Joule Smasher Led Flasher - Update at 18 Months into 10 year Life on AA

It has been a surprising 6 months since I last reported on this project. Shortly after reporting at 12 months the discharge rate increased significantly. It was a hot summer so perhaps that was the reason? With a sudden transition to cooler winter temperatures in the last few weeks the discharge rate has slowed and is now running at essentially the same rate it was prior to summer.

I had expected the discharge rate to plateau based on what I had seen with AG3 cells. However, it is yet to happen. At present the discharge rate for the last 12 months is 0.3 mV/day and the forecast life is 8.5 years. Hopefully the plateau does eventuate.

Li Ion Pulse Charging - Proof of Concept Test Part 6

I changed the software to try and reduce the reported capacity variation due to the variation in cell voltage after resting. 

After resting for 60 secs I started the pulse charging cycle again. Once the termination threshold is again reached rest for 10 secs then start the discharge test. That has seen a significant reduction in the standard deviation of the reported capacity for each run. It's only reduced the impact of the starting voltage when discharging begins. 

 

I'm surprised that these two severely abused batteries show no signs of declining capacity after some 400 partial discharge cycles. I'm going keep them on the pulse charger until that happens. I will refine(?) the charging algorithm again to keep them on the pulse charge cycle until the voltage stays above 4.0V for 5 seconds while resting. How long the modified algorithm takes me to do is unknown since tonight when I attempted this the workshop PC malfunctioned. I suspect the power supply :(

I'm started assembling the better hardware. Soon I will be putting some brand new cells through the testing process with control cells and CC/CV charging versus pulse charging.

Testing Anko CR2032 cells in a Joule Smasher Led Flasher - Update 4

119 days have passed and the Anko CR2032 cell appears to be on track for a good result. A back of the envelope estimate was for a life of 400 days. The cell voltage has just dropped below 3.0V and if the decline in voltage over the last two months holds true then 2.8V will be reached in 280 days time.

Li Ion Pulse Charging - Proof of Concept Test Part 5

 I wasn't going to bore you further with the results of the "test" runs but today I noticed something. One of the outcomes from testing cells which are now abused is that the low cell capacity gives a fast cycle turn around. I was looking at cell F, which has now been though 300+ pulse charge cycles, and I was trying to understand the cyclical variation in capacity.

It appears that the reported cell capacity for a charge/discharge cycle is highly correlated (r=0.974) with the cell voltage at the first discharge measurement. This measurement is taken 20 seconds after discharge commences. 

 The first handful of observations demonstrate this:

	Cycle	mAh	Voltage at t=20 sec
	1	77.3	3.779
	2	69.7	3.773
	3	68.3	3.766
	4	67.2	3.760
	5	65.8	3.753
	6	63.6	3.747
	7	63.4	3.747
	8	62.2	3.740
	9	62.1	3.740
	10	62.2	3.740
	11	62.3	3.740
	12	68.1	3.760
	13	74.1	3.779

Comparing cycle 11 and 12 a 20mV difference in the cell voltage existed after 20 secs of the discharge test. The ADC step size is 6.5mV per bit. So it's a much larger error than I expected and some experimentation with the charge termination is required.