Hi Malte,
You are right. The problem I have had is reproducing the problem. As I said
earlier this fault wouldn¹t show itself for hours on end. I finally realized
the thing to do was to actually play the P6 which very quickly reveals the
problem. Especially if you bash out a few chords and sweep the filters and
VCO controls as you are doing it. In my case I used manual¹ mode as the
control for the experiment. When the fault appeared bank lights A-D would
extinguish and lights 2, 4, 7 would light. What kind of patch is that?
The CPU is clearly very susceptible to noise on the clone board, so much so
than the old 367 for some reason. Even getting you hand close to the
oscillator for testing sends it into a spin. If you compare this with the
original 367 the circuitry is ultra stable. Whilst measuring the clock
input at C22 I realized that in doing this the whole CPU would appear less
likely to crash when I was probing. If I leave the scope probes in place and
then bash out a few chords whilst sweeping the filter, etc, I cannot get the
board to leave manual mode. I am guessing that as the scope probes are very
high impedance then it must be the capacitance of them which is stabilizing
the circuit. This is also the case with just the probes attached, not even a
scope! My probes are around 47pF, so that in parallel with the 22pF I am
testing would be 68pF (or as near the E series value as you can get). This
is much along the same lines as your initial experiment, but with a more
clear view of what might be happening. Perhaps we should suggest people
leave the scope probes attached before they screw down the lid! Can anyone
else think of any property that might be introduced the probes? In the mean
time I¹ll trying some more modifications to the oscillator.
Cheers,
Andy
On 02/09/2010 19:31, "Malte Rogacki" <gacki@gacki.sax.de> wrote:
> Andrew;
>
> I've reverted to my previous version now (with the 10pF cap for C23) and
> the board has been runing fine now for several hours again.
>
> The problems with that particular KLM-367 also were manifest in another
> Polysix so I doubt my case has something to do with a specific machine.
[Non-text portions of this message have been removed]