Originally Posted by
RossM
Possibly not a waste of time - I'll bet you tested the battery without any load??
If you measure the batteries with a 150 ohm resistor in parallel to provide some load, you will possibly see a significant difference.
Also I omitted to ask if the PC is powered on most the time, or if it is powered off when not in use. The CMOS battery should not affect the clock while the PC is powered on as the RTC would be using the PC power supply. If its losing time when the PC is powered on for days on end, there are a plethora of possibilities. Anyway, the fix I proved should sort that out.
Given the age of the machine its possible the there is a capacitor in the RTC circuitry that is degrading, or some issues with stray capacitance around the RTC oscillator that is having an impact. Sadly, second law of thermodynamics gets to everything eventually (including us) - damn that entropy!