How to design a good gate driver?
Why here's another in-depth technical topic by yours truly. 
I need some help on designing a fast gate driver. As you know i have built a 2-transistor forward 70 volt supply using nothing but the 555 timer chip, a LM311 comparator and a TL431 as voltage reference. Well, it was intended to be 70-0-70 volts actually but i got lazy and never wound the second 70 volt winding as i didn't really need the supply at that time (it was intended for an amplifier i didn't get around to building yet).
The supply uses a base drive transformer from an ATX power supply. However it needs heavy snubbing in the primary to produce an acceptable waveform, otherwise it goes all crazy. The two secondaries of the transformer simply drive the switching MOSFETs directly. It works quite well, but the fall time seems a little slow to me. So maybe i need something better.
I'll have to check again what "a little slow" means in that circuit, but the 555s that i am using clock in at 100ns rise time and 20ns fall time. That is quite respectable given that in the datasheet both times are rated for 100ns. Anyway the timer is most definitely NOT the bottleneck in my supply.
Could it simply be that i'm using a transformer intended for a bipolar transistor halfbridge supply in a 2-transistor forward MOSFET supply? I think that the funky waveform without snubbers is due to the core saturating, since i'm basically pushing 15 volts into a transformer intended for a lot less, as BJTs are driven in current not voltage, the base-emitter voltage is basically fixed at ~0.7v.
I'll wind some new transformers and see what they do, but in the meantime, feel free to post any drive speedup methods that you have seen or used. I know single transistor forward supplies can simply be driven straight from the controller, but high voltage high current MOSFETs tend to be expensive AND have high Rds(on) thus more heat, so i like the 2-transistor approach better.
In its current form my 70 volt supply runs two IRF740s mounted on an ATX PSU heatsink, no fan. They run at about 65C at 250W load. The supply is intended to do 400W but i'll be able to find out whether i will need a fan only when i'll complete the second winding as well. I hope that if i can make the fall time faster, i'll be able to get by without a fan at 400W.
I need some help on designing a fast gate driver. As you know i have built a 2-transistor forward 70 volt supply using nothing but the 555 timer chip, a LM311 comparator and a TL431 as voltage reference. Well, it was intended to be 70-0-70 volts actually but i got lazy and never wound the second 70 volt winding as i didn't really need the supply at that time (it was intended for an amplifier i didn't get around to building yet).
The supply uses a base drive transformer from an ATX power supply. However it needs heavy snubbing in the primary to produce an acceptable waveform, otherwise it goes all crazy. The two secondaries of the transformer simply drive the switching MOSFETs directly. It works quite well, but the fall time seems a little slow to me. So maybe i need something better.
I'll have to check again what "a little slow" means in that circuit, but the 555s that i am using clock in at 100ns rise time and 20ns fall time. That is quite respectable given that in the datasheet both times are rated for 100ns. Anyway the timer is most definitely NOT the bottleneck in my supply.
Could it simply be that i'm using a transformer intended for a bipolar transistor halfbridge supply in a 2-transistor forward MOSFET supply? I think that the funky waveform without snubbers is due to the core saturating, since i'm basically pushing 15 volts into a transformer intended for a lot less, as BJTs are driven in current not voltage, the base-emitter voltage is basically fixed at ~0.7v.
I'll wind some new transformers and see what they do, but in the meantime, feel free to post any drive speedup methods that you have seen or used. I know single transistor forward supplies can simply be driven straight from the controller, but high voltage high current MOSFETs tend to be expensive AND have high Rds(on) thus more heat, so i like the 2-transistor approach better.
In its current form my 70 volt supply runs two IRF740s mounted on an ATX PSU heatsink, no fan. They run at about 65C at 250W load. The supply is intended to do 400W but i'll be able to find out whether i will need a fan only when i'll complete the second winding as well. I hope that if i can make the fall time faster, i'll be able to get by without a fan at 400W.
Comment