[sdiy] Prophet 10 op-amp swaps?

cheater00 . cheater00 at gmail.com
Wed Jul 31 21:47:45 CEST 2013


Hi Paul and John,

On Wed, Jul 31, 2013 at 8:41 PM, Paul Schreiber <synth1 at airmail.net> wrote:
> Dave is not the one trolling here.

if you have any information relevant to the technical topic at hand,
that would be great, but quips and unsolicited aggression seriously
don't belong in a discussion on synth-diy. Dave flew off on a tangent
with unsolicited aggression and his email wasn't leading to anywhere,
and it wasn't the first or second time this happened, so I felt
compelled to answer. I think it would have been better to just have
ignored it, I'm sorry. Please let's stop going in this direction right
now, before Eric's thread gets derailed further. Let's also not talk
about the angle between Dave and myself.

I contend that new op amps put in the place of very old op amps can
start oscillating. If you have experience to the contrary, that would
be an interesting topic of conversation. Perhaps it depends on the
kind of circuit. But this thread is for Eric's project. I'm curious as
to what his scoping will show up. In my experience, although likely
not as extensive as yours, it can just as likely show the output is
perfect or that there's an issue. I have found that at my level of
understanding it's not always immediately obvious from the surrounding
circuit that a certain part will start oscillating, so I like to give
op amps an additional check. It's not time consuming and if one has a
scope it's very simple, but a high-pass filter and AC volt meter are
good too.

While writing the above, I have received Eric's email.

On Wed, Jul 31, 2013 at 8:58 PM, Eric Frampton <eric at ericframpton.com> wrote:
> D -
>
> I've now swapped out all the 356's except for one in the Lower EQ out.
>
> I've got a CRT scope, 30MHz.

Cool, what model is it?

> Sure 'nuf, you called it. There's a super-high-frequency oscillation happening at the Upper EQ out that isn't happening at the Lower one. You can't hear it, and it doesn't register on VU meters, but you can see it on the 'scope. It seems to only be happening at that one stage; the final outputs do amplify it (no rejection, darn!), but they don't appear to be adding any problems of their own.

Theoretically, you should be able to hear it further on, because the
oscillation will get mixed ("mixed" in RF terms, as in "frequency
shifted") by the nonlinearites down the chain, pulling it slowly into
the audio frequency (even if not in one step, e.g. 10 MHz -> 5 MHz ->
2 MHz -> and so on). If you plugged such an oscillating circuit into a
VCA with a very fast envelope, the click might change timbre somewhat
(easier to hear if you put the click through a very long reverb). But
of course it's easiest to hear that if you create a simple frequency
shifter. Maybe it would be a cool project for me too, now that I think
about it. Being able to not only see but also hear the parasitic
oscillation while you're trying things out could be a great
improvement.

> So, aside from building an active filter circuit (which I'd rather not get into), might this be solved by a well-placed bypass/filter cap, or maybe a not-quite-as-fancy op-amp (though I don't want to drop that 356 back in there)?
>
> e

Hi Eric,
Unfortunately, I can't tell you what the best way is, simply because I
don't have enough experience in analog design yet. You might want to
use a small cap in series with a resistor in the negative feedback of
the op amp. Or if the amp is inverting or it doesn't help for some
other reason try the positive feedback. A bypass cap might help as
well. Power supply decoupling was suggested to me once. Sedra &
Smith's "Microelectronic Circuits" has extensive coverage of feedback
over several large chapters, but I haven't had the option of spending
enough quality time with it yet. Great book, see if you can find one
if you're interested in this sort of thing. Starting page 843 in the
3rd edition, (12.1, subtitle "The Oscillation Criterion") they talk
about the basic idea of why things end up oscillating, and then
there's some theory on how to control oscillation, but they don't talk
about parasitic oscillation. I think this book doesn't talk much about
the practical issues so much.

Here's a resource with some ideas. I've found it after a cursory look:
http://tangentsoft.net/audio/opamps.html - here the author talks about
various op amps in audio use and notes some of them oscillated in his
designs
http://tangentsoft.net/audio/hs-opamp.html - here he talks about
parasitic oscillation and how to prevent it. He mentions limiting the
bandwidth of the feedback loop, and PSU decoupling. He also mentions
high DC bias on output. That's not something I thought was a major
issue, but apparently is worth checking. Consider setting your scope
to DC and checking the voltage right after the op amp. Use a local
ground (possibly the op amp's negative supply if it's not bipolar) for
the clip.

On Wed, Jul 31, 2013 at 8:58 PM, Fast TriggerFish
<fasttriggerfish at gmail.com> wrote:
> I would have thought that if the op amp was self oscillating at a high
> frequency then you would hear *nothing* or at least the performance would be
> very poor.

Hmmm, I don't know. I don't understand op amps fully, but in my head,
when there's no input, the incoming high-frequency oscillation from
the feedback loop is allowed to request all of the op amp's power to
sustain itself. Even at relatively high power, though, it's going to
be of small amplitude (I guess). However, as soon as something
lower-frequency comes in, it easily overtakes the oscillation and
becomes "more important", until it's gone again. This is probably the
fuzziest and wrongest explanation you're going to get, but it's my
understanding for what it's worth. Try not to keep it in your head.
This is how I've noticed many filters operated that were set just
around self-oscillation. Left alone they would self-oscillate but if
you added sound input the oscillation would change frequency or timbre
or completely disappear and yield to the incoming signal.

Maybe Brian can say more about why oscillating op amps still amplify.

Brian,

On Wed, Jul 31, 2013 at 9:20 PM,  <rsdio at sounds.wa.com> wrote:
> I recently designed a 32-channel DAC for ultrasonic frequencies

really curious, 32-channel ultra-sonic DAC - are you allowed to tell
what the application was? I imagine an array of transducers, possibly
for measurement purposes, relating to mechanical modes of rigid
bodies. Alternatively, a software radio but those normally just use a
single output DAC for whatever transmitter you've got (usually just
one!). I'm also reminded of a recent experiment I saw in ultrasonic
levitation, the parts for that are less than $100, but that doesn't
need a DAC, so this would have to be much more sophisticated...

Cheers,
Damian


More information about the Synth-diy mailing list