Soft-switching for high-frequency power conversion

Abstract

A power converter designed for operation at high frequencies includes a soft-switching cell comprising a split inductor, a resonant inductor, a resonant capacitor, two diodes and a controlled semiconductor. Alternatively, the power converter includes a soft-switching cell comprising a transformer having isolated windings, a resonant inductor, a resonant capacitor, two diodes and a controlled semiconductor.

Claims

1. A power converter designed for operation at high frequencies including a soft-switching cell comprising a split inductor, a resonant inductor, a resonant capacitor, two diodes and a controlled semiconductor.

2. The power converter of claim 1, wherein the resonant inductor and the resonant capacitor form a resonant tank connected to an active circuit that is actuated just prior to the turn-on transition of the main switching device in order to event triggers a half-wavelength resonant voltage transition on the resonant capacitor that causes its voltage to transition from a non-zero value to virtually zero.

3. The power converter of claim 1, wherein said two diodes include a first clamp diode is connected from ground to one terminal of the split inductor, and a second clamp diode is connected from another terminal of the split inductor to a known DC voltage bus.

4. The power converter of claim 3, wherein said second clamp diode that is connected from a terminal of the split inductor to a known DC voltage bus is replaced by a zener diode in order to facilitate the demagnetization of said split inductor.

5. The power converter of claim 1 including a lossless capacitive snubber connecting the controlled semiconductor to the resonant capacitor.

6. The power converter of claim 5, wherein said lossless snubber includes a diode and a capacitor.

7. A power converter designed for operation at high frequencies including a soft-switching cell comprising a transformer having isolated windings, a resonant inductor, a resonant capacitor, two diodes and a controlled semiconductor.

8. The power converter of claim 7, wherein the resonant inductor and the resonant capacitor form a resonant tank connected to an active circuit that is actuated just prior to the turn-on transition of the main switching device in order to event triggers a half-wavelength resonant voltage transition on the resonant capacitor that causes its voltage to transition from a non-zero value to virtually zero.

9. The power converter of claim 7, wherein said two diodes include a first clamp diode is connected from ground to one terminal of said transformer and a second clamp diode is connected from another terminal of said transformer to a known DC voltage bus.

10. The power converter of claim 9, wherein said second clamp diode that is connected from a terminal of said transformer to a known DC voltage bus is replaced by a zener diode in order to facilitate the demagnetization of said transformer.

11. The power converter of claim 7 including a lossless capacitive snubber connecting the controlled semiconductor to the resonant capacitor.

12. The power converter of claim 11, wherein said lossless snubber includes a diode and a capacitor.

13. A method for controlling operation of a power converter designed for operation at high frequencies includes: providing a soft-switching cell comprising a transformer with isolated windings, a resonant inductor, a resonant capacitor, two diodes, and an Q.sub.aux controlled semiconductor; and producing a first drive signal for the an Q.sub.aux controlled semiconductor and a second drive signal for the an Q.sub.main controlled semiconductor.

14. The method of claim 13 including terminating said first drive signal upon sensing a voltage transition at any of the windings of the transformer.

15. The method of claim 13 including terminating said first drive signal after a time that is calculated or estimated using known circuit values and parameters.

16. The method of claim 13 including initiating said second drive signal upon sensing a transition on the voltage of the resonant capacitor.

17. The method of claim 16 including terminating said second drive signal by a PWM controller.

18. The method of claim 16 including terminating said second drive signal by a peak-current mode controller.

19. The method of claim 13 including initiating said second drive signal upon comparison of the voltage of the resonant capacitor to a set minimum threshold.

20. The method of claim 19 including terminating said second drive signal by a PWM controller.

21. The method of claim 19 including terminating said second drive signal by a peak-current mode controller.

22. The method of claim 11 including initiating said second drive signal upon detection of a decrease of the rate of change of the voltage of the resonant capacitor.

23. The method of claim 22 including terminating said second drive signal by a PWM controller.

24. The method of claim 22 including terminating said second drive signal by a peak-current mode controller.

25. The method of claim 13 including modulating the operational frequency in response to the feedback control loop error signal.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0035] FIG. 1 illustrates schematically a boost converter with a soft-switching cell (SSC) or module.

[0036] FIG. 1A illustrates separately the soft-switching cell used in the boost converter shown in FIG. 1.

[0037] FIG. 2 illustrates schematically a buck converter with a soft-switching cell.

[0038] FIG. 3 illustrates schematically a buck-boost converter with a soft-switching cell.

[0039] FIG. 4 illustrates schematically a Cuk converter with a soft-switching cell.

[0040] FIG. 5 illustrates schematically a SEPIC converter with a soft-switching cell.

[0041] FIG. 6 illustrates waveforms for the boost converter with the soft-switching cell shown in FIGS. 1 and 1A.

[0042] FIG. 7 illustrates schematically a boost converter with a soft-switching cell featuring Q.sub.aux turn-off snubber and demagnetization.

[0043] FIG. 8 illustrates schematically a SEPIC converter with a soft-switching cell with output not referenced to ground.

[0044] FIG. 9 illustrates schematically a current-fed push-pull converter with a soft-switching cell.

[0045] FIG. 10 illustrates controlling the boost converter with the soft-switching cell as schematically shown in FIG. 7

DETAILED DESCRIPTION OF ILLUSTRATED EMBODIMENTS

[0046] FIGS. 1, 2, 3, 4, and 5 illustrates schematically embodiments of different power conversion topologies modified with a soft-switching cell (SSC) shown in FIG. 1A. The soft-switching cell includes a split inductor, a resonant inductor, a resonant capacitor, two diodes, and a controlled semiconductor. The soft-switching cell alternatively includes a transformer having isolated windings, a resonant inductor, a resonant capacitor, two diodes and a controlled semiconductor. In order to illustrate its operation, FIG. 6 shows the relevant signal waveforms relative to the boost converter shown in FIG. 1. Referring to FIG. 6, starting at a time t.sub.0 when Q.sub.main is not conducting, D.sub.boost is conducting, Lr carries no current and Cr is charged at a voltage equal to Vo, the output voltage. At time t.sub.1, the control logic begins the turn-on transition by driving Q.sub.aux to its conductive state. Resonant inductor Lr begins to conduct current that impinges upon the center tap of the split inductor T; this current separates equally between the two windings of T due to transformer action and a nominal 1:1 turns ratio of the two windings. As both Q.sub.aux and D2 conduct, voltage across split inductor T is approximately equal to the output voltage Vo. Again due to transformer action, the split inductor divides this voltage equally across the two winding. Therefore the voltage at node 1 is approximately one half of Vo. The voltage at node 2, on the other hand is clamped at the initial voltage of Vo given the conducting state of D.sub.boost. During this time interval, the voltage across inductor Lr is constant and equal to one half of Vo, causing the current in Lr to increase linearly.

[0047] At time t.sub.2, the current in Lr reaches the same amplitude as the current through the boost inductor L.sub.boost. As all the input current has been diverted from it, boost diode D.sub.boost turns off, no longer clamping the voltage at node 2, thus allowing Cr to begin resonating with Lr. This resonance is centered about the voltage at node 1, or one half of Vo; therefore, at node 2 the initial voltage Vo will sinusoidally discharge to zero volts without attempting to assume negative values.

[0048] At time t.sub.3, at the end of the discharge time of Cr, the control logic drives Q.sub.main to its conductive state; as the voltage across Q.sub.main is approximately zero at this time, this turn-on transition does not entail any significant energy loss. Now that Q.sub.main is conducting, the voltage across Cr is prevented from oscillating back and is rather clamped at approximately zero volts. The current in Lr now decreases linearly until it becomes equal to twice the magnetizing current of the split inductor T, at time t.sub.4. At this time t.sub.4, diode D2 stops conducting and the voltage at node 3 begins to decrease quickly; simultaneously, the control logic drives Q.sub.aux into its off-state. It is noteworthy that the turn-off current level of Q.sub.aux is limited to a modest value of only twice the magnetizing current of T; thus the turn-off energy loss, including current-tail effects, is very small for Q.sub.aux. With these conditions, transformer T undergoes a classic flyback transition whereby the voltage across it quickly reverses; clamp diodes D3 and D4 limit this excursion to a maximum of twice the output voltage Vo. This negative voltage quickly demagnetizes transformer T until the transformer core is magnetically reset, at time t.sub.4. The transformer voltage now quickly falls to zero volts; at this time the whole boost current flows through Q.sub.main, while Cr and Lr are both discharged. This situation continues for the remainder of the on-time of Q.sub.main, until the control logic determines that Q.sub.main should turn off, at time t5. The turn-off transition of Q.sub.main is very smooth due to the presence of the initially discharged Cr, which now acts as a very effective turn-off snubber, limiting the collector voltage rate of change to very low values. This slow rate of change of the collector voltage reduces EMI emissions and decreases the turn-off energy loss by a significant factor. Once the collector voltage reaches the output voltage value Vo, the boost diode D.sub.boost begins to conduct the boost current. At this stage the circuit parameters are identical to the initial parameters, as the switching cycle is ready to repeat all the transitions described above.

[0049] FIG. 7 illustrates a more sophisticated embodiment of the boost converter shown in FIG. 1. This embodiment takes advantage of the fact that Q.sub.aux turns off while Qmain is conducting and resonant cap Cr is discharged. Therefore, a turn-off diode-capacitor snubber (D5-Cs) can be added to the collector of Q.sub.aux in order to decrease the rate of change of voltage during the turn-off transition; this further decreases switching loss and improves EMI emissions.

[0050] FIG. 7 also shows the addition of power zener Dz in the clamping path. In normal operation this zener simply adds some voltage to the clamp voltage of Q.sub.aux and otherwise does not affect operation as described above. However, its presence becomes critical if the conduction time of Q.sub.main is allowed to approach smaller durations so that Q.sub.main turns off occurs while the magnetic reset of transformer T is not yet completed. In this case, voltages at node 2 and node 1 increase, drastically reducing the reset voltage of T. Consequently, in the absence of Dz, the reset time would become unpractically long, leading to possibly destructive saturation of transformer T. The presence of Dz, on the other hand, introduces a minimal reset voltage that forces demagnetization under all conditions. In practical circuits, the zener voltage of Dz can be relatively small so that the added voltage stress on Q.sub.aux can be limited to an extra 5% or less.

[0051] Other embodiments of the present invention are applications to basic conversion topologies other than the boost topologies shown in FIGS. 2, 3, 4, 5. In such embodiments the basic operation of the soft-switching cell (SSC) is very similar to the one described in detail above. An important element for the proper functioning of the SSC in any converter topology is the presence of a stable DC voltage bus that has a known correlation to the main IGBT's off-time voltage. For instance, in the case of the boost the output voltage represents this stable DC bus; its correlation to the main IGBT's off-voltage is simply 1 to 1. In the case of the buck converter, the stable DC bus is the input voltage Vin; here too the correlation to the IGBT voltage is 1:1. In the embodiments shown in FIGS. 3, 4, and 5, using the buckboost converter, the Cuk converter, and the SEPIC converter, respectively, the stable bus is provided by the sum of the input and output voltages.

[0052] In other embodiments, the needed stable DC bus may be referenced to a different voltage with respect to the common terminal of the soft-switching cell. In this case, the split inductor may be rewired as a transformer with two isolated windings. As example of such embodiments is the SEPIC converter, shown in FIG. 8. In this embodiment, the stable bus is across the series combination of Co and Cs, which is not referenced to ground. In this case, the whole of one winding of T is across the DC bus; because the intent is to produce a voltage equal to one half of this voltage, the transformer turns ratio will be 2:1. FIG. 8 does not show any reset mechanism for transformer T, as this function can be performed using several standard techniques, such as the use an additional reset winding.

[0053] The disclosed technique can be applied to many isolated topologies such as the forward converter or the flyback. For example, FIG. 9 shows the SSC applied to a current-fed push-pull topology (that is, isolated boost). In this embodiment, the reset winding is shown connected to the output voltage through D2B.

[0054] The control method of the soft-switching cell aims at generating the drive signals to both Q.sub.main and Q.sub.aux. In its basic functionality, the control signal for Q.sub.aux requires only a trigger signal from the controller, designating the start of the Q.sub.aux drive pulse (t.sub.0 in FIG. 6). In contrast, the duration of the Q.sub.aux drive signal is solely determined by sensing or predicting the evolution of critical waveforms in the power circuit.

[0055] In one embodiment of the present invention, a voltage sensor may be placed across transformer T or at node 3 (FIG. 1); the drive signal can be terminated when this voltage is sensed to be decreasing rapidly, indicating the onset of the transformer reset, at time t.sub.3.

[0056] In another embodiment, the duration of the Q.sub.aux drive signal may be calculated or otherwise estimated using known information including the output and input voltage amplitude, the switching frequency, the value of the resonant components Cr and Lr, the value of the main inductor (for instance L.sub.boost) and the amplitude of the current it conducts.

[0057] For Q.sub.main the drive pulse begins at a time when the resonant transition of the collector voltage of Q.sub.main has reached its minimum value (t.sub.2 in FIG. 1). Again, in some embodiments this time can be assessed by sensing the instantaneous amplitude of the voltage on the resonant capacitor Cr or its rate of change. Alternatively, it can be calculated or otherwise estimated using known information including the output and input voltage amplitude, the switching frequency, the value of the resonant components Cr and Lr, the value of the main inductor (for instance L.sub.boost) and the amplitude of the current it conducts.

[0058] The drive pulse for Q.sub.main is terminated as determined by the controller, which may utilize pulse width modulation (PWM) or frequency modulation (FM) or a combination of the two. To this point, both FM and PWM are effective control methods when the converter operates in discontinuous conduction mode (DCM), whereas only PWM is properly practical in continuous conduction (CCM). Therefore, PWM is the nominal control method, but FM may also be added in order to simplify the control, especially at light loads, when the converter naturally enters DCM operation. The PWM signal may be generated by means of any of the well-known conventional techniques, including average mode current control, peak mode current control, voltage mode control with or without feedforward, charge mode control and more. Similarly, the FM signal may be generated by a conventional voltage-controlled oscillator or frequency generator.

[0059] FIG. 10 illustrates comprehensively the control for the soft-switching cell. The output voltage is sensed and compared to a reference, generating a frequency-compensated error signal. The error signal is then fed to both a peak-current mode controller and a frequency generator in a way that an increasing error signal produces a decreasing peak current reference as well as a decreasing frequency. This means that in high load conditions, the frequency of operation will also be higher. Operating at high frequency with higher load conditions allows the designer to select a boost inductor that partly saturates with increasing DC bias without suffering increasing ripple current. Such inductor benefits from considerable reduction in energy loss, size and cost. Operating at low frequency with lower loads also affords a substantial benefit that relates to an intrinsic feature of the soft-switching cell: the fact that Q.sub.aux is always driven on and off at the start of every cycle. This operation entails that a minimum amount of energy is always transferred to the output, even when O.sub.main is continuously in the off-state. Consequently, further controlled reduction of the output power can only be achieved by decreasing the frequency of operation.

[0060] We note separately that in the circuits, where the rate of change of rectifier current is controlled by a snubber inductor connected in series with the boost switch and the rectifier, due to the placement of the inductor, the voltage stress of the main switch is higher. This increased voltage stress can be minimized by a proper selection of the snubber-inductance value and the switching frequency. Both the boost and the auxiliary switches in these circuits, as well as the boost rectifier, operate under ZVS conditions.

[0061] To eliminate switching losses in the PWM converter, a zero-current, zero-voltage-switched (ZC-ZVS) cell which includes a snubber inductor, a clamp diode, a clamp capacitor, a main switch, and an auxiliary switch. The ZC-ZVS cell reduces reverse-recovery-related losses of the boost rectifier and also provides lossless switching for the main and auxiliary switches. The reverse-recovery-related losses in the boost topology are reduced by the snubber inductor, which is connected in series with the main switch (boost switch) and the boost rectifier, and which controls the rate of current change (di/dt) in the boost rectifier during its turn-off. Moreover, the main switch operates with zero-current and zero-voltage switching, and the auxiliary switch operates with zero-voltage switching. The proper operation requires overlapping gate drives of the main and the auxiliary switches, where the main switch becomes conducting or non-conducting prior to the auxiliary switch becoming conducting or non-conducting.

[0062] When the main switch of the boost converter with the proposed ZC-ZVS cell becomes conducting, the snubber inductor controls the rate of change of the current in the boost rectifier to reduce reverse-recovery-related losses of the boost rectifier. In addition, since the snubber inductor prevents the main-switch current from increasing immediately, the main switch becomes conducting with zero-current switching. Further, during the conducting period of the main switch, the snubber inductor and the output capacitance of the auxiliary switch form a resonant circuit, so that the voltage across the auxiliary switch falls to zero by a resonant oscillation. As a result, the auxiliary switch becomes conducting when the voltage across it is zero.

[0063] During the period when both the main and the auxiliary switch are conducting, the snubber inductor and the clamp capacitor form yet another resonant circuit through the closed switches. Due to this resonance, the current through the main switch is reduced to zero prior to the main switch becoming non-conducting, while the voltage across the main switch is clamped to zero by the conducting clamp diode and the auxiliary switch. Thus, the main switch turns off with zero-current-zero-voltage switching.

[0064] As pointed out, the auxiliary-switch-controlled resonant circuit of the above-described invention can be employed advantageously with a wide converter and inverter pulse-width-modulated topographies in order to provide soft-switching commutation of both the power modulating switch and the rectifier diode.

[0065] Since the main and the auxiliary switches have their source terminals connected to the circuit ground, a non-isolated (direct) gate drive can be used. In addition, because the proper operation of the circuit requires that the conduction periods of the main and the auxiliary switches overlap, a circuit of the present invention is not susceptible to failures due to accidental transient overlapping of the main and auxiliary switch gate drives. Further, the voltage and current stresses of the components in an active-snubber boost converter are similar to those in conventional hard-switched converters.

[0066] Having described various embodiments and implementations of the present invention, it should be apparent to those skilled in the relevant art that the foregoing is illustrative only and not limiting, having been presented by way of example only. The functions of any one element may be carried out in various ways in alternative embodiments. It is possible to modify further the converters described in U.S. Pat. Nos. 6,005,782 or 6,051,961, both of which are incorporated by reference as if fully provided herein. It is also possible to modify further the resonant type converters (described in U.S. Pat. No. 4,024,453; U.S. Pat. No. 4,720,667; or U.S. Pat. No. 6,496,388), or Phase-Shift-Full-Bridge type converters (described in U.S. Pat. No. 4,855,888; or U.S. Pat. No. 4,860,189) all listed patents are incorporated by reference as if fully provided herein, in order to reduce or eliminate a variety drawbacks that include input/output range limitations, isolation requirements, and complex control.

[0067] Additional embodiments are within the following claims.