Transformer mosfet gate drive?

On a sunny day (Thu, 18 Jun 2009 10:21:19 -0400) it happened Hammy wrote in :

I think their circuit uses the input capacitance of thw power MOSFET to keep the voltage high. The small MOSFET switches on when the T1 secondary goes low (source below its gate), then pulls the gate of the big MOSFET low.

Note Z1, that impedance must be high for those waveforms to happen on T1. So if you drive T1 hard, and it never saturates, nothing much will change in your waveform, square in - square out. Try a resistor in series with the primary of T1... or less turns... Or lower frequency ... ?

Looks tricky. You can usually get away with just a series cap and a resistor||diode from source to gate. That (almost) always applies the transformer p-p voltage to the gate as a unidirectional pulse.

John

I did that very same scheme for a satellite "spin launcher" (Sperry Satellite Systems) around 1990... except I used bipolars for the DC restore section to allow for an accurate and very wide PWM range.

One has to be a little more elegant if you need things to work 100% of the time ;-)

...Jim Thompson

John what you recommended was the first result from a search, and then I read this from Ridley.

"The circuit works well during steadystate operation (a gate load resistor is Recommended), but when the PWM controller shuts off, the dc blocking capacitor Is connected across the gate drive transformer for an indefinite period. This can lead to saturation of the transformer, as shown in Fig. 6b. When the transformer Saturates, the secondary is a short circuit, and the secondary capacitor can turn on the FET. The saturation can be avoided with a gapped core, and smaller value of capacitor, but this will increase the reactive current needed from the gate driver, and may produce other problems."

Fig. 6b

The controllers I'm using tl3844 (flyback) and ucc25600 (llc half bridge) are limited to 50% duty max I knew my Vgs would reduce at 50% duty I was just surprised it reduced as much as did. So I must be doing something wrong.

I'll do some more testing with the input impedance and see if I can't get the Vgs unidirectional. If not I'll try the other restorer that John recommends.