From: email@example.com (Eric Y. Chang)
Subject: Re: Problems with MOSFET drivers
Date: Mon, 18 Nov 2002 00:03:47 +0000 (UTC)
Organization: California Institute of Technology, Pasadena
NNTP-Posting-Date: Mon, 18 Nov 2002 00:03:47 +0000 (UTC)
X-Newsreader: TIN [version 1.2 PL2]
Winfield Hill (firstname.lastname@example.org) wrote:
: I completely agree, an IC experiencing improper voltages, or high
: substrate currents, etc. will blow faster than an instrument fuse.
: The solutions include 1) knowledge and understanding, 2) analysis
: and 3) measurement (reduced voltages are suggested, approaching the
: trouble region slowly), and sadly probably not fuses. Sorry. :)
Hmm. Maybe this is why you see so few "fuses" used in power
electronics design. But I would ask why not? I had a tough time
debugging one design which ate up a bunch of power semiconductors,
but the reason why was that it was intermittent, and it would work
completely within ratings (with oscilloscope and signal generator
tests all OK) for several months. Then, it would die. I had 5
transistors in my order, and I burned through all of them before
finding an intermittent short in a coil. Funny, I whacked and
thwacked and sprayed that darn thing, and it never failed when I
was around, and I couldn't spring for that single shot capture.
But, if the circuit is failing quickly, and often, why not put
a "fuse" (actually I mean fast electronic circuit breaker) on it?
I have heard of people using small diodes, which pop pretty fast,
but they sometimes fail too slowly to protect a much larger and
more expensive MOSFET (presumably because their dissipation is
the first rather than the second power of I). But then, why not
use a small MOSFET. Shouldn't the smaller one burn first? I
didn't mean use an expensive semiconductor fuse. These are pricey,
and, according to their I2t ratings, only fast enough to protect
thyristors. Recently, I have seen some calculations with modern
switching devices that say that the fuse will blow first, still.
But, how about all those fast electronic circuit breakers? They
use isolated gate drive and adjustable turnoff based on current
level or rate of rise? People tell me that you cannot do a design
without burning through a coffee can of switches. The "fuses"
don't work? I have found that a cheap fast diode always blows
before the bipolar transistor.