Li Ion Pulse Charging - Proof of Concept Test

I took my newly hacked relay board and adjusted the hardware and software parameters as follows:

Charge Termination: 4V (before scaling)

Discharge Termination: 3V (before Scaling)

Pulse duty cycle: 250 ms On, 750 ms Off (25% duty cycle)

Charging current of Pulse: 460mA for an average of 115mA.

Discharge current:  Approx. 106mA.

Rest Period: 5 Minutes

The first charge/discharge cycle had the following discharge curve: 

I note the charge voltage at the start of the discharge test is lower than the expected 4.0V. The rest period was 5 minutes and subsequent testing showed the voltage was sagging before the discharge test began. It was a "sacrificial" battery so this was unsurprising.

The calculated mAh capacity being discharged was ballpark with the CC/CV testing I had done so I felt confident enough to let the testing run unattended. Unfortunately, a program error meant the cell stayed on discharge and when I caught it the voltage was down below 2V.

After correcting that I let the testing run overnight. After 8 cycles I had the following discharge curves:

Note the negative step change in capacity and the voltage sag after resting was ever worse. So the cell was well and truly ruined but I noticed the rate of deterioration was slowing. The fall from run 2 to run 3 was larger then the small change between runs 8 and 9.

So I did one more run and found an improvement:

Perhaps that was nothing more than the variation I might expect. I grabbed another cell and after some code house-keeping found I had introduced anther error and again discharged this second sacrificial cell to below 2V.

With all that fixed, and perhaps in hope that the cell would "fix itself" I left the second cell on extended test. I'll report back when more data is to hand.