cupoftea
Advanced Member level 5
Hi,
Nobody ever does anything in SMPS if it means more cost and circuitry for no reason.
So why does everyone use gate drive transformer for top and bottom fets in a Full Bridge?....after all, surely you can get away without bothering for the bottom fets?
This is why, if you design a full bridge SMPS, for 400vdc input, and a 100W or more, then you must drive top and bottom fets with a gate drive transformer. Avoid the temptation of just using the raw fet drive to drive the bottom fets…even though it costs less…..you can see from reverse engineering them, that this is always done.
If you do not do it, then you risk reverse recovery of the upper fets anti-parallel diode, and lots of smoke. The fact is, that, if you just use the raw fet drive for the bottom fet, then it will turn on quicker, ..will then drag a spike of current through the upper fets CDS capacitance…..then there will be a resonant ring with the primary and the lower FETs cds, and the “back-swing” will go right through the upper anti-parallel diode….the top fet then turns on, so its (conducting) diode does not then get properly reverse recovered……and then when the bottom fet eventually turns on…yes , you’ve guessed it, the top diode is still conductive and a massive reverse recovery current spike goes through the bottom fet…..lets out the magic smoke.
This is why we see everyone doing this (driving both top and bottom fets with a gate drive transformer…even though it looks like you could leave it out for the bottom fet.)
Have you encountered this issue?...and have you got into using SiC FETs to get round it (internal diode has no reverse recovery).
And also, we can forget trying to simulate the issue...the sims dont model reverse recovery well enough.
Nobody ever does anything in SMPS if it means more cost and circuitry for no reason.
So why does everyone use gate drive transformer for top and bottom fets in a Full Bridge?....after all, surely you can get away without bothering for the bottom fets?
This is why, if you design a full bridge SMPS, for 400vdc input, and a 100W or more, then you must drive top and bottom fets with a gate drive transformer. Avoid the temptation of just using the raw fet drive to drive the bottom fets…even though it costs less…..you can see from reverse engineering them, that this is always done.
If you do not do it, then you risk reverse recovery of the upper fets anti-parallel diode, and lots of smoke. The fact is, that, if you just use the raw fet drive for the bottom fet, then it will turn on quicker, ..will then drag a spike of current through the upper fets CDS capacitance…..then there will be a resonant ring with the primary and the lower FETs cds, and the “back-swing” will go right through the upper anti-parallel diode….the top fet then turns on, so its (conducting) diode does not then get properly reverse recovered……and then when the bottom fet eventually turns on…yes , you’ve guessed it, the top diode is still conductive and a massive reverse recovery current spike goes through the bottom fet…..lets out the magic smoke.
This is why we see everyone doing this (driving both top and bottom fets with a gate drive transformer…even though it looks like you could leave it out for the bottom fet.)
Have you encountered this issue?...and have you got into using SiC FETs to get round it (internal diode has no reverse recovery).
And also, we can forget trying to simulate the issue...the sims dont model reverse recovery well enough.
Last edited:
