Just to check my understanding of the CV:
I believe the panning works by providing negative gain when you pan away from one side, rather than by adding gain -- is that right? So for instance panning channel 1 hard left applies a 0 CV to the gain cells that sends channel 1 to the left output, and a negative CV to the channel 1 right gain cell.
The gain pots on the panel seem to be similar: they range from 0CV at the maximum to sending a negative CV (positive from the pot, but then inverted in the circuit) to attenuate the output signal.
Meanwhile the gain CV jack provides positive gain.
If all that is correct then the symptoms I'm seeing, which are very similar to Stevie's, can all be attributed to too much gain. With lots of gain, turning the gain pot down even a little bit will attenuate the signal very hard. And obviously, applying positive gain CV will amplify it a great deal too.
If so then reducing the 5k6 will take the edge off the effect, as Ken suggests.
The clipping of signals greater than +/- 2V is more troublesome but that seems to be my problem rather than Stevie's!