Are you sure the sound of the previous note that we hear is not coming from a different voice?
If the poly six does not re-assign the same voice for the same note, but always picks the next voice,
then maybe the issue is with voices not turning off fast enough when you release the key?
So you get the old note and the new note for a bit, then just the new note?
Maybe the sound is gated off, but the voice is still playing when it shouldn't be, then when you
play a new note, the overall sound is gated on again, and you hear the new voice plus the old voice
for a bit. But this implies that you could wait longer between notes and the problem would
go away. You said that wasn't the case.
I think a simpler demo, with you pressing keys, and also telling us exactly what the sequence
is, like press 1, release 1, press 2, etc. might be more useful. Unless the note is being re-assigned
to the same voice, my theory about the capacitor voltage doesn't work.
Also, then you look at IC31 pin 3, you are seeing the voltages for all 5 voices, plus two or three
other voltages that are used to keep the anti-log tuned correctly. Those extra levels are normal
and will not change as you play notes.
When you label your photos, please show the real vertical sensitivity, taking into account the probe.
So if the scope is set to 0.1V/div but you are using a 10:1 probe, then the setting is 1V/div, not 0.1V
Bob