X-Mailer: Mozilla 4.79 [en] (Win98; U)
Subject: Re: Problems with MOSFET drivers
References: <3DD53525.FDFAE50E@ieee.org> <3DD6840C.289DCA3C@ieee.org>
NNTP-Posting-Date: Sat, 16 Nov 2002 22:37:04 GMT
Organization: AT&T Broadband
Date: Sat, 16 Nov 2002 22:37:04 GMT
Winfield Hill, Mike wrote:
Hello Win, this guy is a serious doubting Thomas, isn't he?
>> ... I'm seriosuly considering ... adding 10 ohms in series with
>> each gate. I would be leery of using more than 10 ohms because
>> of the reduced rise time. ...
> I had assumed you were using some series gate resistance, it's
> customary to have some, because it moves much of the switching
> dissipationj out of the driver and into the resistor.
Bingo. And it can prevent ringing induced, multiple switching
transitions of the driven MOSFETs as well as damp out reactive
gate current before it has a chance to ring back into the driver
outputs, possibly causing all kinds of trouble.
> You might try about 1.5 ohms, 2W.
With his gate drive transformer this value would probably be way
too small. The resistance should be around one half to one times
the square root of L/C, where L is the series gate inductance and
C is the gate capacitance. I'm guessing that he could stand to
optimize his drive transformer quite a bit by reducing turns all
around and changing the turns ratio to 1:1:1:1:1. This would trade
off core losses for leakage inductance and allow the use of smaller
series gate resistors.
At some future point, when and if everything is working, he ought
to try increasing the gate resistors until he just starts to notice
an increase in operating temperature of the MOSFETs. He should
then back off from this value by a factor of two or so.
> You could also try a pair of Schottky diodes, one to each driver
> supply rail at the driver's output, to see if you are suffering
> from improper reflected voltages.
Already suggested that, but I don't think he took it to heart.
Adding such diodes would be good insurance, but their lack is not
at the heart of his problem which seems to be a combination of
excessive drive inductance and insufficient damping resistance
coupled with a seriously malfunctioning self-oscillating tesla
coil feedback circuit. But I think he'll come around 'cause
there is no authority on what works and what doesn't like the
breadboard. :) -- analog