Uploaded January 2026 | Updated September 2026, 2 weeks ago
As expected this decent sized 1:1:1 GDT "works" but the circuit needs a good ferrite low leakage transformer for the best performance. Right now it starts to sag too low below 24V when loaded at 8-10W but is acceptable at 5W. It'll be good for more than 10W when I put the right small ferrite transformer in it. I'm hoping to hit about 85-90%. I don't want to just wind one though, and would rather source something I can finalize a circuit with to make more of. After that I will spend time on the output stage trying to filter it as best as I can to reduce the ripple (really just switching noise). IRF540's are not exactly modern switches but meh. To really push efficiency one would need low gate charge and output capacitance fets (sic/gan), of course a very clean board, and the good transformer. 90% then becomes a reasonable baseline whereas even 85% may be difficult using these parts.
TL494 Push-Pull common emitter, collectors pulled to 5v
74HC14 to invert signals
UCC27524 push-pull fed
IRF540's
~90khz ( I could drop this for better performance but..)
The TL494 is not exactly cutting edge technology either but it's a handy chip and it's always good to be able to set some dead time to avoid shoot through. Otherwise I suppose you could use any oscillator that will give you a 50% duty cycle and then try to RC delay the Schmitt trigger to attempt dead time. (I should add that you may not want to run a high duty cycle with a super tiny transformer unless you plan to run lower voltage into the primary). I'm kind of just using a bunch of these thruhole IC's I had for a while but man, all SMD would make everything about it pretty tiny.
As expected this decent sized 1:1:1 GDT "works" but the circuit needs a good ferrite low leakage transformer for the best performance. Right now it starts to sag too low below 24V when loaded at 8-10W but is acceptable at 5W. It'll be good for more than 10W when I put the right small ferrite transformer in it. I'm hoping to hit about 85-90%. I don't want to just wind one though, and would rather source something I can finalize a circuit with to make more of. After that I will spend time on the output stage trying to filter it as best as I can to reduce the ripple (really just switching noise). IRF540's are not exactly modern switches but meh. To really push efficiency one would need low gate charge and output capacitance fets (sic/gan), of course a very clean board, and the good transformer. 90% then becomes a reasonable baseline whereas even 85% may be difficult using these parts.
TL494 Push-Pull common emitter, collectors pulled to 5v
74HC14 to invert signals
UCC27524 push-pull fed
IRF540's
~90khz ( I could drop this for better performance but..)
The TL494 is not exactly cutting edge technology either but it's a handy chip and it's always good to be able to set some dead time to avoid shoot through. Otherwise I suppose you could use any oscillator that will give you a 50% duty cycle and then try to RC delay the Schmitt trigger to attempt dead time. (I should add that you may not want to run a high duty cycle with a super tiny transformer unless you plan to run lower voltage into the primary). I'm kind of just using a bunch of these thruhole IC's I had for a while but man, all SMD would make everything about it pretty tiny.
