Good question, no real answer.
Consider: The trim on the xmitr is there to bias the signal throught the pot on that axis of the joy stick, but the question is why? In your case (and in mine the right stick trim is althe way up to stop reverse motion with zeroed stick), so , either the xmitr is sending too strong a signal (to many encoded pulses) for the resistance in zeroed position, or, more likely, the h-bridge in the driver circuit for the motors is out of balance. They use cheap electronics, which could mean leaky transistors (quite common).

I'm thinking, maybe the cheapest tac to start with would be to try another driver board. The transistor bridge is a simple circuit, but for me, without schematics, fixing that would be quite a task.

No help, but maybe food for thought?

splat