SPWM converter mosfet failure

nikhilmahasvar said:

Split bobbin will increase your leakage inductance which is required to filter the HF. You may apply SPWM 1 to A and 2 to B. Complement of 1 should be given to C and 2 toD
The dead time is to be given for the compliment signal.

Click to expand...

Assuming i have and H bridge with N-Mosfets, If i give the compliment of signal 1 to C and the 2 to D with more dead time, how will the signals be in sync? Dont mosftets A and D have to be open the same time? If one delays what will happen?

Dimitrisvlamis said:

Assuming i have and H bridge with N-Mosfets, If i give the compliment of signal 1 to C and the 2 to D with more dead time, how will the signals be in sync? Dont mosftets A and D have to be open the same time? If one delays what will happen?

Click to expand...

Apply only few micro seconds dead time, just to avoid the cross conduction. It will work.

nikhilmahasvar said:

Apply only few micro seconds dead time, just to avoid the cross conduction. It will work.

Click to expand...

I dont quite understand this, could you explain to me why should i add more dead time? And shouldnt i feed the same signal to the mosfets? The must act like sink source right? So why not put signal 1 to A and D ?

So i found the schematic of the UPS i took the transformer from. I saw it has some resistors and capacitors coming from the drain. Do you think is some kind of snubber or its just for noise?

As you can see it was working like this, simple push pull topology and i keep burn my mosfets. The mosfets on the PCB were IRFZ46N 2 pairs in parallel.

It looks like the diode mirror to sense the MOSFET current. It can only prevent over current.

So i put everything back to a UPS with my transformer and i saw somethin weird. This ups is Modified sinewave and the pulses at the fets were like this. 5ms ON 15ms OFF with dead time of 5ms...isnt that a bit too much? It works like this but 1/4 of the period for dead time isnt a waste of energy you could potentially transfer to the secondary? Maybe it needs that time...?

I did the following test:

I put the transformer back and put in the place of battery my power supply which goes from 10.3v to18V and started the UPS at 13.9V which is almost full battery. I disconected mains cable to simulate a power break, and it started working, the mosfet signal was like the image above. But as i raised the voltage the on time went down, at 16V i got 3.2mS of on time in each signal with a dead time of more than 7mS. And with a voltage of 10.3V (which i think it should have closed because it would damage the voltage but anyways) i got an on time of 6-7mS on each gate.

Anyway the conclusion is that the dead time between the signals is always more than 3mS. Do you think is wise to make an SPWM inverter with a dead time between the spwm signals 3+ms?

Kindly refer some SPWM notes, Please don't get confused between square wave and SPWM switching topologies. when you switch with 16Khz freq. your pulse itself is only 60us you may need to insert 1-2us dead time depending on your driver and MOSFET speed.

nikhilmahasvar said:

Kindly refer some SPWM notes, Please don't get confused between square wave and SPWM switching topologies. when you switch with 16Khz freq. your pulse itself is only 60us you may need to insert 1-2us dead time depending on your driver and MOSFET speed.

Click to expand...

Could i keep burning my mosfets because of a very low gate resistor? I use 4.7Ohms and i saw the same mosfet with 40Ohms and 10ohms. My mosfet is IRF3205

Increasing the gate series resistance will slow down the switching speed of the MOSFET and increase it's heat dissipation. Keep it as low as the gate driver can manage, typically just a few Ohms.

I'm still suspicious about that transformer but I know nothing about the original U2 device. I wonder if it was producing 50Hz and driving the transformer at that frequency, it would tie in with the switching waveform and relatively square wave at the output when unloaded. Pushing it to 15KHz would account for the MOSFETS suffering overload.

This ups is Modified sinewave

Click to expand...

would seem to confirm that, a PWM driven output would be more sine shaped.

Brian.

Lack of dead time, C9 looks like a bad idea as the leakage of the Tx is unknown, it is easy to get DC imbalance on the pri side and then the Tx saturates and then bang - hence why H -bridge most often used ...

how code sir

The failure of ]MOSFETs has been a problem in LLC resonant converters. Poor body diode performance of the MOSFET can cause unexpected system or device failures.
Such as severe through current, body diode dv / dt, breakdown dv / dt, and gate oxide breakdown under abnormal conditions, such as startup, load transients, and output short circuit