Experiments never stop around here, and I’m pleased to present the tenth installment in this series of posts about the capacity degradation of a lithium-polymer based power bank as a function of charge-discharge cycle. This tenth part represents another increase of 50 cycles from the last report, for a total of 516 from new, or 500 from the commencement of the experiment. The power bank has reached and exceeded the 300 to 500 cycle lifetime commonly quoted for lithium-chemistry batteries, and thus we are now into uncharted waters.
A further 50 runs was accomplished with no loss of data, resulting in a reduction of data loss to 0.8% total. The increased variability continues to be displayed, with a less clear downward trend.
The variation seems to have exceeded 100mAh in some cases, and I have several hypotheses which could possibly explain this:
- Increased internal resistance and contact resistance, as alluded to in the previous part.
- Increased dwell time after full charge before testing, as during this test segment, testing of this power bank was interleaved with the Mi 16000mAh power bank, thus some times the power bank was sitting at “full charge” attached to the charger for a day before it was actually tested. This dwell time might have resulted in some loss of charge, which reduced towards the last 15 runs, when no more interleaved testing was undertaken as the Mi had finished its testing.
- Inconsistent charge termination and discharge termination from temperature variation – Sydney is in the middle of a cold spell, and despite my best efforts to keep the temperature to within 21 +/- 5 degrees C, this temperature variation could be enough to cause charging or discharging to be affected.
- Errors introduced by the multimeter’s sensitivity to temperature changes, which could result in slight errors integrated over several hours to result in such variations.
- Possibly loss of capacity could be slightly recovered by subsequent charging as trapped charge carriers somehow become “untrapped”.
It still does seem plausible that the cell is degrading, purely because the peaks are no longer as high as they were, and the troughs continue to deepen as a function of cycles. The predicted lifetime based on the linear model is 1588 cycles, and the lifetime based on the polynominal model is 1115 cycles – all well above twice the expected cycle life.
This might be a good sign to say that cells which undergo heavy cycle loading to 100% depth of discharge might last longer than expected if they are made of high quality materials. It could be that high temperature or high-state-of-charge storage over long periods causes more degradation, so it would make sense to use these cells rather than leave them sitting.
On the whole, the capacity is still a long way from zero, so people don’t need to worry. I suppose if this quality of battery was outfitted to devices, replacing batteries isn’t really a big worry as the devices will become obsolete well before the battery degrades to an unusable level.
This has been a long and ongoing experiment, and it seems that I still haven’t seen the failure that I had expected to see, even though the cell has now expired its “generally accepted” lifetime. This is good news for owners, and good news for users, as it seems heavy 100% DoD run back-to-back continually has not harmed this cell anywhere near as much as I would have expected.
The experiment will continue … possibly until I get utterly bored of it, or there is a failure of the circuitry or battery.