21.9.15

Diagnosing and reviving an old bike battery

My neighbor mentioned that his bike battery didn't have nearly the full range any more. It was flat after only 5 to 10 kilometers. I had just learned about a simple way to measure battery resistance, so I decided to take a look.
It's fairly heavy and contains 5 rows of 6 batteries. Fun fact: the charging pins are wired in parallel to the output pins, so you could power your bike with the battery charger. Don't forget your extension cord.

This Schachner battery costs about 700€ and is filled with 30 NiMH batteries, each with a capacity of 9Ah. They're all wired in series, with the ends connected to the terminals via a fuse. Other than that, there's only a thermistor in the middle of the pack, no other battery management.

I started the diagnosis by removing strips of the blue isolation, and cutting off small segments of the wrapping paper. This way, I could access each cell individually. I also numbered each cell, so that I could keep tabs on the results.


Some of the cells showed some serious discoloration, and at first I thought that the cells had leaked.
I started checking the cells. Measuring battery resistance relies on three values: idle voltage, load voltage and load current. To load the batteries, I chained a couple of wire connectors together, resulting in a resistor of about 0.8Ω. Then I measured the current when this resistor was connected to each cell. But it turned out that 0.8Ω is not nearly enough load for a beefy 9Ah cell, and the load voltage only dropped by a few tens of millivolts.
After ten cells, and very sketchy results from shaky connections, I realized that the cells could take much more load than that (after all, the 250W motor has an equivalent resistance of 0.1Ω) and connected the leads of the multimeter directly to each cell. The two leads, and the shunt in the multimeter together have a resistance of 0.18Ω, which resulted in a healthy current of about 6A per cell. I waited a few seconds during each measurement for the voltage to stabilize, and wrote down the values. This chart was the result:



I used this formula to calculate the internal resistance values:
internal resistance = (idle voltage - load voltage) / load current