[Arzela]

How can you tell if your opamp (opa 548 in a Gilmore
power supply circuit) is oscillating if
you only have basic (Fluke 177) DMM?
Is there an easy test or measurement that would show such problems?

I'd rather not test by listening to the output of an amp...

 

[jeffreyj]

Quote:

Originally posted by Arzela How can you tell if your opamp (opa 548 in a Gilmore power supply circuit) is oscillating if you only have basic (Fluke 177) DMM? Is there an easy test or measurement that would show such problems? I'd rather not test by listening to the output of an amp...

If an op-amp is capacitively loaded and its feedback network is not compensated for such, then it *is* oscillating.

But, how can you tell without a scope? There's a couple of ways....

Firstly, you can simply "measure" the temperature of the op-amp while it is unloaded; if it's getting hot just sitting there, then you know it must be oscillating.

Secondly, you can capacitively couple a 1N4148 rectifier to a small value polypropylene capacitor (1uF or so) and see if there's any DC voltage across the cap This is simply coupling the AC component of the op-amp's output to a half-wave rectifier/filter. Load the cap with, say, 100k to make sure no "soakage" in the dielectric fools you because the presence of more than a few millivolts of DC voltage means the op-amp is producing an AC output with no AC input (besides the residual ripple from the prior linear regulator which, of course, should be something like 100dB down).

 

[jamont]

Quote:

Originally posted by jeffreyj If an op-amp is capacitively loaded and its feedback network is not compensated for such, then it *is* oscillating. But, how can you tell without a scope? There's a couple of ways....

OK, what should you expect to see if you DO have a scope? (putting on dunce cap...)

 

[ppl]

by jefferyj "Secondly, you can capacitively couple a 1N4148 rectifier to a small value polypropylene capacitor (1uF or so) and see if there's any DC voltage across the cap This is simply coupling the AC component of the op-amp's output to a half-wave rectifier/filter. Load the cap with, say, 100k to make sure no "soakage" in the dielectric fools you because the presence of more than a few millivolts of DC voltage means the op-amp is producing an AC output with no AC input (besides the residual ripple from the prior linear regulator which, of course, should be something like 100dB down)."

This method works real good i use it myself to do spot checks while prototyping. Also the Fluke is real good at rejecting DC when in the AC mode so prior to using the above method try measuring with the DMM set on AC 200 mV scaleallow reading to stablise for a min or two.

 

[jeffreyj]

Quote:

Originally posted by jamont OK, what should you expect to see if you DO have a scope? (putting on dunce cap...)

No dunce cap needed, jamont; it's easy to miss, especially if the oscillation is high in frequency but low in magnitude.

What oscillation will look like depends on a number of things, but a few of the more common indicators which are most easily seen when driving the amp with a low frequency square wave (ie - 1-10kHz) are:

1. overshoot at the leading edges - mild instability.

2. ringing at the leading edges (ringing is a damped sinusoid, or one that progresses to zero over a few cycles) - moderate instability.

3. ringing that persists until the trailing edge - major instability/outright oscillation.

4. the flat part of the square wave is indistinct, or fuzzy - strong oscillation at a frequency 100x or more than the square wave.

5. there is no distinct square wave, just a positive fuzziness and a negative fuzziness - destructive oscillation, most likely regenerative; you'll probably be smelling smoke real soon...

Hey - where's the avatar pic, jamont? Going in cognito on us?

 

[aos]

Doesn't the power consumption also go up if the thing is wildly oscillating?

 

[tangent]

Yes, but it's only easy to check that in small things like headphone amps. For a DC power supply, getting the ammeter inline with the AC input is tricky.

I built a "short circuit" years ago for this kind of thing: cut a short extension cord in half, run both ends into a small project box, hook all the wire ends back up except for one pair, which go to a pair of banana jacks. Use standard banana-to-banana jumpers to hook it up to your meter. It might be more work than you want to go to for just this one project, but you can use it on lots of other things. Like finding out how much power your PC really draws from the wall.

 

[jamont]

Quote:

Originally posted by jeffreyj Hey - where's the avatar pic, jamont? Going in cognito on us?

I decided to retire Pedro until next year (thanks Grady

), but I couldn't find the other one - I expected head-fi to have it tucked away in my profile stuff, but no such luck...

And thanks for your informative comments on oscillation!