Weird 42S battery problem

[pdicamillo]

The calculator passes the self-test. I made measurements with LR44 batteries, and they were draining similarly to the 357 batteries. The battery which drains is the one which contacts the negative tab (tab at the left viewed from the back.) I also noticed that with the batteries removed, the drained battery gradually regains its original voltage, eventually matching the other two. But the problem appears to be a very high current when the calculator is off. I measured 5.4ma at either contact tab, and it's consistent over time. I don't remember ever having battery leakage, and there's no sign of it around the battery compartment.

I'm reasonably good at soldering, and I think I could replace a capacitor or some other component that is a likely cause. But I wouldn't know how to find the source of the current drain.