Reply-To: "fred bartoli"
From: "fred bartoli"
References: <3DD39D90.firstname.lastname@example.org> <email@example.com> <firstname.lastname@example.org> <email@example.com> <firstname.lastname@example.org> <email@example.com>
Subject: Re: zero-power toggle circuit; was, how to master electronics
Date: Wed, 20 Nov 2002 20:34:57 +0100
X-Newsreader: Microsoft Outlook Express 5.00.2314.1300
Organization: Guest of ProXad - France
NNTP-Posting-Date: 20 Nov 2002 20:32:18 MET
Terry Pinnell a écrit dans le message :
> Tony Williams wrote:
> > Terry, I see that the old Ferranti transistor
> > equivalent of this circuit had an extra capacitor
> > in it..... across our R3, the g-s of the 2N700x.
> > I suspect this was added to avoid triggering-On
> > due to noise glitches (as power up). Because
> > the gate of the 2N700x has to be reliably pulled
> > to 0v by C1, this new cap should be no more than
> > 1/100th of C1, say 1nF or less. Try it?
> > Note also, this circuit will be reluctant to
> > turn Off with light loads (your 100 ohm) and
> > possibly with capacitive loads. It should be
> > 100% reliable with inductive loads (relays,etc).
> > Load-sensitivity is a natural consequence of
> > taking any feedback directly off the load.
> Thanks Tony. As you know, dodgy CM MOSFET models remains the most
> significant factor confusing my simulations. But I've tried
> experimenting with an extra C across R3, ranging from 500pF to 5nF,
> and reducing load to 10 and then 1 R, but no change.
> One change of technique did prove successful. I thought I'd try
> ramping up the 10V supply, instead of applying it instantaneously. Now
> that I've reduced R4 to 10k, only a few uS ramp proved necessary to
> prevent some types switching on at power up. I'll do that as a matter
> of course from now on, for sensitive circuits.
I don't know CM, but it should be the same as for generic spices.
It's not that you have a sensitive circuit in your case but that you have a
dual state one.
The initial bias point can be whatever of both states, depending how the
solver makes its way through the solution.
The best answer to your *initial* bias point problem is to help the solver
.NODESET control card that set the (some) node(s) into the right bias region
at initial biaspoint computation,
.IC that sets some initial conditions (caps and inductors)
You also have some control words (OFF) that defines the initial states of
semiconductors. Unfortunately the power mosfets models are macro models so
this option is not available.
You use the rampup option when you have convergence problems at the inital
bias point. The reason it works is that the transient solver is more robust
when it has converged on one solution, and the *all zero voltages* is an
Please note that this will change nothing to the "no switching" problem.
> Terry Pinnell
> Hobbyist, West Sussex, UK