One MOSFET is heated and another MOSFET is cool

[rckanta]

I am designed PWM inverter using sg3525, o/p frequency 55Hz and am using iron core transformer(I take the transformer from the ups , I think this transformer can handle up to 800W).


when I increased the output voltage to 200V(by varying the value of R8), one mosfet is heated and another mosfet is cool and the transformer is also cool.
I have already checked my circuit so many time, nothing is wrong with the circuit and the connection.

I don't know how to solve this problem. Please help me

[NiHaoMike]

There doesn't seem to be any provisions for leakage inductance or flux balance.

[Circlotron]

My guess is the oscillator resistor from pin 7 to pin 5 is too low value, despite the datasheet saying it can be zero ohms. Does the diagram show 10 ohms? Hard to read it. I had a similar problem - fixed it here -> https://www.eevblog.com/forum/projects/sg3525-oscillator-asymmetry-when-error-amp-is-maxed-out/

[David Hess]

Flux imbalance in the transformer is what comes to mind first.

[rckanta]

this 10R is for dead zone

[rckanta]

this 10ohms is for the dead zone not for the oscillator.

[rckanta]

how to balance the flux.
Please send me the circuit

[rckanta]

Is that can be caused by Back EMF.

[Circlotron]

this 10ohms is for the dead zone not for the oscillator.

It *is* part of the oscillator circuit though.
10 ohms is too small; try 390 ohms or so like in my link.
Trust me.

[rckanta]

Deadtime Resistor Range = 0 to 500R

if I use 300R, then o/p voltage will be very low because of large Deadtime(small duty cycle)

[David Hess]

how to balance the flux.
Please send me the circuit

The easy way is probably to use a current mode controller in place of the SG3525 like the UC3846.

Sometimes an air gapped transformer which can sustain any flux imbalance without saturation is used.

If the primary side voltage is high enough, then a half bridge with a series capacitor for balance is feasible.

There is probably some weird way to measure the current on both sides and adjust the drive to the MOSFETs to enforce current balance but a UC3846 would be easier.

[rckanta]

Can I solve this problem by discharging the primary back EMF using a diode

[Circlotron]

Deadtime Resistor Range = 0 to 500R

if I use 300R, then o/p voltage will be very low because of large Deadtime(small duty cycle)

Put your scope on pin 5 and see if the sawtooth waveform is *exactly* identical in duration and height from one cycle to the next at the setting that causes one mosfet to get hot.

[T3sl4co1l]

Primary EMF is discharged during dead time.

If the gate drive waveforms are asymmetrical, you must have dead time set to at least the same duration, otherwise the resulting DC bias is shorted across the transformer primary and bad things happen.

Drive should not be asymmetrical.  This can happen due to poor layout, too little compensation, etc.

Tim