Thanks for the insights, especially the notes about conflicting ‘errors’. For what it’s worth, neither of my batteries pass the EMMPathy or Lind testing/programming (yes, the memory is probably faulty) but this doesn’t stop the batteries from working well enough for me - they charge and they run the machine for an acceptable time (1.5-2.5hrs depending on power saving, at least for the one with the 2000mAh AA cells: the one with smaller 1500mAh 4/5AA cells does not work well, but I think this is because smaller cells just cannot deliver the proper current or accept the rate of charge). The low-power alarm seems correct, estimated runtime seems correct, and the sleep drain is reasonable (not exactly iPad-standard, lol) in that it seems to use about 33% charge being asleep for 24 hours.
I’m lucky to have several hard drives from these machines, so I would try the sleep/wake with another drive. It’s a long shot but if the drive does not respond, the machine seems to wait. Yes, the sleep fix extension that’s part of EMMpathy is to reduce the drain while asleep, 3lectr1cPPC is right.
I traced my power supply fault to a shorted diode, which I couldn’t identify (coloured bands on glass, 1N64 germanium type perhaps but I don’t wish to risk a substitute). I’ve decided to use a 16V supply and will test what happens with a single supply rail.