I'm not into circuit design on this scale,
but looking at the schematics, it looks like
the battery backup is powering the USB
controller (which does the processing), the
transceiver section, and the power-line
interface.
I don't see any need to power the
transceiver board or the power line
interface, UNLESS the CM15A is smart
enough to keep track of commands received
during that time.
The batteries are fed through a low-loss
diode, and THEN into a 5-volt regulator.
I don't know enough about the regulator used;
If I remember correctly, SOME regulators
would cut out if you didn't feed it enough
power ( Output + .5 volts )
If this regulator drops out below 5.5 on the
input, then this could be the source of the
battery problem. Anything less than a set
of fresh batteries might brown out the
whole module. But this is just speculation.
Honestly, I'm not sure why they chose to
route the batteries through the regulator.
A 1N4001 has about a .6 - .7 voltage drop;
they could have fed the batteries through
one or two IN4001's (unregulated) to the
OUTPUT side of the regulator. The USB
controller and the other chips on-board
will take up to 5.5 volts on the input. . .
I might have to try that . . . .
T