The
article linked to by SilverFox describing an inflection method for charge termination on NiMH cells made me curious enough to do an experiment of my own to see how it might work out.
For the test I used a "Chicago Electric" cell of 2000 mAh label capacity, which I measured at 1700mAh actual capacity using break-in mode on the Powerex MH-C9000 charger. This cell is from a set obtained at $3.49 for 4 at Harbor Freight, so I didn't feel too upset about torturing it in the pursuit of science.
The test itself consisted of charging the cell at a rate of 1600 mA (0.95C) and taking regular readings until the C9000 terminated the charge and showed "Done".
Since I lack sophisticated test equipment I had to improvise a bit. I measured time, voltage, and charge by reading the C9000 display, and I measured temperature using a thermocouple sensor attached to the wall of the cell and covered by a layer of paper and aluminium foil for insulation purposes. The C9000 only displays down to 10 mV resolution where 1 mV would have been ideal, but I was able to get a reasonable trend line by smoothing the data using a rolling average.
The first chart (below) shows the trend of voltage and temperature. Charging terminated after 82 minutes at which point the cell temperature had reached 51°C and the total charge input was 1971 mAh. After an 8 hour rest for the cell to cool down, a discharge test at 500 mA give a measurement of 1586 mA. This gives an estimated charging efficiency of 1586 / 1971 = 80%. We can also note that even at the -dV cutoff point, the cell was left apparently 100 mAh short of a full charge (unless the self-discharge over 8 hours was abnormally high, which needs another test not yet done).
The second chart shows the slope of the voltage curve (smoothed, to show the trend). It can be seen there are two inflection points, and it is the second one we are interested in. This occurs at a charge of about 1600 mAh, and this is where the charge would terminate if using the inflection method. Comparing with the first chart, the temperature at this point was 41°C, so ten degrees cooler than the maximum reached at -dV. If we assume the same charging efficiency of 80% applies, the actual stored charge would have been 0.8 x 1600 = 1280 mAh, now over 400 mAh or 25% short of a full charge.
Conclusions? Well, this is only one test on one cell of course, and it might not be the best quality cell either. On the other hand, real chargers cannot pick and choose which cells people will try to charge, so they need to be able to deal with whatever is thrown at them. Given that, it does seem that the inflection algorithm can be implemented and does provide a termination signal that can be detected.
With this particular cell the temperature rises throughout the charging process and therefore the cell still gets quite warm by the time the inflection point is reached. It also seems that the cell will not be fully charged at this point, so any algorithm of this nature would need to follow up with a significant top-off charge to ensure full charging.
In the future, if time and enthusiasm permit, I might repeat the experiment with an Eneloop. I do know, however, that the C9000 will not reach the -dV signal when charging an Eneloop, so it might actually terminate before the inflection point is reached. I suspect the low internal resistance of an Eneloop will lead to lower temperatures during the early part of the charge cycle. I also suspect the charging efficiency will be higher than 80%.
