Paul,
Thanks - this is very helpful. The first thing I will do next is make sure the Pulse Width/PW CVs are working correctly for the troubled VCO. A bad signal shape (like no pulse output) might screw up the auto-tune process based on your description of the digital counter. The second VCO on the same voice board seems to tuneup just fine - so this is isolated to this VCO or the auto-tune components/software for this VOC.

While I will continue to work this issue on the hardware side, I have this nagging fear that the EPROM code that runs the auto-tune process is corrupted, but that is mostlikely my due to my "what things can I not fix if they break" phobia more than anything else.

Thanks again. I will let you know what happens.

--- In oberheim@yahoogroups.com , "Paul J. White" <pjwhite@...> wrote:
>
> At 11:31 AM 10/15/2009, you wrote:
> >Does anyone have a good (engineering level) understanding on how the auto tune works on the OB-X
> >
> >I have one OSC on one voice that is getting a strange auto-tune control voltage causing it to be out of tune. After swaping around op-amps to no avail, my next guess is a bad cap supporting the voltage at the op-amp.
> >
> >However, I don't understand what goes on during an auto-tune. I assume it compares the OSC frequency to a reference signal. I don't understand who the comparison in made, or how the software turns off voices, gates the voice, etc...
> >
> >Thoughts
>
> I don't have the schematic handy, but I'm pretty sure the auto-tune process is something like this:
>
> 1. Mute all the voices at the audio output (so you don't hear the auto-tune process).
>
> 2. Set the VCAs to turn on one voice at a time. The filter is set wide open and the oscillator pulse output signal is shaped to a logic level and mixed on the OSCMUX bus which gates a high frequency digital counter on the control board. This counter is capable of measuring the oscillator period to within a microsecond or two.
>
> 3. The computer uses the digital count to determine the period of the waveform and adjusts the oscillator control voltage until the count corresponds correctly to the desired ideal frequency.
>
> 4. Store the correction voltage in memory, to be used in choosing the right control voltage for any given note pitch during normal operation.
>
> The process is repeated for each oscillator and each voice, probably across a range of at least two different control voltages in order to establish a linear relationship between desired frequency and control voltage.
>
> -- Paul White
>