The following is a timeline of how I rebuilt my failing IMA battery with a junk pack. Not a definitive guide, just info for others looking into it. I lucked out where most of my existing pack still had good cells.

Why you should not do what I did

• Time and labour intensive
• Results are not guaranteed based on matching cell performance or self discharge of multiple sticks.
• Risk of electrocution - Treat everything as a live voltage source.

Backstory

Bought a used 2007 civic hybrid with 170,000 km. Tried grid charging and light bulb discharging with limited sucess. IMA light would return and go away 1-2 weeks after. Constant recalibrations dropping to 1-2 bars with every drive. Didn't really care as long as the 12V battery is still charged and car runs. Found a used battery with 300,000 km for $100 CAD knowing most of the cells are junk. Has some spare time over the holidays to experiment and learn about the IMA pack. Wouldn't recommend everyone try this as self discharging voltage measurement is not the best way to measure cell performance under load as some cells may lose voltage only under high load.

Pictures of the process:

https://photos.app.goo.gl/aKTRbkbwE4U6HCF26

Data

First thing I did was disassemble the junk pack and measure the voltage of each pair. The voltages are severely discharged as the pack was sitting for an extended amount of time. Some people would say the sticks are already damaged due to cell reversal. Voltages increased on day 1 because the batteries warmed up inside my house. Outside is -15 C.




Next thing I did was take out my existing pack which hasn't been grid charged in the past 3 months. Note how stick I2 is severely discharged compared to the highest stick. Skeith stated in the past 0.1 V difference is suspect, 0.2 V is junk. I also measured the difference between the positive and negative cell. All cells were within 0.01 - 0.05 V of eachother with the exception of stick I2.

Measured voltages the next day. Results similar to day 0.

At this point I knew stick I2 was a bad stick, but not sure how bad compared to my junk pack. So I replaced I1 (junk pack) with stick I2 and grid charged and drove it for a couple of days. Next I let my existing pack self discharge for a week and found H2, J2, and K2 are weaker than the other cells.



After a week of driving the junk pack with bad I2 cell, I pulled it and measured the voltages to find the strongest cell (E1) which I could replace my I2 cell with. Note the I2 cell has 16.50 of charge and it appears to be a strong cell compared to the rest of the pack. (Turns out it wasn't).





Next I put cell E1 back into my existing pack and grid charged overnight. So far after 7 days, the pack has more assist and auto-stop than ever before. I should have waited at least 7 days to find the strongest cell, but I didn't have anymore spare time to work on it. I measured the voltages of the junk pack after 3 days and found the I2 cell is actually very weak and self discharged the most. In the future I might replace H2, J2, and K2 with B1, D1, and H1 if I get more recalibrations.



Conclusion

• Replacing the weak stick pair appears to improve battery performance.
  • benefits could be temporary as other sticks could fail getting me back to square 1.
• The weak cell might have gotten excess wear over time through constant recharge and discharge, Saving the capacity of the remaining cells.
• Self discharge over 3-7 days seems to be predictive of cell performance.
  • Better ways to assess stick capacity would be through high current amp loading or low current Imax B6 to get a mAh reading, but I don't own neither.
  • The method I used is partly based on luck and guessing the strong stick from the weak pack will have comparable performance to my existing pack.
• My results seems to show sticks at the center and botttom of the pack have more wear compared to the rest of the pack.

Here is an interesting russian video of someone replacing failed D cells and soldering them together. I personally wouldn't take the risk doing this, but its a fun experiment.


Shoutout to skeith for giving background posts about it.

 

Bummer... quoted and in-context reply ready to go, and then computer crash... here it is in a nutshell.

Only big mistake you made was not cycling the junk pack 3 times. Dormant, deeply self-discharged NiMH needs to be cycled to assess performance.

Those people would be talking out their butt. Had you punctured the shrink and measured the individual cells, you would have seen they were all normal polarity and > 0.1V.

Quoted out of context, but the spirit of the content is in the right direction.

While benefits may be temporary, if you can get away with an overnight or soaking grid charge once every 90 days and a SINGLE discharge to 96V once a year, then it's a big win.

Weak cell was the weak cell. It's exaggerated behavior during discharge registered empty first (the negative 2 bar recalibrations you experienced), thus the other subpacks were not discharge as much, i.e., their depth of discharge was reduced (not as deep), thus less wear and tear on those cells.

Self discharge should be assessed at a 7 day MINIMUM to be truly meaningful. Longer is always better.

Almost nothing is good about the IMAX B6... it's a piece of sh!te. Its results can only be considered useful IF you validate with a load test, so your OR proposal should be AND.

Yes. You got lucky, BUT you took a sensible approach based on reasonable criteria. All too often, initial results are amazing, but it goes to turds shortly. It remains to be seen if you avoid that, but your time to turds should be a longer wait than most.

Recommend you:
soak charge the junk pack for 25 hours (assuming 350mA charge) with a box fan or carpet dryer blowing through it.
Discharge to 96V (any current down to 158V, 400mAh max down to 96V)
Charge for 25 hours
Discharge to 96V (any current down to 158V, 400mAh max down to 96V)
Charge for 25 hours
Discharge to 96V (any current down to 158V, 400mAh max down to 96V)
Charge for 25 hours
Log voltage after 24 hours, 7 days, 14 days, 30 days, 60 days and 90 days

Once you believe you have a truly good candidate(s) after the above, puncture the shrink with your voltmeter leads and check all twelve cells. < 0.01V variation = good subpack, up to 0.03V variation = workable with periodic grid charging, and it gets worse above that.

Lastly, don't do what the crazy russians did at 12:36... parallel wire all the sticks and then charge them in parallel. GREAT way to enjoy thermal runaway.

P.S. Nice work!