[sdiy] OB-8 voice glitching, warble

[Eric Frampton]

Hey y'all, thought I'd update you guys on this. I (we!) figured out this problem, but another has taken its place.

On Apr 3, 2012, at 2:41 PM, g m montalbano wrote:

> Hola Eric --
> Did you look at the key CV for that voice? U28 demultiplexer & U29 opamp buffer.
> A scope should show if these are behaving correctly.

So I pulled out my scope, and after the machine warmed up I could see some bleed through the TL084 on CV3 (the keyboard CV for that voice) as a low-amplitube sawtooth sort of thing. So then reading Bob's note...

On Apr 3, 2012, at 3:55 PM, Bob Weigel wrote:

> Hey Eric I need to brush up on them again to get a couple fixed here at long last.  This sounds mildly familiar.  So with just one voice doing this it eliminates all power supply questions, all modulation questions...all dac questions since indeed that would tend to affect all of them and not just one.  So it's got to pretty much be something (given that you already replaced relevant 4051's and 084's)... the keyboard CV?  That one is the only thing in common isn't it?  Have you looked for the signal on the lines correlating to the disturbance?  Looks like it would have to be a resistor going open or something..not likely a poly cap I wouldn't think.  What did I have a while back..one high reliability looking cap failing in a korg delta.  Someone else thanked me for posting that.  They had the same issue in a Delta!  LOL.  What are the odds of that?    Anyway  if you can find the signal that is causing the issue it would sure help.  You might be able to hear it if the rates of transition are hitting the audio range by just putting a little cap on the end of a 1/4" jack to an amp if you dont' have a scope.

...I thought, hey, there's nothing between the 4051 and the TL084 except a cap, let's try swapping that out. I had a .022mF polyester I dropped in there to test, and voila, it works.

However, during this process somehow I took out voice 5 (CV1). Removing U30 (the TL084) to take anything downstream out of the equation, I probed pin 1 of U29 (the 4051, the output for CV1), but instead of seeing a sawtooth-looking wave as in the other 7 outputs, I see a flat-line punctuated by little positive-biased blips.

Also odd: although the output on U29 for CV2/voice 6 is clean, neither voice 5 nor 6 will pass the Autotune routine now.

Another bad polystyrene perhaps, or something in the data stream upstream?

e