Sanyo Heater Prediction Routine

drbacke

Member
Joined
Apr 3, 2019
Messages
60
The problem with the Sanyo heaters is much discussed, butin my opinionone thing is clear:
You dont want to have a single heater in your powerwall, it's simply dangerous.

Wolf has developed a criterion based on the internal resistance of the cells, which I would like to verify with my own data and give you an advice how to actually use it. In his "Harvested Cell Analysis -> IR by Part#" Google Table are 2 IR Limits for a lot of known cells listed. One is called "optimal IR" and one is called "marginal IR", optimalIR < marginalIR and so optimalIR isthe "harder" criteria.
I checked both criteria with my own database which contains actually 41 Sanyo cells from which are 11 Heaters. The Test setup is mainlythe same as wolf used.

So lets begin with my results:

Average relative IR:

  1. The relative IR (measuredIR /marginalIR) for the heater cells 1.022
  2. The relative IR for the non heater cells: 0.789
This looks very promising, the heater cells seem to have a much higher average relative IR. But how accurate are the single criteria?



How well can I detect heaters with marginalIR as criteria?

The prediction:If the measured IR is greatern than the marginalIR from wolf its a possible heater.
Of my 11 cells, I wouldn't have recognized 4 as heater. So its a failure rate of about 36%. Thats to bad for using it as a detector.


How wellcan I detect heaters with optimalIRas criteria?
The prediction:If the measured IR is greatern than the optimalIR from wolf its a possible heater.
Of my 11 cells, I wouldn't have recognized1as heater. So its a failure rate of about 9%.
This is a very good result !!!


How wellcan I detect heaters with 55mOhm as fixedcriteria?

The prediction:If the measured IR is greatern than the 55mOhmfrom wolf its a possible heater.
Of my 11 cells, I wouldn't have recognized 3 as heater. So its a failure rate of about 27%.


Are all cellswith smaller IR thanoptimalIRnon heaters?
The prediction:If the measured IR is smaller than the optimalIR from wolf its non heater.
Of my 30 non heater cells, I wouldn't have recognized 1as non-heater. So its a failure rate of about3%.

Conclusion:
There is a strong correlationbetween theoptimalIR criteria of Wolf and the rate of heaters. Buthow canI use it practically in my harvesting process?

  1. As alone pre detection method for heaters?If you just use this criteria you will detect the most of thembut you willnot detect all heaters. So I would not recommend this. You dont want to have a single heater in your powerwall.
  2. As alone pre detection method for non heaters? It's the same problem, you might predict one wrong and so you end up witha heater in your powerwall.
So what to do?
First check the IR and throw cells with IR greater than optimalIR to the bin or check them verycautious by charging.
Then you can start the capacity test.
Whilecharging your cells,do your standard testing by measure the temperature or just feel with your hands regulary. Of course this is also not a 100% safe method to detect heaters, maybe you just checked the temp in the wrong moment and so you willend up with a heater in your powerwall. But if you use the IR measurement in combination with the"hand checking temperature",you will have a very verygood chance to detect all heaters. Also it's not 100%
 
Valid scientific based approach .. :D

... with one flaw:
Thats to bad for using it as a detector.

That solves 2/3rds of the problem you have, in a simple, time saving and safe way, and reduces the problem to one/third. :D
 
Cherry67 said:
Valid scientific based approach .. :D

... with one flaw:
Thats to bad for using it as a detector.

That solves 2/3rds of the problem you have, in a simple, time saving and safe way, and reduces the problem to one/third. :D
Yes you are right :sIhoped for an easier solution. But actually its not easy.
ButI also wanted to point out that it is not enough to rely on just one value.
AndI wanted to confirm wolf's results.
 
drbacke said:
Conclusion:
There is a strong correlationbetween theoptimalIR criteria of Wolf and the rate of heaters. Buthow canI use it practically in my harvesting process?

  1. As alone pre detection method for heaters?If you just use this criteria you will detect the most of thembut you willnot detect all heaters. So I would not recommend this. You dont want to have a single heater in your powerwall.
  2. As alone pre detection method for non heaters? It's the same problem, you might predict one wrong and so you end up witha heater in your powerwall.
So what to do?
First check the IR and throw cells with IR greater than optimalIR to the bin or check them verycautious by charging.
Then you can start the capacity test.
Whilecharging your cells,do your standard testing by measure the temperature or just feel with your hands regulary. Of course this is also not a 100% safe method to detect heaters, maybe you just checked the temp in the wrong moment and so you willend up with a heater in your powerwall. But if you use the IR measurement in combination with the"hand checking temperature",you will have a very verygood chance to detect all heaters. Also it's not 100%
Very good synopsis and well written up.

In your conclusion I only have some criteria to add.(Not criticism just criteria) :p

As alone pre detection method for heaters?If you just use this criteria you will detect themost of thembut you willnot detectallheaters. So I would not recommend this. You dont want to have a single heater in your powerwall.

I don't think I suggested that my IR measurements should be the lone means to check for heaters, only that cells that fall within thosemeasurements are most likely not to be heaters. Also a manual or electronic temperature check of any cell esp. the infamous Sanyos should be performed without a question. Additionally if you are using these "Sanyo" cells in a high demand application to be very weary of them. I have charged them at 1A,at 0.5A and at 0.05Aand to be honest the "heaters" will show up at any amperage especially when checking them with a thermal camera.A tenth or two of a degree will show up like a lightbulb on the thermal image screen.
These are pictures of all Sanyo cells being charged on a CC/CV parallel board at 4.2V at 50mA. Can you spot the heaters?
We are talking about 1 or 2 C barely can feel it by hand if at all.

image_zohfnn.jpg

image_dazumc.jpg

image_npyrxn.jpg


As alone pre detection method for non heaters?It's the same problem, you might predict one wrong and so you end up witha heater in your powerwall.

Again it is not the lone criteria but a guide to go by. Your mileage may vary but you will at least be playing in the ballpark. You may get a strike or two but in the long run you will win the game. This was not an exercise in prediction but an exercise in avoidance.
Avoiding cells with a potential of being a heaternot predicting cells with a potential of being a heater.


So what to do?
First check the IR and throw cells with IR greater than optimalIR to the bin or check them verycautious by charging.
Then you can start the capacity test.

Perfect could not have said it better myself. :) :) :) :)


Wolf
 
Just remembered that I have an IR device for night hunting which would be excellent for cell temp detection. Strange that I did not think of this before.
 
Back
Top