Keep burning out MOSFET (25 of them)

Hi everyone!

I'm working on a boost converter circuit and I keep burning out my main MOSFET, the one responsible for switching the inductor current. The RDS-on is very low, around 0.008 ohms, and very low gate capacitance for this particular chip (FDS6670A - N channel), SOIC8 package.

I have adequate heat sinking, the drain pins are soldered to a copper bar (for the prototype so far).

So here's my primary question... I've been testing different duty cycles and PWM frequencies (many of which I can get to work, 20 or 30 volts on the output with no load)... But as soon as I put around a 1 amp load on the output... the MOSFET gets real hot real fast and my power supply jumps up to its max of 15 amps, it's fans kick on, and MOSFET smokes.

All I want to be able to do is to take a 4 volt source and convert it to about 8 volt and around 2 amp, it's not critical, but I'm not getting anywhere close to it right now. My understanding is that if the MOSFET is turned completely on and off fast enough, I shouldn't even be having this issue at all. 2 amps is well below the FDS6670A max current, AND with an RDS-on of 0.008 ohms the power dissipated should be practically nothing.

Also, on the oscilloscope I've manually tuned away almost all voltage spikes (when measured between source and drain, and there's very little ringing) I don't know what else to look for!

What am I doing wrong????

Attached is the circuit.

What is the best PWM frequency? Duty cycle?
What are the PMOS/NMOS predriver transistors?
What is the pull-down on the gate of the FDS6670?

I'm betting that the problem is the 20K pull-up on the predriver.

The PWM I've been using is about 30KHz, this is slow enough that heat wouldn't be an issue right? Duty cycle 20%

The predriver transistors are the same as my power transistor and also it's P complement.

Thanks for your suggestion. I've also tried op amp predrivers (got the signal looking real nice on the scope too (perfect square wave)... I've also tried direct from the microcontroller, it worked about the same. It just over heats when there is a load attached to the boost output....

At 30kHz, the inductor needs to be 100uH to 330uH, rated for a peak DC current of several Amps.

With only 1uH, look at the currents:



Only change is inductor now 220uH:

Thank you so much Mike!!!!!! You saved me !!

BosonCannon said:

Thank you so much Mike!!!!!! You saved me !!

Click to expand...

Make sure you look at the sims again. The first time I ran them I had the pwm freq at 3KHz, not 30KHz. That made the 1uH currents even more dramatic.

You also need to look a the peak current rating of the inductor. If the inductor saturates, the peak current will go way up and likely zap the MOSFET.

Edit: You can check the peak inductor current by adding a small resistor (say 0.05Ω) in series with the MOSFET source to ground and measure the voltage across that with an oscilloscope).

Love this site man... So many knowledgeable people willing to help.

Often a small amount (10-20R) of resistance in series with the gate of the MOSFET helps prevent ringing. I'd also suspect the 20K resistor may be a little high for some clean switching