EP0847614A1

EP0847614A1 - Direct current voltage converter with soft switching

Info

Publication number: EP0847614A1
Application number: EP96929380A
Authority: EP
Inventors: Henri Huillet; Didier Ploquin
Original assignee: Gaia Converter
Current assignee: Gaia Converter
Priority date: 1995-08-30
Filing date: 1996-08-29
Publication date: 1998-06-17
Also published as: CA2227966A1; AU6879696A; US5903446A; FR2738417B1; FR2738417A1; WO1997008812A1

Abstract

The invention relates to a direct current voltage converter with soft switching, comprising a transformer (Np, Ns) of which the primary is particularly of the half-bridge mount type and susceptible of being connected to an input voltage source (Ve) through two electronic switches (INT1, INT2) and of which the secondary, of the monoalternating type, is susceptible of being connected to a load through a series inductance (L2) and means (1-3, G1, G2) to alternatingly control the two switches, at fixed frequency, according to a regulation through pulse width modulation as a function of the output voltage (Vo) so as to achieve at the primary a zero voltage switching, said converter being characterized in that said secondary comprises, additionally, a resonant circuit (L1, C1) so as to perform at said secondary a zero current quasi-resonant switching.

Description

CONTINUOUS VOLTAGE CONVERTER WITH SOFT SWITCHING

The present invention relates to the conversion of electrical energy and more specifically to the creation, from a DC input voltage, of one or more DC output voltages and aims generally to provide regulation the most perfect possible of the output voltage (s) with respect to, on the one hand, variations in the input voltage, and, on the other hand, variations in the output current (s) absorbed by the load (s) , and, this, with the highest possible yield.

Various techniques have been implemented to date in order to ensure such regulation. The first types of voltage converters were linear regulators which heat dissipate the excess energy between the needs of the load and the capacities of the power supply. Such converters are bulky and have a poor efficiency, of the order of 50%.

Then appeared the switching converters with forced commutation in which the transfer of the just necessary energy, of the entry towards the exit (s), is done by quanta of energy taken periodically on the source, stored in reactive components then restored to the charge (s) by techniques known as switching techniques using electronic switch switching. The major drawback of these converters comes from the fact that each switching is accompanied by losses which increase with the number of switching, that is to say with the switching frequency.

However, with the evolution of the characteristics of the components, it has been possible to reach the optimum with this type of converter with devices operating at switching frequencies of the order of 200 KHz with an efficiency of up to 80%.

These converters have recently been supplanted by devices implementing resonance cutting techniques, the principle of which is to exploit, even amplify, the effects of parasitic elements by the addition of passive components forming resonant circuits. These structures then make it possible to create conditions where the voltages at the terminals of the switches or switching diodes, or the current flowing through them, are zero at the times when their switching is commanded. Switching losses are reduced accordingly.

This made it possible to increase the switching frequency, which exceeded 500 KHz, while at the same time reducing the weight and the volume of the converter.

Such techniques nevertheless remain difficult to implement and have drawbacks and limits. At each switching, high currents or overvoltages appear in the resonant circuits and therefore conduction losses. If we have reduced switching losses, we have increased conduction losses.

In addition, this drawback has made it necessary to over-size the cutting components compared to a conventional power supply, by a factor of 1.5 to 2. More recently, new conversion techniques have appeared with soft switching switching regulators. They are distinguished from resonance techniques by the absence of an expensive resonant circuit in loss by conduction and by switching conditions at zero voltage, therefore with lower switching losses, obtained from elements of conventional switching techniques. These new techniques are for example exposed in the documents US-A-4,441 .1 46, US-A-5,057,986 and US-A-5.1 26,931, as well as ^" in the publications:

. "Utilization of an active-clamp circuit to achieve soft switching in flyback converters" by R. Watson, F.C. Lee, G.C. Hua; Conference

PESC 1 994; . "Characterization of an active clamp flyback topology for power factor correction applications" by R. Watson, F.C. Lee, G.C. Hua; PESC Conference 1 994. However, the effectiveness of these techniques remains limited as regards the range of variation of the input voltage which is insufficient in most industrial applications. In fact, for large variations in this input voltage, the conditions for soft switching are no longer fulfilled and, in order to restore these conditions, arrangements must be made, in particular resonant circuits, causing serious problems of conduction losses.

The present invention aims precisely, in order to adapt these soft switching techniques to a wider range of input voltages, to restore said soft switching conditions but by simpler means on the contrary reducing losses by conduction, as well as the cost and size of the converter.

To this end, the subject of the invention is a soft-switching DC voltage converter, comprising a transformer, the primary of which is in particular of the half-bridge mounting type and capable of being connected to an input voltage source by by means of two electronic switches, the secondary of which, of the mono-alternation type, is capable of being connected to a load by means of an inductor in series and means for alternately controlling the two switches, at frequency fixed, according to a regulation by pulse width modulation as a function of the output voltage, so as to achieve a switching of zero voltage at said primary, said converter being characterized in that said secondary comprises, in in addition, a resonant circuit so as to achieve at said secondary a quasi-resonant switching at zero current.

According to one embodiment, the secondary circuit comprises, in addition to a rectification-filtering circuit of conventional type, a resonant circuit comprising a capacitor and a low value inductance, said resonant circuit being capable of creating in the secondary winding of the transformer , at each switch opening / closing cycle, a sinusoidal current which is zero or passes through a zero value in the time intervals during which the two switches are both open, so that the current of the primary winding either in the direction which favors the soft switching, without loss, of that of the two switches which closes.

According to this mode of implementation, said means for alternately controlling the two switches deliver square signals of identical and constant frequencies, the duration of a slot of one determining the duration of closure of one of the switches, this last duration being modulated as a function of the difference between the output voltage and a reference voltage, while the second signal commands the opening of the second switch before the closing of the first switch and the closing of said second switch after the opening of the first switch, the offsets between the opening of one of the switches and the closing of the other switch being equal and constant.

Such a converter combining zero voltage switching at the primary of the isolation transformer with a quasi-resonant switching at zero current at the secondary ensures excellent regulation of the output voltage, even with large variations in input voltage , without appreciable alterations of the conditions of the soft switching at the level of the primary switches, with as a consequence very reduced switching losses, but also almost non-existent conduction losses. Other characteristics and advantages will emerge from the following description of a preferred embodiment of the device of the invention, description given by way of example only and with reference to the appended drawings in which:

. Figure 1 is an electrical diagram of a converter of the prior art, of the soft switching type; . Figures 2a to 2g are diagrams of the times relating to the circuit of Figure 1; . Figure 3 is an electrical diagram of a converter according to the invention; . Figures 4a to 4h are diagrams of the times relating to the circuit of Figure 3; . Figure 5 is an electrical diagram of the secondary of a converter of the prior art; . Figures 6a to 6d are diagrams of the times relating to the circuit of Figure 5; . FIGS. 7a to 7d are diagrams of the times relating to the control circuit of the switches of the device of the invention, and

. Figures 8a to 8g are timing diagrams relating to an alternative embodiment of the converter of the invention. FIG. 1 represents a converter with soft switching switching regulation, of known principle, with primary circuit isolation transformer with half-bridge mounting.

The transformer includes a primary core part Np and a secondary core part Ns.

The polarized end of the primary winding, where the potential Vc is available, is connected via capacitors to the terminals of application of the input voltage Ve, while the other end of said winding, where is available the potential Vp, is connected to these same terminals via two electronic switches, respectively INT1 and INT2.

The polarized end of the secondary winding is connected by a diode D1 and an inductance L to one of the output terminals of the converter between which the output voltage Vo is available.

The other end of the secondary winding is connected to the other output terminal. A rectifier diode D2 is mounted between one of the ends of the secondary winding and the diode D 1 -inductance junction L. Finally, a capacitor Cs is mounted between the output terminals of the converter.

The control chain of the switches INT1 and INT2 includes a differential amplifier 1 connected to the output of the converter, a circuit 2 of regulation by pulse width modulation, two circuits G 1 and G2 for generating dead time tm connected, the one, directly to circuit 2, the other, indirectly to the latter, via an inverter 3, each circuit G 1, G2 controlling the opening / closing of one of the switches INT1, INT2. FIGS. 2a, 2b illustrate the shape of the opening / closing control signals of the switches INT1 and INT2, generated by the circuits G 1, G2.

FIGS. 2c and 2d respectively represent the variations of potential Vp and of the voltage Vs at the terminals of the secondary winding; Figures 2e, 2f and 2g respectively illustrate the variations in current

IL crossing the inductance L, of the current Is of the secondary winding and of the current Ip of the primary winding.

In FIG. 2a, T is the cutting period (opening / closing cycle of each switch INT1, INT2) an equal dead time tm being provided between the alternating actuations of these switches so that a simultaneous open state of the latter is obtained twice in each cutting period.

On the diagram of figure 2g we observe that the commutations during the two dead times tm, respectively at point A (just before switch INT2 closes) and at point B (just before switch INT1 closes ) operate smoothly because the current Ip is in the right direction, that is to say it is in the direction which will naturally charge or discharge the parasitic capacitors of the switches INT1 and INT2 so that the potential Vp will evolve simultaneously as shown in the diagram in Figure 2c.

However, such a device has its limits. When in fact the range of variation of the input voltage Ve is likely to be large, at high of Ve corresponds to a shape of the current Ip in the zone of point A such that it becomes positive and increases just before the switch INT2 closes at the end of the 1st dead time tm in FIGS. 2a, 2b.

One thus moves away from the conditions of a soft commutation and the fact of sufficiently increasing the amplitude of the current Ip so that the point A remains in negative values would pose serious problems of losses by conduction.

To remedy these problems, the invention proposes to modify the secondary circuit as illustrated in FIG. 3. To this end, in this FIG. 3, to the rectification circuit - conventional filtering L2, C2, (corresponding to the circuit L, Cs of FIG. 1) is added a resonant circuit L1, C1 at the output of the diode D 1, the diode D2 of FIG. 1 being deleted. The inductance L1 is preferably of low value.

The wave diagrams of Figures 4a, 4b, 4c, 4d, 4th, 4g and 4h correspond respectively to those of Figures 2a, 2b, 2c, 2g, 2d, 2f and 2e.

The nature of the secondary circuit of the converter of the invention makes it possible to give the current Is brought back to the primary a sinusoidal shape having

(Figure 4g), unlike a rectangular shape with a stiff front (Figure 2f), soft fronts. These soft edges maintain, in the switching zone A (just before the switch INT2, figure 4d, closes) the current Ip below zero, the current tending towards zero, which ensures a soft switching, however that in the switching zone B, said current Ip tending towards zero is always positive just before the switch INT1 closes. This sinusoidal form of said current Is brought back to the primary is precisely obtained by the type of circuit resonating at zero current of the secondary stage of the converter.

For more details on such a circuit resonating at zero current, known in itself, reference is made to FIG. 5 and to the associated waveform diagrams of FIGS. 6a to 6d. FIG. 5 represents a secondary circuit of a forced-switching chopper converter of the type described in US-A-4.41 5.959 and intended for input voltages not exceeding a few tens of volts.

Such a secondary is similar to that of FIG. 3 (the homologous components carrying the same reference numeral) with in addition a diode D2 in parallel with the capacitor C1.

FIGS. 6a to 6d represent diagrams respectively of the voltage Vs at the terminals of the secondary winding, of the voltage VC1 at the terminals of C1, of the current Is of the secondary and of the current IL2 passing through the inductance L2.

In such a circuit, the shape of the current Is (FIG. 6c) is indeed sinusoidal when the voltage Vs is established.

During phase 1, the voltage Vs is established, the inductor L1 charges linearly in current Is up to the value of the current IL2 passing through the inductor L2 and which is close to the current lo supplying the load, the voltage VC1 across the capacitor C1 remaining zero.

During phase 2 the elements L1, C1 are in resonance, the voltage VC1 (figure 6b) and current Is (figure 6c) having a sinusoidal shape, the voltage VC1 rising to twice the voltage Vs. In phase 3, the diode D1 is blocked in reverse, first of all by the current Is which is canceled, then by the fall of the voltage Vs, the capacitor C1 being discharged linearly by the current IL2.

In phase 4, the current IL2 continues to flow through the freewheeling diode D2, the voltage VC1 remaining zero. Thus, the voltage VC1 is filtered by L2, C2 so that its average value is equivalent to the output voltage Vo.

The regulation of the output voltage Vo therefore depends on the form of VC1, that is to say on Vs, lo and on the repetition frequency of the cycle of phases 1 to 4. If the voltage Vs is imposed by the input voltage Ve of the converter, the regulation, in this type of device of FIG. 5, of the output voltage Vo as a function of lo and of the input voltage Ve therefore consists in varying said frequency. On the contrary, in the assembly of the secondary (FIG. 3) of the converter of the invention, the freewheeling diode D2 of FIG. 5 has been deleted, so that the cycle of phases 1 to 4 is replaced by the cycle (FIG. 4h) with two phases 1 and 2. Phase 1, corresponding to phases 1 and 2 of FIGS. 6a to 6d, is a resonance phase and phase 2 is a phase of linear discharge of the capacitor C1 by the current IL2, which is almost constant, corresponding to phase 3 of Figures 6a to 6d. The disappearance of phase 4 is brought about by the suppression of the diode D2.

The difference in mounting between the two secondaries of Figures 3 and 5 shows the following essential consequences.

The shape of the voltage VC1 (figure 4f) is symmetrical around the value of the voltage input slot Vs (figure 4e). The filtering of VC1 at its average value by L2, C2 gives the voltage Vs which is none other than the output voltage Vo. Consequently, Vo = Vs whatever the repetition frequency and the output current lo. Incidentally, it should be noted that VC1 can be negative provided that during phase 2 VS remains in an algebraic value lower than VC1, the diode D1 being blocked in reverse. This characteristic fundamentally distinguishes this device from that of FIG. 5 (cf. also FIG. 6b). The regulation of the output voltage Vo is therefore very simple. It does not depend on said frequency but only on the amplitude of the voltage input slot Vs,

In the device of FIG. 3, the voltage Vs will therefore be regulated as a function of the input voltage Ve by the time t1 (FIG. 4a) defined by a period T of appropriate cutting of the control signals of the switches INT1 and INT2 of the assembly. primary bridge.

The regulation function is simple and independent of the load. Only the peak-to-peak amplitude of the voltage VC1 is linked to that of the output current lo and the no-load operation with lo = 0 poses no problem. Furthermore, the whole of phase 1 (FIGS. 4f and 4g) consists of sinusoidal signals. The duration of phase 1 is constant and breaks down (Figure 4g) in a half-period of duration Tr / 2 (Figure 4g) at the resonance frequency given by L1, C1 and two intervals located on either side.

The duration of the input slot Vs can be indifferently between the duration of phase 1 and the duration of phase 1 plus half the duration of phase 2, or even more. This possibility of variation will be exploited in the half-bridge stage to vary t1 (FIG. 4a) as a function of the input voltage Ve and obtain Vs = constant = Vo.

Finally, the removal of the diode D2 brings the advantage of having fewer losses by conduction or reverse current and leads to a reduction in the size of the device and its cost.

FIGS. 7a to 7d illustrate by way of example the waveforms generated by the control circuit of the switches INT1 and INT2 of the primary of the converter of FIG. 3.

FIG. 7a represents the square signal Va generated by the circuit 2, when the amplifier 1 detects a difference between the output voltage Vo and a reference voltage greater than a predetermined threshold.

FIG. 7b represents the signal Vb which is the inverse of Va and FIGS. 7c and 7d respectively represent the signals Vd for controlling the switch INT1 and Vc for controlling the switch INT2, delivered respectively by the circuits G 1 and G2.

The circuit (1 to 3, G 1, G2) performs a pulse width modulation t / T = function of said difference detected by the amplifier 1 in order to obtain the time t1 defining the length of the voltage pulse Vs. Vc is generated from Vb and

Vd from Va, the falling edges being transmitted without delay and the rising edges being transmitted with a delay tm.

In summary and in other words, the converter of the invention makes it possible to highlight the following observations.

There is a soft switching to zero voltage at the primary. Indeed, at the opening of one of the switches (for example INT1), at the start of an interval tm, the current in the winding of the primary persists thanks to the magnetizing current and to a current in the secondary reduced to the primary (Is ) of sinusoidal shape. This current is in the direction which favors the natural commutation without loss (transition A) of the other switch (INT2). At transition A, in fact, when the switch INT1 opens, the current of the transformer primary (Ip) is negative and flows from point Vp to point Vc. It charges the stray capacitance of the switch INT1 and discharges that of INT2, the potential across the terminals of INT2 decreases to cancel itself. The primary current then continues to flow in the stray diode of INT2 at the same time as it decreases in amplitude. The INT2 switch is then closed before it becomes positive. The power lost during switching is zero.

By adjusting the amplitude of the magnetizing current, by determining the value of the inductance of the transformer, so that this amplitude is greater than twice the average value of the secondary current brought back to the primary, we will have a current Ip which will be, during the time tm during which the switch INT1 (transition B) is closed sufficiently positive to also carry out a soft switching for the other switch. These natural, lossless switches are obtained without adding a resonant circuit.

The switches INT1 and INT2 are for example MOS transistors. They can be made up of other components such as in particular bi-polar, GTO or IGBT transistors. MOS transistors naturally exhibit parasitic components (capacitance and diode) and if the switch types chosen did not include such components, in particular the diode function, it would then be necessary to add this function, which is necessary for the proper functioning of the device.

The diode D 1 is switched to the locked position in reverse with conditions di / dt and dv / dt lower than in a conventional technique, therefore with less loss by reverse overlap.

The regulation law is Vo = Vs = n * Vc = n * Ve * t1 / T. It is linear and makes it possible to obtain a range of variation of Ve in a ratio of 2 minimum. The regulation process is obtained by varying t1 while retaining tm = cte and t1 + t2 + 2 * tm = summer = 1 / F where F = switching frequency. Theoretical calculations show that the magnetizing current is increasing with Ve while conserving through the regulation Vo = ete. This is favorable to the soft switching criteria.

Fixed frequency operation allows synchronization of several converters.

Finally, the measurement of the amplitude of VC1 is a good way to measure the output current lo and to implement functions such as: current limitation, current and short-circuit security, servo control with current loop in parallelization of several modules for an equal and controlled distribution of the respective currents.

In accordance with the invention, a 30 watt converter of the type in FIG. 3 has thus been produced by way of example, having several outputs of different values: 3.3; 5; 1 2; 1 5; 24 and 28 volts, regulated in a range of input voltage values between 200 and 400 volts, the switching frequency being 500 KHz.

Such a converter had an efficiency of 92% for 15 volts of output and had a volume of 35.6 x 52.5 x 12.7 mm, which is two to three times less than that of the converters of the moment.

It should be noted that the converter can have several identical outputs, for example mono-alternation as shown in the drawings.

Furthermore, the resonance inductor L1 can be a discrete component or be an integral part of the leakage inductance of the transformer.

FIGS. 8a to 8g represent wave diagrams corresponding to an alternative embodiment of the converter consisting essentially of a modification of the dimensioning of the resonant circuit L1 C1 of the device of FIG. 3, this circuit being moreover unchanged.

Regardless of the different dimensioning, implying a lower resonant frequency, the device works on the same principle with the same commands and the same controls. The aim sought by such a different dimensioning is to carry out a voltage zero switching, hereinafter denoted ZVS, of each primary switch (INT1, INT2) following the opening of the other switch. Referring to FIG. 4d, it can be seen that the condition ZVS is fulfilled if the primary current is negative in A and positive in B all the time of this switching which occurs from the start of the time intervals tm.

The ZVS switching to A is ensured by two parameters: the quasi-resonance in current at the secondary brings back to the primary a current at A which is only gradually increasing. In the converter of the invention, the primary current at A remains easily negative the time that the ZVS switching is completed. The balance of the primary capacitive bridge means that the average primary current is zero. The presence of the output current lo leads to a shift in the negative of the average magnetizing current of the transformer and an even more negative value of the primary current at A. Consequently, the more power the converter provides, the more the condition ZVS at A assured.

With regard to ZVS switching at B, the current brought back from the secondary is zero and only the primary magnetizing current must fulfill the condition of being positive. Unlike point A, but for the same reason, the more power the converter provides at output, the more the offset in the negative of the magnetizing average current causes the primary current towards the negative at point B. Also, in order to have at point B a primary current which remains positive, so as to fulfill the condition ZVS at said point B, the resonant circuit L1 C1 is dimensioned with a lower resonant frequency, therefore a longer period. At point B, a non-zero secondary current is thus obtained which superimposes on the primary magnetizing current a positive reduced value contributing to the conservation at B of a positive primary current. Clearly, the current quasi-resonance phase is interrupted by early opening of the switch INT2. It is thus possible to minimize the amplitude of variation of the magnetizing current as well as the conduction losses which result therefrom while retaining the switching property ZVS at B. FIGS. 8d, 8f and 8g illustrate the above, FIGS. 8a, 8b, 8c, and

8e being identical to FIGS. 4a, 4b, 4c and 4th. In particular, one will note on ia figure 8d the very marked shift towards the positive of the current Ip at the point B and on figure 8g the phase of quasi-resonance in interrupted current of INT2.

The invention applies of course to other primary assemblies than that of FIG. 3 and, in general, to any primary, in particular of the Buck type, comprising two alternately controllable switches so as to obtain switching gentle at zero voltage.

14.T

LEGEND ASSOCIATED WITH FIGURE 2g

- Ip of zero mean value

^ Is brought back to primary

Magnetizing current of average value = - average value of% 'a

Claims

1. Soft-switching DC voltage converter, comprising a transformer (Np, Ns), the primary of which is in particular of the half-bridge type and can be connected to an input voltage source (Ve) via iary of two electronic switches (INT1, INT2) and the secondary of which, of the mono-alternation type, is capable of being connected to a load via a series inductor (L2) and means (1 to 3 , G 1, G2) to alternately control the two switches, at fixed frequency, according to a regulation by pulse width modulation as a function of the output voltage (Vo), so as to achieve a zero switching of the primary voltage, said converter being characterized in that said secondary comprises, in addition, a resonant circuit (L1, C1) so as to achieve said secondary quasi-resonant switching at zero current.

2. Converter according to claim 1, characterized in that the secondary circuit comprises, in addition to a rectification-filtering circuit (L2, C2) of conventional type, a resonant circuit comprising a capacitor (C1) and a low inductance (L1) value, said resonant circuit (L1, C1) being capable of creating in the secondary winding (Ns) of the transformer, at each opening / closing cycle of the switches (INT1, INT2), a current of sinusoidal shape (Is) which is zero or goes through a zero value in the time intervals (tm) during which the two switches (INT1, INT2) are both open, so that the current (Ip) of the primary winding (Np) is in the direction which promotes gentle, lossless switching of the two switches which closes.

3. Converter according to claim 2, characterized in that said alternating control means of the two switches (INT1, INT2) deliver square signals (Vc, Vd) of identical and constant frequencies, the duration (t1) of a slot of one (Vd) determining the closing time of one of the switches, this last time being modulated as a function of the difference between the output voltage (Vo) and a reference voltage, while the second signal (Vc) commands the opening of the second switch before the closing of the first switch and the closing of said second switch after the opening of the first switch, the offsets (tm) between the opening of one of the switches and the closing of the other switch being equal and constant.

4. Converter according to one of claims 1 to 3, characterized in that said resonant circuit (L1, CD of the secondary is dimensioned with a resonant frequency sufficiently low to obtain a primary current of suitable sign allowing switching to zero voltage of each switch (INT1, INT2) following the opening of the other.