''special'' mosfet driver...

[fredos]

Back again!

I need to drive a pair of (power!) P and N mosfet, source to ground, from LM319 comparator at + and - 10V. I already try 1.5K pull up resistor and a pair of PNP and NPN transistor as buffer, but near maximum duty cycle, this seem unsuficient...I try a TC4420 ( at 20V), this work very well, but its over the maximum rating of manufacturer....Anybody has already do some discret and SIMPLE driver that can provide + and minus 10 to 15V to drive mosfet?

Thanks

Fredos


Yeah it's for a class d amplifier!

[Workhorse]

Fredos,

Did you tried using IRFD9110/110 complementary mosfets in Common source configuration to drive the large power mosfets!!!

Amazed to see a guy capable of designing KW level PWM amps is searching here on this forum for drivers..........

Kanwar

[alfsch]

hi fredos,
a year ago i had this in mind: (pic)
you think about something like this?

for small mosfet, as trans i tried: bc516/17 , 1A darling

for bigger gates, komp.fet driver: irf7389 , 30v, 30a peak

monster , fast driver: tc4422

[fredos]

Thanks for reply...

alfsch

That's exactly the way I have try, but at 250Khz with a LM319, at maximum duty cycle, I can't drive correctly mosfet...It work very well with a TC4420, but I'm over maximum rating of IC....

Workhorse

I have already try the source follower, but with voltage drop I can't bring power mosfet in full saturation...I have try common drain configuration with reverse input on my comparator, but introduce too much delay in rise and fall time....I'm looking to design a class d with less than 100ns response time...just the LM319 introduce nearly 40 ns, TC4420 35 ns, and mosfet itself nearly 40 ns....So with TC4420 solution, I'm out of my spec...With a pair of PNP and NPN driving transistor, I can reach my goal...But prob at maximum duty cycle....


At least, I will simply limit duty cycle and raise PSU voltage to keep same power....

Thanks

Fredos

[lumanauw]

Hi, Alfsch,

Interesting design ClassD version of Transnova. How to adjust the dead time in your design?

[Eva]

I don't understand how a direct coupled gate-drive circuit can exhibit duty-cycle-dependent performance.

[Workhorse]

But what exactly is the problem when driving with full duty cycle........

[mzzj]

Originally posted by fredos
class d with less than 100ns response time...just the LM319 introduce nearly 40

Isn't LM319 a bit slow if you want less than 100ns total response time?

And please try to be more specific what is exactly your problem with high dyty cycles?

[alfsch]

#lumanauw: depending on gate-drive /resistors you get a slew rate , about 0,5v/ns at the gates; from +4 to -4 v , one fet closing, other gets on, about 16ns "dead-time" ; using res+diode at gates you can adjust the dead-time even more;

and fets switch off faster , because negative driven

next is, because there is some stray ind. between the windings of the fets, the shoot, if any, is much smaller...

the problem is, you need n + p fets, so its only for small amps, lets say max. +/- 40v, with 100v fets like irf540/9540

[lumanauw]

Hi, Alfsch,

Nice Did you eventually built it? I noticed something. The feedback is taken after the LC filter. Is this UcD?
The parameter difference between Nch (IRF540) and Pch (IRF9640), will this be a problem in real cct?

Hi, Fredos,

Q4-Q5=common collector bipolars, driving Q2-Q3=common source mosfets, and the drains are driving Q1 (the big mosfet)