From: Winfield Hill
Subject: Re: Problems with MOSFET drivers
Date: 15 Nov 2002 15:52:23 -0800
Organization: Rowland Institute
X-Newsreader: Direct Read News 4.10
Mike Poulton wrote...
> Winfield Hill said:
>> Comparators with hysteresis no doubt will. BTW, we should ask,
>> what's the driver repetition frequency, FET load capacitance,
>> and the supply voltage for the drivers?
> Frequency is 550-650kHz, depending on output terminal capacitance.
> It will drop quite a bit more once I get a decent toroid on there.
> The total gate capacitance for all the fets combined is 8000pF.
> They are IRFP350LC's. The fet drivers were getting 12V from a
> computer power supply. This gave 18V peak on the mosfet gates
> through a 1.5:1 ratio transformer.
Hmm, the IRFP350LC is actually over 3000pF at low voltages,
according to their curves. Whatever.
OK, the transformer, if perfect, makes the FET capacitance look
like 1.5^2 x 8800pF = 30000pF. That's the capacitance you'll be
driving to 12V and back at 650kHz. The gate power required is
1/2 CV^2 f or about 1.4 watts, not too bad. However, bad news.
Look at the Supply-Current vs Capacitance curves in the TC4421
data sheet, page 5, and you'll see that for Vdd = 12V and Cload
= 30000pF the current begins to skyrocket at high frequencies.
For the 632kHz line it has gone off the chart above 180mA. If
we assume its about 230mA, that's 2.8 watts, all of which is
dissipated in the driver. :(
Now examine the Supply-Current vs Frequency curves, and imagine
a curve between the 22,000pF and 47,000pF curves. Their curves
stop at 200 and 600kHz respectively, so our imaginary curve must
stop near 400kHz. Clearly you're operating in a past-dangerous
region for these driver ICs.
Hmm, why do you feel the need to drive the FET gates to 18 volts?
The FETs are well on at 10V gate drive (IR specs Rds(on) at 10V).
Also, why not use four transformers and drivers instead of one?