly capacitor C Fx If the load becomes unbalanced, the associated control circuit will detect the fly capacitor C FxIn order to help balance the inputs, the M-level converter cell 401 is forced to inject a sub-harmonic signal (here shown starting at time T3). This injected sub-harmonic ripple modifies the duty cycle of the switching states of the M-level converter cell 401.

[0076] In the example shown in FIG. 6, referring to Table 1, a particular V OUT On the other hand, the three-level converter has a 0V L X A voltage level 1 (L1) having a voltage V IN / 2L X The L1 duty cycle alternates between voltage levels 2a or 2b (2L) with a voltage. A complete cycle (CyX) consists of a certain amount of time spent on L1 and a certain amount of time spent on L2. In the illustrated example, in a complete cycle Cy1, the voltage level L2 duty cycle increases and the voltage level L1 duty cycle decreases. A longer L2 duty cycle in a complete cycle charges the inductor L more, increasing the inductor current to become positive in the next complete cycle Cy2 (about 1.5A in this example). Conversely, in a complete cycle Cy2, the voltage level L1 duty cycle increases and the voltage level L2 duty cycle decreases. A longer L1 duty cycle in a complete cycle discharges the inductor L more, decreasing the inductor current to become negative in the next complete cycle Cy3 (about -0.5A in this example). In a complete cycle Cy3, the L2 duty cycle increases again and the L1 duty cycle decreases again, which makes the average current of the inductor L positive in the next complete cycle Cy4. Similarly, in a complete cycle Cy4, the L2 duty cycle decreases again and the L1 duty cycle increases again, which makes the average current in inductor L negative in the next complete cycle (not shown).

[0077] Importantly, the positive and negative excursions resulting from the injection of a subharmonic signal are Fx It is sufficient to allow the voltages to be balanced by a rebalancing circuit (not shown) in a known manner.

[0078] Note that this process requires the control circuit to take into account whether the average current through the inductor L is negative or positive. A negative average current requires reversing the required state of the switches in the M-level converter cell 401; therefore, when a negative current is present, the switches that are normally in a charging state should be set to a discharging state (see also the discussion below regarding FIG. 8).

[0079] In some embodiments, it may be useful to inject the subharmonic signal only when the average current through inductor L is about 0 amps. In other embodiments, it may be useful to scale the degree of injection of the subharmonic signal as a function of the absolute value of the average current through inductor L. This approach allows for a smoother transition and avoids abrupt mode changes that may cause problems in the system response. Scaled injection of the subharmonic signal can be relatively easy to implement in current mode control systems where the target current is readily available. The scaled injected subharmonic signal may be scaled in amplitude, or may be scaled by changing the duty cycle of the injection (i.e., full scale injection for a period and then off for a period to obtain an average amount of injected subharmonic signal), or may be scaled by some combination of these two methods.

[0080] For example, Figure 7 is a graph 700 of the current through the inductor of a three-level converter as a function of time under a second set of load conditions. The three-level converter may include a control circuit as shown in Figure 5 with a sub-harmonic signal selectively injectable at an appropriate insertion node, which may be similar to the converter shown in Figure 3A, for example.

[0081] In the illustrated example, the average current through inductor L of the three-level converter is about 8 amps until time T1. At time T1, the output load drops and the average inductor current decreases (shown as the time between T1 and T2) and may approach 0 amps (shown as the time between T2 and T3). In the illustrated example, the average current through inductor L between times T1 and T2 is about 1.1 A. Therefore, it may be useful to start injecting some less subharmonic signal than the maximum amount available to allow some charge balance to occur. If the absolute average current continues to drop (e.g., as shown after time T3), the amount of subharmonic signal injected may increase to the maximum amount available.

[0082] In some implementations, the multi-level controller needs to predict whether the current subharmonic injection cycle will result in a positive or negative average current. In the case of a negative average current, the controller will increase the fly capacitor(s) C Fx For example, referring to FIG. 3A and Table 1 above, voltage level 2a would normally flow through switch QH2, fly capacitor C1, and switch QL1 to V IN Voltage level 2b is typically considered to be a charge state coupling circuit ground through switch QL2, fly capacitor C1, and switch QH1 to inductor L1. However, these state indications assume an average positive current flow through inductor L. If the average current flow through inductor L is negative, voltage level 2a would be a charge state coupling inductor L to V IN terminal is considered to be in a discharging state, with current flowing through the inductor L, and voltage level 2b is considered to be in a charging state, with current flowing through the inductor L to circuit ground. More generally, when the average current is negative, the fly capacitor C Fxinstead reverses its switch state to the complementary setting, normally discharging the fly capacitor C Fx instead reverses its switch state to its complementary setting.

[0083] 8 is a graph 800 of current through an inductor of a multilevel converter as a function of time under conditions that may result in a negative current flow through the inductor. Graph 800 illustrates a subharmonic injection cycle having a starting current that may be determined by measurement and a target current that is determined by control circuit 400 taking into account the injection of the subharmonic signal. The method of predicting whether the subharmonic injection cycle will result in a positive or negative average current may depend on the type of converter controller.

[0084] For a fixed frequency peak current mode controller, for each cycle, for a rising edge (i.e., between times T1 and T2 in the illustrated example), the current at the beginning of the rising edge (e.g., at time T1) is measured. The current at the end of the rising edge (e.g., at time T2) is the voltage V output by comparator 405. COMP can be easily estimated from (see Figure 4). COMP The value of is proportional to the target current and can therefore determine what the final output current should be – essentially, V COMP The voltage is used as a proxy to predict the output current. The average of these two values ​​is then calculated. If the calculated average current is negative, the controller will turn on the fly capacitor(s) C, as described above. FxThe definitions of the switch states that charge and discharge L must be reversed. For a falling edge (i.e., between times T2 and T3 in the illustrated example), the current at the start of the falling edge (e.g., at time T2) is measured, and the end current (e.g., at time T3) is estimated by knowing how much time is left in the fixed frequency cycle and the inductance and voltage of inductor L. The average of these two values ​​is then calculated. Again, if the calculated average current is negative, the controller will de-energize the fly capacitor(s) C, as described above. Fx Therefore, the definition of the switch states for charging and discharging needs to be reversed.

[0085] For a fixed frequency voltage mode controller, for each cycle, for a rising edge (i.e., between times T1 and T2 in the illustrated example), the current at the start of the rising edge (e.g., at time T1) is measured. The time required to reach time T2 is known - this is what the COMP voltage sets. The inductance of inductor L and the voltage across it can be used to calculate the final current at the end of the rising edge (e.g., at time T2). The average of these two values ​​is then calculated. If the calculated average current is negative, the controller reduces the fly capacitor(s) C, as described above. Fx The definitions of the switch states that charge and discharge L must be reversed. For a falling edge (i.e., between times T2 and T3 in the illustrated example), the current at the start of the falling edge (e.g., at time T2) is measured and the end current (e.g., at time T3) is estimated by knowing how much time is left in the fixed frequency cycle and the inductance of inductor L and the voltage across it. The average of these two values ​​is then calculated. Again, if the calculated average current is negative, the controller will increase the fly capacitor(s) C as described above. Fx Therefore, the definition of the switch states for charging and discharging needs to be reversed.

[0086] Second embodiment

[0087] As an alternative to injecting a subharmonic signal into the control loop feedback path of the multilevel converter, the same result can be achieved by sensing the current through inductor L in real time and adding additional switching cycles (note that the current flowing through inductor L can be sensed at multiple paths in the converter). All that is needed is to sense when the current through inductor L approaches or crosses zero (i.e., not significantly positive or negative) from negative to positive (positive transition) or positive to negative (negative transition) and then switch the M-level switch state selector 414 (see FIG. 5) to the fly capacitor C Fx The only thing that needs to be done is to reprogram the capacitor to alternate between charging and discharging states. In other words, additional switching cycles are added at near-zero current levels that result in the voltage on the fly capacitor moving towards the target value.

[0088] 9A is a graph 900 of current through inductor L of a multilevel converter as a function of time, illustrating the addition of switching cycles to improve light load recovery. A negative to positive transition, such as the inflection point of the example circle 902, indicates the start of charging current flow into inductor L. A positive to negative transition, such as the inflection point of the example circle 904, indicates the start of discharging current flow from inductor L.

[0089] In the "added cycle" embodiment, when the output load drops, an additional inductor charge or discharge cycle is added. For example, at time T1 in FIG. 9A, conditions are favorable for adding an additional cycle. Thus, at zero-crossing points, such as the zero-crossing points of the example circles 906 and 908, the state of the then-current charge cycle is reversed. For example, at zero-crossing point 906, when the current reverses, the switch state of the then-current discharge cycle is reversed to discharge or maintain the discharge of the fly capacitor. As another example, at zero-crossing point 908, the switch state of the then-current charge cycle is reversed to provide an additional discharge cycle. The number and interval of added charge / discharge cycles can be adjusted as needed to scale the degree of light-load recovery required for a particular state of the multilevel converter.

[0090] FIG. 9B is a graph of the voltage of a particular fly capacitor C of a three-level multilevel converter as a function of time. The target voltage of the fly capacitor C may be 5V, but at time T1 (corresponding to time T1 of FIG. 9A), the voltage across the fly capacitor C may be, for example, only 4.8V. The level selected for the converter at time T1 is set to discharge the inductor L, and the switch state is selected to charge the fly capacitor C. However, at zero crossing point 906, when the inductor current reverses, the switch state for the inductor discharge cycle at that point is reversed (flipped). This allows the fly capacitor C to continue charging with the reversed current flow while maintaining the output level. Note that when the switch state is set to level 3 or level 2b, the inductor L can be discharged, but only level 2b can charge the fly capacitor C during the inductor discharge cycle. In this example, the fly capacitor C is charged to about 4.9V between times T1 and T2, as shown by graph line 920.

[0091] At time T2 in Figure 9B, the voltage across the fly capacitor C is still below the 5V target, so at zero crossing 908, the switch state of the current inductor charging cycle is reversed to provide yet another capacitor charging cycle - note that while the switch state can be set to level 1 or level 2a to charge the inductor L, only level 2a can charge the fly capacitor C during the inductor charging cycle. In this example, the fly capacitor C charges to approximately 5V between times T2 and T3, as shown by graph line 920. Without an additional capacitor charging cycle, the charge on the fly capacitor C decreases, as shown by graph line 922.

[0092] The advantages of the “added cycle” embodiment are that it eliminates the need for a ripple generator circuit, it is easy to implement, there are no additional transition losses because switching occurs at zero load, and there is no need to worry about the fly capacitor C when sufficiently large inductor ripple is present. Fx In some embodiments, cycles may be added only when necessary. For example, the fly capacitor C Fx Additional cycles may be added only if the voltage is sufficiently unbalanced from the target voltage, thus reducing the capacitance of the capacitor C Fx is within an acceptable range, additional cycles are avoided. Thus, for example, at zero crossing point 910, additional discharge cycles may not be needed. As another example, the controller may decide to only perform a charge cycle and / or a discharge cycle if the cycle is long enough. In some scenarios, the charge / discharge cycle is very short, such as with a minimum off time, and the "extra cycle" process provides little benefit.

[0093] The “additional cycle” technique can be used in multilevel converters that include circuits for subharmonic signal generation. For example, FxFor a small fraction of the voltage of C, it may be useful to use the add cycle technique because it is faster and has less impact on the controller loop. Fx For larger excursions in the voltage, it may be more efficient to use a sub-harmonic signal generator. In some cases, it may be useful to use both the extra cycle technique and a sub-harmonic signal generator at the same time - the sub-harmonic signal can give a larger increase in current, and the extra cycle (zero crossing) technique can be used to ensure that all the current through the fly capacitor is used to charge or discharge.

[0094] Example of a subharmonic signal generating circuit

[0095] 10 is a schematic diagram of one embodiment of a subharmonic signal generating circuit 1000. The illustrated subharmonic signal generating circuit 1000 may be used, for example, at any of the insertion nodes A1, A2, B, or C of the control circuit shown in FIG. 5. A pair of resistors R1, R2 between current sources I1, I2 are connected to the input signal V IN Give positive and negative I×R voltage drops for V IN A time-varying output signal V having an average voltage equal to OUT In an exemplary embodiment, current sources I1, I2 may provide a current of approximately 10 μA, and resistors R1, R2 each have a value of approximately 1 KΩ.

[0096] A first MOSFET M1 is coupled in parallel with a resistor R1 and its gate is modulated by a subharmonic frequency (e.g., half the switching frequency of a switch in an associated converter cell). Similarly, a second MOSFET M2 is coupled in parallel with a resistor R2 and its gate is modulated by the subharmonic frequency. In the illustrated example, the subharmonic frequency is applied to the gates of MOSFETs M1 and M2 in the form of complementary square waves. Thus, MOSFET M1 in the on state experiences a positive offset voltage V IN +I×R1 to V OUTwhile the on-state MOSFET M2 couples to a negative offset voltage V IN -I×R2 to V OUT Therefore, the output voltage V OUT (1) V IN +I×R1 and (2)V IN -I × R2 and V IN is averaged to

[0097] 11 is a schematic diagram of one embodiment of a subharmonic PWM edge delay circuit 1100. IN is shown applied to a first input S1 of multiplexer 1102, a first input of OR gate 1104, an input of delay circuit 1106, and a clock input of a conventional SR flip-flop 1108. An output of delay circuit 1106 is coupled to a second input of OR gate 1104. An output of OR gate 1104 is coupled to a second input S2 of multiplexer 1102. A Q output of SR flip-flop 1108 is coupled to a select input C of multiplexer 1102. A Q-BAR output of SR flip-flop 1108 is coupled to a set input S of SR flip-flop 1108. An output D of multiplexer 1102 is coupled to an input signal PWM IN Edge-delayed version of PWM OUT It is.

[0098] In operation, PWM IN When the PWM IN (Triggered by the positive-going edge of the PWM transition.) IN Assume that the first (e.g., odd) positive pulse of the signal drives the Q output of the SR flip-flop 1108 low. Thus, the first PWM IN For a pulse, multiplexer 1102 selects the first input S1, and the output D of multiplexer 1102 is the unmodified positive PWM IN It becomes a pulse. PWM INA second (e.g., even) positive pulse of the signal toggles the SR flip-flop 1108, causing the Q output of the SR flip-flop 1108 to go high. Thus, the second PWM IN For a pulse, the multiplexer 1102 selects the second input S2, and the output D of the multiplexer 1102 is fed to the PWM IN A long stretched positive PWM by ORing with a delayed version of IN Therefore, the SR flip-flop 1108 is a PWM OUT As a signal, a normal PWM IN Either a positive pulse or a stretched positive pulse may be selected instead.

[0099] Additional Control and Operational Considerations

[0100] In power converters, especially multilevel power converters, power switches may be implemented using FETs, especially MOSFETs. For each power FET, a driver circuit is generally required. Furthermore, for some power FETs, a level shifter may be required to convert a ground-referenced low-voltage logic on / off signal from an analog or digital controller to a signal with the same voltage swing, but referenced to the source voltage of the power FET it is driving, in order to charge or discharge the gate of the power FET, thereby controlling the conductive or blocked state of the power FET. In some applications, the functions of the level shifter and the driver circuit may be combined into one circuit.

[0101] It may be desirable to provide additional control and operating circuitry (or one or more shutdown procedures) that enable reliable and efficient operation of a power converter utilizing a multilevel converter cell designed according to the present disclosure. For example, in a step-down power converter, the output voltage of the converter cell is lower than the input voltage of the converter cell. Shutting down or disabling a converter cell having a design inductance connected to its output while the output load current is non-zero (e.g., due to a fault event such as a short circuit) generally requires some means to discharge the inductor current. In some embodiments, a bypass switch may be connected in parallel with the design inductance connected to the output of the converter cell and controlled to open during normal operation and close when shutting down the converter cell or in the event of a fault event. Ideally, the bypass switch may be closed before disabling switching of the converter cell to prevent transient ringing and provide a safe discharge of the inductor current. In an alternative embodiment that uses MOSFETs for the converter's main switches, an intrinsic body diode connected between the body and drain terminals of each MOSFET may also discharge the inductor current. Details of these solutions, as well as alternative shutdown solutions, are taught in U.S. Pat. No. 10,686,367, entitled "Apparatus and Method for Efficient Shutdown of Adiabatic Charge Pumps," issued June 16, 2020, and assigned to the assignee of the present invention, the contents of which are incorporated by reference.

[0102] Another consideration when combining converter cells in parallel is to control multiple parallel power converters to avoid inrush currents (e.g., during the soft start period of the power converters) and / or overstress of switches when not all of the power converters are fully operational, such as during start-up or when a fault condition occurs. Conditional control may be achieved by using a node state detector coupled to a selected node in the parallel-connected power converters to monitor voltage and / or current. Such a node state detector may be configured in some embodiments to operate in parallel with an output state detector that measures the output voltage of the associated power converter during start-up. The node state detector ensures that the voltage across critical components (e.g., fly capacitors and / or switches) in the converter cell(s) of the power converter are within a desired range before allowing full power steady-state operation of the parallel power converters, otherwise preventing full power steady-state operation. The node state detector may be coupled to a master controller that controls one or more of the parallel power converters using one or more common control signals. To facilitate the master controller configuration, the parallel power converters may each report a power good signal (Pgood) when they are ready to leave the start-up phase for full power steady-state operation. The master controller may essentially "AND" all such Pgood signals together, possibly along with one or more status signals from other circuits, so that the master controller will not allow full power steady state operation of any parallel power converter unless all of the parallel power converters are ready for that state. Essentially, the Pgood signals from each parallel power converter are all tied together such that the parallel power converters do not transition out of the start-up phase until all Pgood signals indicate they are ready to transition to steady state operation. Additionally, the parallel power converters may transition from steady state operation to automatic restart or shutdown operation if a fault condition in one or more of the parallel power converters causes the Pgood signal to change.Details of these solutions, as well as alternative shutdown solutions, are taught in U.S. Pat. No. 10,992,226, entitled "Startup Detection for Parallel Power Converters," issued on April 27, 2021, and assigned to the assignee of the present invention, the contents of which are incorporated by reference.

[0103] Another consideration in operating a multilevel converter cell is to achieve (i.e., precharge) and maintain essentially perfectly proportionally balanced fly capacitor voltages so that all switches experience similar voltage stresses. This is because unbalanced fly capacitors can lead to switch (especially FET switches) failure due to exposure to high voltages. One solution to both precharging the capacitor voltages and operationally balancing the capacitor voltages in a multilevel DC-DC converter circuit is to provide parallel "shadow" circuits that conditionally couple the fly capacitor to a voltage source or other circuit to precharge the capacitor, or conditionally couple two or more fly capacitors together to transfer charge from a high voltage capacitor to a low voltage capacitor, or conditionally couple the fly capacitor to a voltage sink to discharge the capacitor, all under the control of real-time capacitor voltage measurements. Each parallel "shadow" circuit may include a switch and resistor coupled in parallel to a main switch that is part of the multilevel converter cell (in some cases, one switch-resistor pair may span two series-connected switches). This particular solution for precharging and / or charge balancing the fly capacitor is very fast, provides slow precharging of the fly capacitor during the precharge period, protects the switches from inrush current, and provides a stable voltage to the converter cell switches. Details of this solution, as well as alternative precharging and charge balancing solutions, are taught in U.S. Patent No. 10,720,843, entitled "Multi-Level DC-DC Converter with Lossy Voltage Balancing," issued July 21, 2020, and assigned to the assignee of the present invention, the contents of which are incorporated by reference.

[0104] Another solution for balancing the capacitor voltages of a multi-level DC-DC converter circuit is to provide a lossless voltage balancing solution that allows out-of-order state transitions of the multi-level DC-DC converter cells to occur during normal operation. The net effect of the out-of-order state transitions is to increase or decrease the voltage across a particular fly capacitor, thereby preventing voltage overstress on the main switch of the DC-DC converter. In some embodiments, restrictions are placed on the overall sequence of state transitions to reduce or avoid toggling of the transition states, thereby allowing each capacitor the opportunity to receive voltage steering as needed, rather than allowing one capacitor to be voltage balanced before the voltage balancing of another capacitor. Details of this solution and alternative charge balancing solutions are taught in U.S. Patent No. 10,770,974, entitled "Multi-Level DC-DC Converter with Lossless Voltage Balancing," issued September 8, 2020, and assigned to the assignee of the present invention, the contents of which are incorporated by reference.

[0105] An additional consideration for some embodiments is to enable operation of the multi-level converter cells such that voltages can be generated within a boundary zone between voltage levels. A "boundary zone" represents an output voltage that is unattainable for a conventional multi-level DC-DC converter circuit. To generate an output voltage within the boundary zone, some embodiments essentially alternate (toggle) between adjacent (or proximate) zones by setting the states of the converter cell switches to a boundary zone transition pattern. For example, a three-level DC-DC converter circuit may operate in zone 1 at a selected time and in adjacent zone 2 at a selected time. Thus, zones 1 and 2 are treated as a single "super zone." More generally, in some cases, it may be useful to create a super zone using non-adjacent zones or using two or more zones (adjacent and / or non-adjacent). Details of this solution are taught in U.S. Pat. No. 10,720,842, entitled "Multi-Level DC-DC Converter with Boundary Transition Control," issued July 21, 2020 and assigned to the assignee of the present invention, the contents of which are incorporated by reference.

[0106] Yet another consideration of some embodiments is the protection of the main power switches and other components in the power converter from stress conditions, particularly voltages that exceed the breakdown voltage of such switches (especially FET switches).One means for protecting a multi-level power converter uses at least one high voltage FET switch while allowing all or most of the other main power switches to be low voltage FET switches.

[0107] As will be apparent, the embodiments of the multilevel power converter described in this disclosure can be synergistically combined with the teachings of one or more of the additional control and operating circuits and methods described in this section.

[0108] method

[0109] Another aspect of the present invention includes a method for improving light load recovery in an M-level multilevel converter (where "light load" means zero or very low load, e.g., less than about 10% of the absolute value of the maximum current specification of a typical converter). For example, FIG. 12 is a process flow diagram 1200 illustrating a first method for improving light load recovery in an M-level multilevel converter. Starting with an M-level multilevel converter configured to convert an input voltage applied to an input terminal into an output voltage at an output terminal in response to an input signal from a controller defining a charge cycle and a discharge cycle of the multilevel converter (block 1202), the method includes: injecting a subharmonic signal into a signal path of the controller to induce a subharmonic ripple at an output terminal of the M-level multilevel converter (block 1204).

[0110] Additional aspects of the above method may include one or more of the following: the controller includes a comparator having an input coupled to the output terminal, and further including selectively injecting a sub-harmonic signal into a signal path between the output terminal and the comparator; the controller includes a comparator having a reference voltage input, and further including selectively injecting a sub-harmonic signal at the reference voltage input; the controller includes a compensation circuit, and further including selectively injecting a sub-harmonic signal into a signal path before and after the compensation circuit; the controller includes a pulse width modulation generator, and further including selectively injecting a sub-harmonic signal into a signal path after the pulse width modulation generator; further including selectively injecting a sub-harmonic signal when an average current through the output terminal is about 0 amps; further including selectively injecting a sub-harmonic signal when a light load is present at the output terminal; further including selectively injecting a sub-harmonic signal scaled as a function of the absolute value of the average current through the output terminal; the M-level multilevel converter has a switching frequency and is the subharmonic signal has a frequency approximately half a switching frequency of the M-level multilevel converter; the M-level multilevel converter has a switching frequency and the subharmonic signal has a frequency less than half a switching frequency of the M-level multilevel converter; the subharmonic signal enables charge balancing of at least one fly capacitor in the M-level multilevel converter; the subharmonic signal generates sufficient transient positive and negative currents to enable charge balancing of at least one fly capacitor in the M-level multilevel converter; the subharmonic signal is an AC waveform; the subharmonic signal is an AC waveform having an average voltage of zero; the subharmonic signal is injected by modifying timing of a digital waveform generated by a controller that controls operation of the M-level multilevel converter; and / or the input signal from the controller includes a pulse width modulated waveform, further comprising selectively injecting the subharmonic signal by modifying timing of the pulse width modulated waveform.

[0111] 13 is a process flow diagram 1300 illustrating a second method for improving light load recovery in an M-level multilevel converter. Starting with an M-level multilevel converter configured to convert an input voltage applied to an input terminal into an output voltage at an output terminal in response to an input signal from a controller defining a charge cycle and a discharge cycle of the multilevel converter (block 1302), the method includes: adding an additional charge cycle near or at a zero current crossing point after a discharge cycle (block 1304); and adding an additional discharge cycle near or at a zero current crossing point after a charge cycle (block 1306).

[0112] General Benefits and Advantages of Multilevel Power Converters

[0113] Embodiments of the present invention improve power density and / or power efficiency when incorporating circuits and circuit modules or blocks. As should be appreciated by those skilled in the art, system architectures can be beneficially impacted utilizing embodiments of the present invention in significant ways, including lower power and / or longer battery life. Thus, the present invention specifically embraces system level embodiments that are creatively enabled by inclusion in larger system designs and applications.

[0114] More specifically, multi-level power converters provide or enable a number of benefits and advantages, including:

[0115] - Suitability for applications where the input and / or output voltage may have a wide dynamic range (e.g., varying battery input voltage levels, varying output voltage);

[0116] - Improving the efficiency of run-time of devices that run on portable electrical energy sources (batteries, generators or fuel cells using liquid or gaseous fuels, solar cells, etc.);

[0117] - Thermal management, improving efficiency especially where efficiency is important to protect other components (display, surrounding ICs, etc.) from excess heat;

[0118] - Enabling design optimization of power efficiency, power density, and form factor of power converters - for example, the small size of multi-level power converters may allow for placing power converters closer to the load, thus improving efficiency and / or reducing the overall bill of materials.

[0119] - Ability to leverage the performance of smaller, lower voltage transistors;

[0120] - Suitability for applications where the power source may vary widely, such as batteries, other power converters, generators or fuel cells using liquid or gaseous fuels, solar cells, line voltage (AC), and DC voltage sources (e.g. USB, USB-C, Power over Ethernet, etc.);

[0121] - Suitability for applications where the loads can vary widely, such as general ICs (including microprocessors and memory ICs), electric motors and actuators, transducers, sensors, displays (e.g. all types of LCDs and LEDs);

[0122] - The ability to be implemented in several IC technologies (e.g., MOSFET, GaN, GaAs, and bulk silicon) and packaging technologies (e.g., flip chip, ball grid array, wafer level scale chip packages, wide fan-out packages, and embedded packages).

[0123] The benefits and advantages of multi-level power converters enable their use in a wide range of applications. For example, applications of multi-level power converters include portable and mobile computing and / or communication products and components (e.g., laptops, ultrabook computers, tablets, and mobile phones), displays (e.g., LCD, LED), wireless-based devices and systems (e.g., cellular systems, WiFi, Bluetooth, Zigbee, Z-Wave, GPS-based devices), wired network devices and systems, data centers (e.g., for battery backup systems and / or power conversion for processing systems and / or electronic / optical network systems), Internet of Things (IOT) devices (e.g., smart switches and lights, safety sensors, security cameras), home appliances and electronics (e.g., appliances with built-in wireless transceivers such as set-top boxes, battery-powered vacuum cleaners, washers, dryers, refrigerators, etc.), AC / DC power converters, electric vehicles of all kinds (e.g., for drivetrains, control systems, and / or infotainment systems), and other devices and systems that utilize portable power sources and / or require power conversion.

[0124] Wireless system usage includes wireless RF systems (including base stations, relay stations, and mobile transceivers) using a variety of technologies and protocols, including various types of Orthogonal Frequency Division Multiplexing ("OFDM"), Quadrature Amplitude Modulation ("QAM"), Code Division Multiple Access ("CDMA"), Time Division Multiple Access ("TDMA"), Wideband Code Division Multiple Access ("W-CDMA"), Global System for Mobile Communications ("GSM"), Long Term Evolution ("LTE"), 5G, and WiFi (e.g., 802.11a, b, g, ac, ax), as well as other wireless communication standards and protocols.

[0125] Manufacturing Techniques and Options

[0126] In various embodiments of a multilevel power converter, it may be beneficial to use certain types of capacitors, particularly fly capacitors. For example, such capacitors are generally useful for having low equivalent series resistance (ESR), low DC bias degradation, high capacitance, and small volume. Low ESR is particularly important for multilevel power converters that incorporate additional switches and fly capacitors to increase the number of voltage levels. The selection of a particular capacitor should be made after considering specifications such as power levels, efficiency, and size. Various types of capacitor technologies may be used, including ceramic (including multilayer ceramic capacitors), electrolytic capacitors, film capacitors (including power film capacitors), and IC-based capacitors. Capacitor dielectrics may be modified as needed for a particular application and may include dielectrics that are paraelectric, such as silicon dioxide (SiO2), hafnium dioxide (HFO2), or aluminum oxide Al2O3. Additionally, the design of the multilevel power converter may advantageously utilize inherent parasitic capacitances (e.g., inherent in power FETs) in combination with or in place of the designed capacitors to reduce circuit size and / or improve circuit performance. Capacitor selection for a multilevel power converter may also consider factors such as capacitor component variations, reduced effective capacitance with DC bias, and ceramic capacitor temperature coefficients (minimum and maximum temperature operating limits, and change in capacitance with temperature).

[0127] Similarly, in various embodiments of a multilevel power converter, it may be advantageous to use certain types of inductors. For example, it is generally useful for inductors to have low DC equivalent resistance, high inductance, and small volume.

[0128] The controller(s) used to control the start-up and operation of the multi-level power converter may be implemented as a microprocessor, a microcontroller, a digital signal processor (DSP), a register transfer level (RTL) circuit, and / or combinational logic.

[0129] Although some exemplary embodiments include a three-level converter, the present invention generally includes multi-level converters, including four-level and five-level converters.

[0130] The term "MOSFET" as used in this disclosure includes any field effect transistor (FET) having an insulated gate whose voltage determines the conductivity of the transistor, and encompasses insulated gates having metallic or metal-like, insulating, and / or semiconducting structures. The terms "metal" or "metal-like" include at least one conductive material (such as aluminum, copper, or other metals, or highly doped polysilicon, graphene, or other conductors), "insulator" includes at least one insulating material (such as silicon oxide or other dielectric material), and "semiconductor" includes at least one semiconducting material.

[0131] As used in this disclosure, the term "radio frequency" (RF) refers to frequencies ranging from about 3 kHz to about 300 GHz. This term also includes frequencies used in wireless communication systems. RF frequencies may be the frequencies of electromagnetic waves or the frequencies of alternating voltages or currents in circuits.

[0132] With respect to the figures referenced in this disclosure, dimensions of various elements are not to scale. Some dimensions have been significantly exaggerated vertically and / or horizontally for clarity or emphasis. Additionally, references to orientations and directions (e.g., "top", "bottom", "up", "down", "lateral", "vertical", "horizontal", etc.) are with respect to the illustrative drawings and are not necessarily absolute orientations or directions.

[0133] Various embodiments of the present invention may be implemented to meet a wide variety of specifications. Unless otherwise noted above, the selection of appropriate component values ​​is a matter of design choice. Various embodiments of the present invention may be implemented in any suitable integrated circuit (IC) technology (including, but not limited to, MOSFET structures) or in hybrid or discrete circuit forms. Integrated circuit embodiments may be fabricated using any suitable substrate and process, including, but not limited to, standard bulk silicon, high resistivity bulk CMOS, silicon-on-insulator (SOI), and silicon-on-sapphire (SOS). Unless otherwise noted above, embodiments of the present invention may be implemented in other transistor technologies, such as bipolar, BiCMOS, LDMOS, BCD, GaAs HBT, GaN HEMT, GaAs pHEMT, and MESFET technologies. However, embodiments of the present invention are particularly useful when fabricated using SOI or SOS based processes, or processes having similar characteristics. Fabrication in CMOS using SOI or SOS processes allows circuits with low power consumption, the ability to withstand high power signals during operation due to FET stacking, good linearity, and high frequency operation (radio frequencies up to and beyond 300 GHz gigahertz). Monolithic IC implementations are particularly useful because careful design can keep parasitic capacitances generally low (or at the very least, uniform across all units, allowing them to be compensated for).

[0134] Voltage levels can be adjusted and / or polarity of voltages and / or logic signals can be inverted depending on the particular specifications and / or implementation technology (e.g., NMOS, PMOS, or CMOS, and enhancement or depletion mode transistor devices). The voltage, current, and power handling capabilities of components can be adapted as needed, for example, by adjusting device sizes, "stacking" components (especially FETs) in series to withstand higher voltages, and / or using multiple components in parallel to handle higher currents. Additional circuit components can be added to enhance the capabilities of the disclosed circuits and / or to provide additional functionality without significantly changing the functionality of the disclosed circuits.

[0135] Circuits and devices according to the invention can be used alone or in combination with other components, circuits, and devices. Embodiments of the invention can be fabricated as integrated circuits (ICs), which can be housed in IC packages and / or modules for improved handling, manufacturing, and / or performance. In particular, IC embodiments of the invention are often used in modules where one or more of such ICs are combined with other circuit components or blocks (e.g., filters, amplifiers, passive components, and possibly additional ICs) into a package. The ICs and / or modules are then typically combined with other components on a printed circuit board to form part of an end product, such as a mobile phone, laptop computer, or electronic tablet, or to form higher level modules that can be used in a wide range of products, such as vehicles, test equipment, medical devices, etc. Through various configurations of modules and assemblies, such ICs typically enable a mode of communication, often wireless communication.

[0136] conclusion

[0137] Several embodiments of the present invention have been described. It should be understood that various modifications may be made without departing from the spirit and scope of the present invention. For example, some of the steps described above may be order independent and therefore may be performed in an order different from that described. Furthermore, some of the steps described above may be optional. Various activities described with respect to the methods described above may be performed iteratively, serially, and / or in parallel.

[0138] It is to be understood that the above description is intended to illustrate, but not to limit, the scope of the invention, which is defined by the scope of the following claims, and that other embodiments are within the scope of the claims. In particular, the scope of the invention includes all feasible combinations of one or more of the processes, machines, manufacture, or compositions of matter set forth in the following claims. (Note that parenthetical labels of claim elements are for ease of reference to such elements and do not, in themselves, indicate a particular required order or enumeration of the elements. Moreover, such labels may be reused in dependent claims as references to additional elements without being considered the start of a conflicting label sequence.)

Claims

1. A multi-level converter, (a) an M-level converter cell configured to convert an input voltage applied to an input terminal into an output voltage at an output terminal in response to a control input; (b) a controller coupled to the M-level converter cell, the controller configured to monitor a node of the M-level converter cell and generate the control input to the M-level converter cell in response to the monitoring; and (c) a subharmonic signal generator coupled to the controller and configured to selectively inject a subharmonic signal into a signal path of the controller; Multilevel converter.

2. the sub-harmonic signal induces a sub-harmonic ripple at the output terminal of the M-level converter cell; 2. The multilevel converter of claim 1.

3. the M-level converter cell includes at least one fly capacitor, and the sub-harmonic signal induces a sub-harmonic ripple at the output terminal of the M-level converter cell to facilitate balancing of charge across the at least one fly capacitor.

2. The multilevel converter of claim 1.

4. the controller includes a comparator having an input coupled to the output terminal of the M-level converter cell, and the sub-harmonic signal generator is configured to selectively inject the sub-harmonic signal into the signal path between the output terminal and the comparator.

2. The multilevel converter of claim 1.

5. the controller includes a comparator having a reference signal input, the subharmonic signal generator being configured to selectively inject the subharmonic signal at the reference signal input; 2. The multilevel converter of claim 1.

6. the controller includes a compensation circuit, and the subharmonic signal generator is configured to selectively inject the subharmonic signal into the signal path after the compensation circuit.

2. The multilevel converter of claim 1.

7. the controller includes a compensation circuit, and the subharmonic signal generator is configured to selectively inject the subharmonic signal into the signal path prior to the compensation circuit.

2. The multilevel converter of claim 1.

8. the controller includes a pulse width modulation generator, and the subharmonic signal generator is configured to selectively inject the subharmonic signal into the signal path after the pulse width modulation generator.

2. The multilevel converter of claim 1.

9. the subharmonic signal generator is configured to inject the subharmonic signal when an average current through the output terminal is about 0 amperes; 2. The multilevel converter of claim 1.

10. the subharmonic signal generator is configured to inject the subharmonic signal when a light load is present at the output terminal; 2. The multilevel converter of claim 1.

11. (a) An M-level converter cell, comprising: (1) an input terminal; (2) a set of switches coupled in series between a voltage source and a reference voltage; (3) an output terminal coupled to an innermost pair of the set of switches and configured to be coupled to an inductor; (4) a switch control input coupled to the set of switches; (5) at least one fly capacitor coupled in series with each particular switch and in parallel with a switch located between said respective switches; Including, the M-level converter cell is configured to convert an input voltage applied to the input terminal into an output voltage at the output terminal in response to a control signal at the switch control input; M level converter cell; (b) a controller coupled to the inductor and the switch control input, the controller configured to monitor an output of the M-level converter cell and, in response to the monitoring, dynamically generate a set of switch control input values ​​for the M-level converter cell; and (c) a subharmonic signal generator coupled to the controller and configured to selectively inject a subharmonic signal into a signal path of the controller; Multilevel converter.

12. the controller includes a comparator having an input coupled to the inductor, the subharmonic signal generator being configured to selectively inject the subharmonic signal into the signal path between the inductor and the comparator.

12. The multilevel converter of claim 11.

13. the controller includes a comparator having a reference voltage input, the subharmonic signal generator being configured to selectively inject the subharmonic signal at the reference voltage input; 12. The multilevel converter of claim 11.

14. the controller includes a compensation circuit, and the subharmonic signal generator is configured to selectively inject the subharmonic signal into the signal path after the compensation circuit.

12. The multilevel converter of claim 11.

15. the controller includes a compensation circuit, and the subharmonic signal generator is configured to selectively inject the subharmonic signal into the signal path prior to the compensation circuit.

12. The multilevel converter of claim 11.

16. the controller includes a pulse width modulation generator, and the subharmonic signal generator is configured to selectively inject the subharmonic signal into the signal path after the pulse width modulation generator.

12. The multilevel converter of claim 11.

17. the subharmonic signal generator is configured to inject the subharmonic signal when an average current through the inductor is about 0 amperes.

12. The multilevel converter of claim 11.

18. the subharmonic signal generator is configured to inject the subharmonic signal when a light load is present at the output terminal.

12. The multilevel converter of claim 11.

19. the subharmonic signal generator is configured to inject a scaled subharmonic signal as a function of an absolute value of an average current through the inductor; 12. The multilevel converter of claim 11.

20. the set of switches has a switching frequency, and the subharmonic signal has a frequency that is approximately half the switching frequency of the set of switches.

12. The multilevel converter of claim 11.