[Re: thermal tips re expo converters:]

[Ian Fritz]

Harry --

I have many years of experience in laboratory temperature control
systems. I can assure you that Martin is correct, and for the reasons he
states. A system with an extremely slow cooling rate will *not*
generally achieve steady state, as you claim. Instead, it will
oscillate. The only way to stop the oscillation would be to use a very
low loop gain, which would result in poor regulation.

It seems to me that it is not easy to get comparable heating and cooling
rates using a plastic IC package. This is undoubtedly why the commercial
VCOs use such a high temperature, e.g., 85deg C for the SSM2033, 60deg C
for some Moog VCO I remember seeing, etc.

  Ian

"Bissell, Harry" wrote:
> 
> Correct, no doubt. The techniques I'm suggesting serve to decouple the
> external temp changes from reaching or disturbing the expo converter. If we
> change ambient and the converter takes a half hour to notice, we win. With
> high heat loss, or low heat loss the circuits will achieve steady state. You
> could even turn the heater power down and still stabilize the loop. Which
> factor is more significant, good thermal isolation, or ability to bleed off
> excess heat ?  Beats me... what does anybody think ? :-)  Harry Bissell
> 
> > -----Original Message-----
> > From: Martin Czech [SMTP:martin.czech at intermetall.de]
> > Sent: Tuesday, February 09, 1999 2:21 AM
> > To:   varner at k-online.com; harrybissell at netscape.net
> > Cc:   synth-diy at mailhost.bpa.nl
> > Subject:      Re: [Re: thermal tips re expo converters:]
> >
> > I'd just like to mention that you need some thermal loss in order to
> > cool the device down again. If it would be perfectly isolated, this
> > would mean that heating is very fast, but cooling very very slow. The
> > slightest regulation loop overshoot, or self heating of the expo pair
> > would need very long to settle down.
> >
> > A peak detector is an electrical analogy to this behaviour.
> > Without bleeder the control loop goes nuts.
> >
> > m.c.