This file shows data on an experiment I performed on self discharge rates of NiMH cells.
The cells had been used in a photoflash unit as well as an LED lantern (the RiverRock LED lantern from Target) over the course of months.
Several incidents occured where one or more of the cells in the pack had reversed voltage due to depleting before the other cells, and yet still under load. This situation results in current flowing through the cells in the normal discharge direction, but with the cell voltage reversed resulting in electrochemical damage.
There is a great deal of uncertainty amongst NiMH users about just what effect temporary bouts of cell voltage reversal have on cell performance. It should be fairly clear from the data that 3 of the cells are experiencing very high self-discharge rates. However, notice that their initial capacities are very close to normal.
This is a very crude test which of course doesn't quantify the degree of voltage reversal experienced by the cells which are performing badly. In fact, it is not even known for certain which of the cells were the ones which experienced voltage reversal. Though it is a fair assumption to say that cells B and D probably were abused the most, with cell C not quite as bad, and cell A being the one that held up under load and may have never reversed.
It would be nice to have measures of the actual remaining capacity in the cells at various points through the discharge curve.
My experiences with NiMH cells have been disappointing in general. It seems that with multiple cell packs in consumer devices meant to work with AA alkaline batteries, there is no way to guarantee that cells will not experience voltage reversal. They most often do, unless one only operates equipment for short periods before recharging cells, thus never realizing the full capacity of the cells. This might still be preferrable to the costs of primary cells in some cases, but it is not ideal.
I am convinced that these cells can only be properly used in series packs with some sort of intelligent supervisory electronics. For small packs of a few cells it may be sufficient to cut off the load when the pack voltage has decayed to some value. Though there are uncertainties about temperature with this approach. The best way is to have a circuit that monitors each cell voltage individually and cuts the load before any single cell collapses.
The only bright side I can see is that cell capacity doesn't deteriorate from voltage reversal so much as self-discharge rate is compromized. I have experienced this as devices working fine with the freshly recharged cells, but after a little more than one week, the device shows weakened performance. Of course, as soon as the device is used with one or more weak cells, it is those cells which get reversed again. So the process is self-reinforcing.
I wonder if a carefully matched set of NiMH cells would avoid this? Probably not. And who has the time to match them?