Multi-mode switching control circuit and method for improving light load efficiency in switching power supplies

[0001] This application is based on and claims priority to U.S. Provisional Patent Application Ser. No. 60/797,766, filed on May 4, 2006 and entitled METHOD AND APPARATUS FOR IMPROVING LIGHT LOAD EFFICIENCY IN SWITCHING POWER SUPPLIES, the entire contents of which are incorporated by reference herein.

[0002] The present invention relates to light-load efficiency of power converters, and more particularly to smooth transition between heavy- and light-loads of power converters.

[0003] Light-load efficiency of power converters is becoming increasingly important for switch mode power supply (SMPS) powered devices. Improved light-load efficiency helps save energy and extend battery life of the SMPS powered devices. Conventional methods for improving the light-load efficiency of the power converters include: reducing the circulation energy and associated losses by running a power converter's circuit in a discontinuous conduction mode (DCM); reducing switching losses by reducing switching frequency at a light load; and shutting down unnecessary phase(s) at a light load of a multi-phase converter. When the load current is high, it is preferable to operate a power converter in a fixed frequency continuous conduction mode (CCM), which allows a fast transient response, higher efficiency, and a narrower-spread noise spectrum. The key to achieving the best overall performance is knowing when and how to transition between the two modes.

[0004] Currently existing solutions for improving light-load efficiency of power converters include a constant on-time with valley-mode control, a hysteresis control, and a duty cycle control with pulse skipping/burst mode at light load methods. In the constant on-time with valley-mode control method, the control switch is turned on with a fixed period of time. The off-time is determined by a time at which the output voltage reaches or hits the valley threshold of the comparator. The disadvantages of this control are: variable frequency control for all input and load condition; instead of the average output voltage, the minimum output voltage is regulated to reference; the control is affected by the comparator delay; reliance on a reasonably large equivalent series resistance (ESR) of the output capacitor to generate the output voltage ramp for off-time comparison; and a need of a high-gain, high-bandwidth, low-offset differential amplifier to sense the output voltage ramp.

[0005] The hysteresis control method is similar to the constant on-time control, having the same advantages and disadvantages. However, both on-time and off-time of the control switch are controlled by the hysteresis of the comparator.

[0006] In the duty cycle control with pulse skipping/burst mode at light load method, a fixed-frequency duty cycle control is used when the load current is high. At light load, the control switch is turned ON only when the output voltage is less than the threshold and within the hysteresis of a comparator. This results in the control switch turning on for a few consecutive cycles and then keeping silent until the output voltage drops below the threshold. This control is not desirable because a burst mode at light load can cause the switching noise spectrum spread to a wide range, imposing EMI issues.

[0007] It is clear that the existing control methods either have variable frequency control at all time or have an abrupt transition between fixed and variable frequency when transitioning between heavy load and light load condition. The ideal solution is to have constant-frequency duty cycle control at heavy load condition, while using constant on-time variable frequency control at light load condition through a smooth transition. As a result, both the efficiency and switching noise concerns can be taken care of.

[0008] It is an object of the present invention to provide best overall performance at all load and input conditions.

[0009] It is another object of the present invention to provide constant-frequency duty cycle control at heavy load and constant on-time variable frequency control at light load.

[0010] It is yet another object of the present invention to provide a natural and smooth transition between heavy and light loads.

[0011] A circuit is provided for transitioning between a discontinuous and a fixed frequency continuous conduction mode (DCM) and (CCM) of a power converter having a driver receiving PWM signals and controlling a switching stage comprises a control switch and a sync switch connected at a common switching node for driving a load. The circuit including a PWM modulator for providing PWM signals; and a mode selector for receiving a duty cycle value, a preset switching period, and a duty cycle at a critical conduction point having a switching frequency equal to the preset switching period and providing an on-time of the control switch and a switching period to the PWM modulator, wherein if the duty cycle value is greater than the duty cycle at a critical conduction point, the PWM modulator will drive will provide the PWM signals to operate the switching stage in the CCM with constant-frequency duty cycle control, and if the duty cycle value is less than duty cycle at a critical conduction point, the PWM modulator will drive will provide the PWM signals to operate the switching stage in the DCM by turning off the sync switch when a load current becomes negative.

[0012] Other features and advantages of the present invention will become apparent from the following description of the invention that refers to the accompanying drawings.

[0017] In FIG. 1 a synchronous buck converter is used for illustrative purposes. However, the same principle may be applied to other topologies, as well. In FIG. 1 circuit 10 includes a driver 14 connected to gate terminals of a half-bridge stage 16 having transistors S1 and S2 connected at a common node 18. Second terminals of the transistors S1 and S2 being connected to a source 12. The driver 14 controlling the half-bridge stage 16 in accordance with PWM signals received from a PWM modulator 20. The PWM signals are based on a computed duty cycle value D provided by a compensator 22.

[0018] A load L is connected to the common node 18 and series connected to a resistor R1. The resistor R1 is connected at a second node to a resistor R_C that is series coupled to a capacitor C and to a resistor R. An amplifier 24 receives its positive input from the connection between the load L and the resistor R1 and the negative input from the second node. The compensator 22 receives a predefined reference voltage V_REF, a load current I_L from the amplifier 24, and a voltage V_O from the second node.

[0019] The value D is defined as:

[0020] $D=\frac{T_{\mathrm{ON}}}{T_S},$
where T_ON is the on-time of the control switch S1 and T_S is a switching period. In this conventional control scheme with a single compensator, either the on-time T_ON of the control switch S1 (constant on-time) or T_S (duty cycle control) are fixed. That is why a smooth transition between the two control modes is difficult.

[0021] As illustrated in FIG. 2, a circuit 30 includes a mode selector 32. The mode selector and T_ON controller 32 receives the duty cycle value D from the compensator 22. If D>D_MIN, where D_MIN is the duty cycle at the critical conduction point with the switching frequency equals to a preset switching period T_SET, the mode selector controller block 32 will output T_S=T_SET and T_ON=D×T_SET. The power stage circuit 30 will operate in the CCM with constant-frequency duty cycle control.

[0022] Alternatively, if the D<D_MIN, the mode selector controller block 32 will turn off the sync switch when the inductor current becomes negative and push the power stage into the DCM. In this mode, the output T_ON=D_MIN×T_SET. The output T_S can then be computed using

[0023] $T_S=\frac{D_{\mathrm{MIN}}·T_{\mathrm{SET}}}{D}.$
This way, the mode selector T_ON controller modulates the switching frequency at light load, instead of the duty cycle, to regulate the output voltage.

[0024] Because the power stage is reduced to a single-pole system when it is running in the DCM, the same compensator used for CCM can be directly used for DCM as well without raising issues of stability. Hence, no extra circuit complexities are required for the compensator design. It is clear that the transition between the two modes is smooth because of the nature of the power stage circuit.

[0025] The modified PWM modulator 34 requires an ability to change both, the on-time and the switching period. But it only needs to change one of them at any specific switching cycle.

[0026] The control proposed with reference to FIG. 2, can be implemented by an analog control, which, from a practical point of view, is complicated. Employing a digital control in the form of a microcontroller, microprocessor, DSP, or other digital implementation (“processor”), combined with a mixed signal input acquisition circuitry, it is possible to achieve a practical implementation of the power converter having the control structure of an embodiment of the present invention.

[0027] FIG. 3 shows an implementation of the mode selector T_ON controller 32 using basic logic and arithmetic cell units. The minimum duty cycle D_MIN is approximated by an arithmetic unit 40 that accepts a reference voltage V_REF and an input voltage V_in and calculates D_MIN=V_REF/V_in. Prior to being accepted in the arithmetic unit 40, the input voltage V_in is processed in an A/D converter 38. A more accurate result may be obtained if the load current and parasitic circuit parameters are taken into consideration.

[0028] A logic unit 42 compares the input duty cycle D with the minimum duty cycle D_MIN. An arithmetic unit 44 multiplies the minimum duty cycle D_MIN with the preset switching period T_SET and divide the product by the input duty cycle D.

[0029] The result of the logic unit 42 is used by two multiplexers 46 and 48 to control the on-time T_ON of the control switch S1 and the switching period T_S outputs. The multiplexer 46 provides a value selected between the minimum duty cycle D_MIN and the input duty cycle D to an arithmetic unit 49, which multiplies that selected value with the preset switching period T_SET to output the on-time T_ON of the control switch S1.

[0030] The multiplexer 48 simply selects the switching period T_S output between the preset switching period T_SET and the product of the arithmetic unit 44.

[0031] FIG. 4 shows a digital hybrid PWM implementation of the PWM multiplexer 34 with T_ON and T_S control. In this circuit 34, a high-bit free-running counter is replaced by a counter CT₁ 58 having enable and clear controls. The counter CT₁ 58 counting a clock signal clk1 to produce an output signal Q_CT1.

[0032] A comparator CP₁ 50 compares the output signal Q_CT1 to the switching period T_S; a comparator CP₂ 52 compares the output signal Q_CT1 to 0; and a comparator CP₃ 54 compares the output signal Q_CT1 to T_ON[m+1 . . . n]. At the beginning, the signal Q_CT1 is zero, so a comparator CP₂ 52 outputs 1 to set the PWM latch SR₁ 62. This sets a PWM output at high. The counter CT₁ 58 starts to count when the signal clk1 is received.

[0033] When the signal Q_CT1 reaches T_ON[m+1 . . . n], this is detected by the comparator CP₃ 54, which issues an enabling signal to a comparator CP₄ 56 and a NAND circuit 64. The NAND circuit 64 performs a nand operation on the enabling and PWM signals to provide an sclr signal to the counter CT₂ 60.

[0034] When the low-bit counter CT₂ 60 is enabled and starts to count the high frequency clk2 signals to produce a Q_CT2 signal, the comparator CP₄ 56 compares the Q_CT2 signal to the T_ON[0 . . . m] signals. When the Q_CT2 signal equals to T_ON[0 . . . m], the PWM latch SR₁ 62 is reset and PWM returns to a low state. This implements the T_ON control.

[0035] The high-bit counter CT₁ 58 continues to count until the output Q_CT1 equals T_S, at which point the counter CT₁ 58 is reset to set the PWM latch SR₁ 62 and bring the output PWM to a high state. This implements switching period T_S control.

[0036] Although the present invention has been described in relation to particular embodiments thereof, many other variations and modifications and other uses will become apparent to those skilled in the art. It is preferred, therefore, that the present invention not be limited by the specific disclosure herein.