From: John Larkin
Subject: Re: Gate drive transformer ferrite?
Date: Sat, 19 Oct 2002 18:37:51 -0700
Organization: Posted via Supernews, http://www.supernews.com
References: <3DB09F8E.4070509@BOGUS.earthlink.net> <3DB0A0B8.DCC38BFD@rica.net> <email@example.com> <firstname.lastname@example.org>
X-Newsreader: Forte Agent 1.91/32.564
On 19 Oct 2002 15:50:28 -0700, email@example.com (Winfield Hill
>John Larkin wrote
>> my knowledge of this is almost all experimental. ...
>> Where this kind of transformer is really useful is with gaasfets. ...
>> The bane of my life (kids excepted) seems to be translating very fast,
>> small-voltage-swing signals into very fast, large-swing thingies.
> It sounds like you've done well. Using straight-forward techniques
> (common-base cascode level-shifter, balanced emitter-follower drive)
> I developed a 2.5V to 12V 1A FET driver with 5ns delay and risetime,
> which was a fun exercise and solved a 500V accelerator deflection-
> pulse problem. One irritating issue, a small slip of the scope probe
> and BANG, instantly burnout five painfully-installed components.
> So I find your GaAsFET stories impressive and highly entertaining.
>> Hey, the sun came out! That's novel around here. Maybe I'll go out
>> for a latte.
> Have one on me, virtually that is. :-)
> - Win
You would think that enough people need fast gate drivers that we
shouldn't have to cobble them out of discretes. I guess what we need
(several amps, 20 volt swing, few ns rise/time) must require a semi
process that doesn't exist. As the speeds go up, the voltages keep
Maybe I'll try just a PNP on the high rail and an NPN pulling to
ground, with mostly-AC coupled drive. Some of the Philips fast
high-current (the GHz Ft guys) transistors seem to come out of
saturation almost instantly, almost like a fet. Surprised me, they
did. They could be driven directly from a fast CMOS logic gate. The
same could be done with complementary fets, given the right parts.
Your cascode driving complementary emitter followers might not work
for me, as I only have an 18 volt supply and I want all of that to go
into the switching fets; they seem to die horrible and noisy deaths if
they're not driven hard.
I was about to suggest this (L stores the turnoff current) but then
realized the gaasfet couldn't stand the voltage. What a waste of ascii
v+--------------- ------------------| mosfets
| | | | |
| T || T | |
| p || s R |
L r || e | |
| i || c _ |
| | | ^ diode |
IN -------------| gaasfet
I wonder if anybody has tried using the LMOS RF parts in switching
apps? Most of them are 'tuned', but maybe that can be tolerated.