0
I have designed a Push-pull converter as shown at the attachment. The input is 30V and the output is 400V with the power of 400W. I have designed the MPPT algorithm, magnetic components, etc and the circuit is operating well. In the starting, there is a soft start such that the duty cycle is increasing step by step from 0 to 30% with an increment of 1.5%. Then MPPT algorithm starts to run and makes the duty cycle between 5% and 45%. Sometimes, usually during the soft start, one of the MOSFET is shorted and the operation fails. I have examined the drain current and drain-source voltage there are huge spikes especially during the low duty cycle which occurs at the soft start. When the duty cycle of MOSFETs is getting higher the current ad voltage spike decrease. ( current spike is at the beginning of the MOSFET on time and the voltage spike is at the end) I know how the voltage spike occurs due to leakage inductance and hard switching. But I could not really understand the reason for the current spikes at the beginning. I think they are because of the MOSFET and transformer parasitic capacitance but I am not sure. And obviously, I could not find a solution and I think the MOSFET is being shorted because of this problem. Even, I have disconnected the output LC filter and directly connected the load to the bridge rectifier but the problem still continues. Can anybody give me a clue about the problem and save me, please?
What is step-up ration of the transformer? E.g. if it is 1:10 this circuit cannot properly produce more than approx 300V. Note that when one transistor closes and its drain voltage spikes (like in flyback) the potential of the drain on the other transistor goes down and may go below ground thus opening the body diode. If it happens you waste energy, may overheat transistor. The body diode often is very slow and that gets the circuit more in trouble.
The quick fix might be to add two Schottky diodes in series with the drains (between drains and transformer), although some power will be lost in them. Otherwise, you need to be very exact with the transfomer step-up ratio and circuit may only work in rather narrow range of voltages.
Dear Gökhan Bayraktar
Hope you are well.
The think the current spike in the beginning of the start may be due to the parasitic capacitance of the transformer because only current spikes can be caused by charged capacitive switching. Current interruptions in the circuit can cause voltage spikes across the inductors.
As for the the short circuit of one of the transistors it can only be affected by continuous on gate voltage on the transistor. So, you must take care of your pulse width control not to cause 100 percent duty cycle.
I think then after this analysis that you have to trouble shoot your control algorithm. You have to prevent the occurrence of this dangerous mode of operation.
Best wishes
I have added some waveshapes, gate driver circuit and more detailed push-pull schematics
Hi Gökhan Bayraktar ,
in contrast to your description, your scope images seem to indicate that the problematic current spikes are absent without LC filter. If so, the LC filter is the suspect.
Where did you measure the Lout current in your last picture? Since it has a DC component it has to be behind the rectifier. Because the initial current spikes can be seen in Lout, too, the culprit has to be located behind your point of measurement.
If the point of measurement is located between rectifier and filter, I guess that the parasitic capacitance of the inductor of the filter is too high. Possible solution: Either buy an inductor with high resonance frequency or, if you are building your own Ls, change the layout of the winding in such a way that the ends of the wire are spatially separated as much as possible (a good solution would be a single layer winding). If you have to use several layers, and there are considerably more turns per layer than number of layers, try to do the winding not in horizontal layers but in vertical "slices", if possible separated by sheets of paper or plastic.
Joerg Fricke thanks for the reply.
But, except for the last one "experiment with LCfilter", the images are without LC filter. And yes with the LC filter that spike is more available but without it still the spike exists as seen in "WhatsApp Image 2021-06-08 at 20
.08.28.jpeg".
I agree the LC filter does something but also I think at the primer side parasitic capacitance of transformer and MOSFETs and the leakage inductance of PCB layout cause something also. Because when I take away the secondary winding from the transformer core and run the circuit current spikes occur at low duty cycle also.
Lout current is measured after rectifier before the inductor as you suggested
Additionally, the figure "named experiment with LC filter" the inductor current has a sudden fall and rise during it is at its peak. At that moment mosfet is turning off. Why does that behaviour occur? I observed that the increase of Mosfet drain-source voltage and decaying of seconder voltage of transformer are very long. Such that Mosfet VGS signal is decaying at 20n but the VDS voltage is increasing 550ns. At this time sudden falls and rise occurs.
Gökhan Bayraktar ,
if the resonance frequency of the 6 mH coil in the filter is located somewhere around 2 MHz you probably would get currents with such spikes as you have. The coil gets excited by both switching on and switching off, only more so by switching on.
Regarding your last remark about VGS and VDS, to which picture are you referring? I can find VGS only in lowduty.jpg, and there VGS reaches the threshold some ns before t = 0, while the VDS peak starts at around t = 0. So the delay might be about 50 ns.
Joerg Fricke I did not post the gate-source signal here but it is decaying from 15V to 0V within 20-30ns. The lowduty.jpg is the starting transient of the circuit but I am not referring it. The problems about the drain and inductor currents are always there both low duty and high duty ratio operations.
The problems can be seen from the drain current and Lout current figures of "experiment with LC filter". First problem is huge current spike and oscilation at turn on. Second is the little fall and rise at the Lout inductor current when it becomes its peak value at the mosfet turn off. In theory it should increase when one one the mosfet is conducting then it should decay to minimum value. But in the maximum value there is a fall and rise.
I cannot find anything about the first problem why it behaves like that?
But for the second the fall and rise time is 550ns. I see this 550 ns at the Mosfet turn off. At the turn off there is a voltage spike as expected but then it becomes nearly zero then increase the supply voltage as suppossed to be. This behaviour also lasts 550ns. So I think the the problem at inductor current may be from this behaviour. But I cannot be sure and could not find a solution.
Joerg Fricke And yes the frequency of current oscillation is 2.1 Mhz. So Can you make it a bit clearer please?
Joerg Fricke As you suggested (thank you for that again ) I redesign the output filter in order to reduce the parasitic capacitance. So I have reduced the inductor current sudden fall and rise ( the part shown with a question mark in the above picture). But the current oscillation when the MOSFET turn-on still continues. I think it's because of the total resonance effect of parasitic inductance and capacitance of the components at the primer side.
I put a ferrite bead series with the output filter inductor but nothing changed.
What should I do? Can you please help me again?
Gökhan Bayraktar Note that you have negative spike in the drain current - likely it means you get current through body diode of MOSFET (source to drain).
E.g. something like :
https://www.5scomponents.com/pdf/copyright_ieee_01mr.pdf
Note that in push-pull configuration if voltage on the drain of one transistor goes up, it goes down on another transistor's drain. And it may go below the rail which usually means opening the MOSFET body diode and that is not good situation at all.
- Badges
- Science topic