gate driver transformer setup

I want to drive an isolated half-bridge of IRF540 using gate drive transformers from an SG2525 or similar. I have made a scetch of how it could look, and would appreciate pointers and help on this matter.

mosfet drive

Is the setup sane? The half-bridge will operate from +-47V, the frequency is 53kHz (~9 us per mosfet). Low switching losses is more important than low ESR-losses as the device wont have any current to conduct for most of the time (purpose is voltage balancing due to energy backflow from the load). The reason for IRF540 is that I have 25 pcs left of them..

Second, am I overengineering this thing using a totem-driver after the SG2525 to drive the transformer? The SG2525 is in itself capable of driving and sinking 500mA.

Any hints on component values? I guess I could pick them out using the oscilloscope, but a priori knowledge is always better.

/Daniel
 
guitar_joe said:
If you are switching really fast into a high gate capacitance then those totem poles might be necessary.
Also if the transformer steps up the voltage at all calculate what is reflected back to the totem poles in current drive requirements for the fets.

Yeah, besides - the mosfets will be backdriven as I will use their body diodes as rectifiers.

The transformers will be 1:1 as I intend to run the thing off a car battery - but I suspect I still should use a sufficient amount of violence just to be safe.
 
What about this circuit? try #2 Still to be driven for constant duty cycle without feedback from a totem pole BJT-output stage (SG2525).

I guess the turnoff current should be controlled enough to avoid gate oscillations during turn-off? I use that scheme in my UcD-amp without any resistors in the turnoff path with no such oscillations.

Have I forgot anything?
* catch diodes after the totem-pole stage to handle the dead bands introdued by having one BJT totem-pole drive another BJT totem-pole
* flux imbalance capacitor on the primary (is this really needed if I replace with a resistor? should do the same job, but with voltage drop as the xfmr walks up to flux imbalance)
* zeners on the gate-source on the mosfets to protect against leakage inductance energy
* diodes directly on the secondaries of the xfmr to break the current path during turn-off to avoid exciting the leagage inductance more than necessary
* finally, a bit more than 1:1 winding ratio to compensate for the 2.8V drop from the totem-pole stages, and the controller outputs, aswell as the dual diode drops of 1.4V on the xfmr secondary sides

How close am I to something that actually holds up for prototyping?
 
Doesn't the Totem pole transistors in most cases protect the PWM IC chip?
IMO I believe it does with the right components. Since you are driving a transformer especially, strong outputs are a benefit, the totem pole makes more sense to me. :)

I've seen SG3525 IC blow before when directly driving MOSFETS in a small power inverter. The MOSFETS were fine, but it blew the chip because of the excess load :rolleyes: I switched the IC to another good one and it was fine.

With a PWM testing/mosfet driver circuit I built with a TL494, if the outputs are shorted (in case of driving shorted MOSFET) the 1 ohm, 1/8W gate resistors go up in smoke, :hot: but the chip stays alive, and so do the totem pole transistors because they are overrated (MJE15034/35), the 1 ohm 1/8W resistors go up in smoke before any other parts would. That's what I like about the totem pole, because I've seen a lot of commerical SMPS without totem poles, and dead PWM chips.:att'n:
 
switchmodepower said:
You also need protection diodes from primary of drive xfmr to the +15V rail (across top trasistor drive). :smash:

I would also but a 10k or so directly across the secondary winding- used during dead time..

Other than that it looks like one of my circuits:D


Is the purpose of the 10kOhm to catch the leakage inductance spike during turn-off of the mosfets?

Btw, how high should I aim with the turn-off buffer PNP? The FZT790A seems ideal, but it not very easy to get hold of over here. Unfortunately all other orderable parts I can find inside swedish borders have lousy gain aboce Ic = -0.5A. I havent done much testing to compare fall times between say FZT790A (I used them in my UcD-amp) and lesser ones such as BC640. Will the fall-time difference be significant for a switching frequency of 53kHz?

I also plan on trying to drive the transformers without the totem pole-buffer to see how much that trade-off costs in performance.