From: email@example.com (carltons)
Subject: Re: When does a filter become an oscillator?
User-Agent: NewsWatcher-X 2.2.3b2
NNTP-Posting-Date: Fri, 18 Oct 2002 00:32:52 GMT
Date: Fri, 18 Oct 2002 00:32:52 GMT
In article , "Dave VanHorn"
> > > Are not the general criterea for oscillation:
> > > 1) a gain greater than 1; and
> > > 2) the absolute value of the phase shift greater than 90 degrees and
> > > than 270 degrees?
> Correct, but remember: If it's not working, either you designed it wrong,
> or you built it wrong.
> Don't argue with the hardware! :) Spice and such tell you what the ckt
> should do, if the circuit is built as designed, but the hardware tells you
> what it IS doing, as built.
> Look for a ground that isn't, a connection that dosen't, or a connection
> that shouldn't be there.
> If you're using a white board, one easy thing to miss is that the long rails
> don't go the whole way, you have to jump them across in the middle.
Good idea to check the layout. I would also see if other models may be
better for the opamp if you're doing a simulation. Take your present
opamp model and put it in a simple circuit comprised of a voltage follower
driving a 2000 pF cap in parallel with a 1 Gohm resistor and use a step or
square wave function on the input to stimulate it. Use the same supply
voltages that you would use for your circuit and if you don't have as much
caps to drive as the 2000 pF then put it down to what you're trying to
drive. Look at Vo on the leading edge of the waveform using Gear as your
solution for time step. You can even give the opamp a break if you
normally operate it with a single supply. Use a + and a - supply in the
simulation, because you can. It doesn't cost anything. :-)
Hope this provides some help.