From: "Peter Larsson"
Subject: Re: Transient problem !
Date: Sun, 12 Jan 2003 11:44:50 +0100
References: <3E1F3BA5.7202C132@earthlink.net> <3E1FD1C3.640B7E8C@earthlink.net> <3E21360B.33AD0412@earthlink.net>
NNTP-Posting-Date: Sun, 12 Jan 2003 10:45:37 +0000 (UTC)
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
"Robert Baer" skrev i meddelandet
> Peter Larsson wrote:
> > > Well, for starters, your FET is switching a load on and off fairly
> > > rapidly.
> > > So that part is like a signal generator, and (maybe) you have
> > > wires to the battery (both the "ground" wire and the supply wire).
> > > If not properly bypassed, that is more than enough to cause problems
> > > like you describe.
> > >
> > > Next is the generator (MAX 627 driver); i will assume that
> > > bypassed, and so the length of the "ground" and supply wires to its
> > > PS then are not an issue.
> > > But this signal has to travel many inches, feet, miles, furlongs, ??
> > > to the load (FET gate and source).
> > > Well, a twisted pair is better than 2 randumb wires, and a coax is
> > > even better, because the gate should be switched in less than 100nSec.
> > > BUT, a signal from a 50 ohm source, thru a 50 ohm coax cable 9
> > > long that is *not* terminated, will reflect back to the generator
> > > 2nSec later.
> > > And if the generator is not 50 ohms, it will bounce back to the FET
> > > gate, and ping-pong maybe a half a dozen times before it becomes
> > > insignificant on a scope.
> > >
> > > So, that is why i made mention of (1) careful layout and bypassing
> > > each circuit, and (2) proper termination.
> > > BTW, that FET gate is a *large* capacitor.
> > > A 10 ohm resistor (minimum) in series from drive signal to gate will
> > > help slow down the risetime a little, and limit the current drive to
> > > the gate (will not act like a transient short circuit to the drive
> > > signal).
> > > If the driver was very close to the gate, then connecting via a 10
> > > resistor (as opposed to a very short wire) would be ideal.
> > > But with a long coax, make it (say) 82 ohms and use a (say) 68 ohm
> > > termination resistor.
> > > Not perfect, but it is difficult to get good termination with a
> > > capacitive load.
> > > If this was an RF CW circuit, an inductor in series would be better
> > > (complex conjugate loading for you tube guys).
> > > For gate safety, connect a resistor from the gate to source
> > > anything from 10K to 20Megs will prevent static buildup if that driver
> > > gets removed or worked upon.
> > Well !!
> > I do not see any problems on my scope with the signal to the gate (no
> > ringing and no transient).
> > The length to the gate from the driver is ~20mm with a 10 ohm resistor
> > included. A resistor for gate safety is probably good to have !
> > The coax is from my signal generator to the input of the MAX driver, and
> > there is also the clips attached.
> > I will try to put some caps over the supply line and see what happens.
> > seems to be some problem with a stray inductance somewhere in the
> > snubber solves the problem but should it really be necessary?
> > Peter
> A snubber is used in "high" power work, to decrease current thru the
> FET clamp diode when the LC load swings below ground.
It can also be used to reduce the switched stress with voltage in mind. If
the voltage stress is lower a lower voltage rated transistor can be selected
with a lower Rdson.