RM> So any Darlington pair is a potential HF oscillator?

 BL> Not Darlingtons so much, which is the reason for their
 BL> popularity. It depends on the configuration. Adding a funny
 BL> emitter circuit can set them off.

 RM> More info? This converter had a current limiter looking at one
 RM> of the 3055 emitter resistors, but that was one of the first
 RM> things I disabled.

  An emitter-coupled oscillator... only needs a capacitance to earth
and a high-Q inductor which a long wire can provide. The base-emitter
capacitance then does the rest. I've seen power supplies with a choke
in there too. 

  Your problem is to explain the broadcast-band 1MHz oscillation. This
needs hundreds of microhenries or thousands of pf's, and it is
therefore unlikley to be layout, and more likely to be an R-C
oscillator with an unfortunate conjunction of the poles.

 RM> I'll say it again: the unstable circuit was a TIP41 driving
 RM> three 3055s with emitter resistors. Adding a 10nF cap from the
 RM> TIPs collector to its base killed the oscillation. The opamp
 RM> that drove the TIP through a 470 ohm wasn't the culprit. Where
 RM> was the positive feedback coming from?

  ROFL! If you are hiring me as a consultant, I will need the actual
thing in my hands. Other than that, I'll have to continue to speak in
generalities (at a much lower hourly rate). As I said originally, I'd
bet that there is an inherent instability in the design, and that it
is probably load-related.

 RM>  I wonder if a cap across the 470 ohm resistor between
 RM> the opamp output and the TIP41 base would have done any good?

  I thought you said the op-amp was out of it? 

  Finding these poles and loops is not easy.

  Years ago, I bought Pye a vector voltmeter (it cost the earth) when
we needed to design the chroma amplifiers and get the phase right for
colour. It worked from audio up to 1000MHz, so I ran it through one of
our production hi-fi amplifiers for the hell of it!

  Whoops! There was a pole at 1MHz - right on the verge of
oscillation!

  This thing had been in production for years and never caused any
trouble by the grace of good luck. So... with my new toy I set about
fixing it and it was a total bastard! I could move it all over the
place with little capacitors and things, but what it *really* needed
was one of the amplifier stages removed (which was not possible
unfortunately) or a nice big cap that totally stuffed the amplifier!

  It was not layout related...

  In the end I jsut fiddled everything along a bit and went back to my
TV set with a wave of my arm.

loop:

 RM> Poor layout includes poor ground layout. Poor ground layout can
 RM> cause hf instability. Ask any digital designer. Very poor
 RM> ground layout can cause audio instability.

  I *am* a design engineer. 

  Bob?
  Yes?
  Do earth loops cause oscillation at 1MHz?
  No.
  Roy says they do
  Tell him to ask a design engineer

goto loop;

  Your problem seems to be that you can't tell the fucking difference
between an oscillation at 1MHz and one at 10-100MHz. They are
different, dicko. And they change again, once you go above 150Mhz or
so, when silicon devices reach their Ft and drop their guts suddenly.

  Ask a design engineer.  

 BL> I suppose you've confirmed this by fixing an oscillating amp or
 BL> DC-DC converter by changing the layout, and then changed the
 BL> layout of a good one to make it oscillate? Pig's arse, you
 BL> have!

 RM> Yeah, and I've fixed floods by asking God to change the weather
 RM> pattern. Get real.

  But *I* have. I spent a whole week on the hi-fi pole, desperately
trying to get out because I was supposed to be doing the TV set, but
unable to leave because it might blow up in production at any moment,
even after years. It's a whole different attitude that has been lost 
to Australia because no one makes things in quantity any more, where 
a few hundred hours on a problem is money well spent if it avoids 
problems in production.

 RM> I recall you suggesting a cause. I don't recall you suggesting
 RM> a cure. Real helpful.

  You've *got* the cure. Slap a bloody-great capacitor across the most
likely collector-base or base-emitter. I was merely pointing out that
layout was probably not the cause... that whatever it was, is probably
inherent, and a genuine design fault.

  The only problem with a large capacitor from collector-base, is that
you can end up charging it up on transients, and the whole amplifier
shuts down for a while. This is not so important in a power converter,
but I would have been inclined to use 1000pF or so, rather than 0.01. 

 RM> 1n probably would have worked, but I was running out of
 RM> patience. To do it right, I should have found the lowest value
 RM> that could keep it just stable, then stick in one about 10
 RM> times bigger for a safety margin.

  That's exactly right.

  You got up my nose, Roy, so I put on my engineer's hat for the
broadside. I was actually agreeing with you. Most techs are nervous
about modifying something rather than fixing it, and I meant to say
that I basically agreed that you'd done the right thing...

Regards,
Bob
 
___ Blue Wave/QWK v2.12
@EOT:

---