Correlation of internal resistance (IR) and self discharge

paddy72

Member
Joined
Nov 17, 2021
Messages
143
Hi all,

it is generally agreed (i guess?) that there is a correlation of a li-ion cells quality (Soh) and its IR. In most cases an increasing IR indicates a decreasing Soh, decreasing capacitance and increasing self discharge rate (SDR). I recently tested all my salvaged 18650 with my new YR1035+ to measure actual voltage and IR.
I came across a few cells from an ebike battery pack made of Sony SE US18650V2 (NMC, 2100 mAh) from which i knew they had quite high SDR and others which are fine SDR-wise. When i talk of high SDR i mean something like dropping 400 - 600 mV in 6 months! The IR of these cells is only insignificantly higher (43...44 mR) compared to the good cells of the same pack (with an IR or 40...41) which only drop around 10 mV in the same time (6 months).

At least, there is an increasing IR, but its not as significant as i would expect.
For other cells the IR difference is much bigger for the bad cells and goes up to 100...200 mR but often with less SDR than the ones mentioned above.

Anyone with similar experience?

BTW: this topic should be in "general cell testing and recycling" :)

P.S. i charged some bad cells to nearly full and did the measurement again: even the slight difference in IR nearly disapeared so they really look like the other good cells with no increased SDR. Probably the temperature went also up a bit - so that might bias the IR readings. As a result: there is not always a direct relation between high SDR and high IR.
 
Last edited:
Definitely a @Wolf thread for response. He's the pack leader on IR measurements and the various correlations.

But yes, generally speaking, the IR has a pretty profound effect on SOH. A few mOhms variance from stock isn't going to have much of a big impact, but 10's or 100's definitely
 
it is generally agreed (i guess?) that there is a correlation of a li-ion cells quality (Soh) and its IR. In most cases an increasing IR indicates a decreasing Soh, decreasing capacitance and increasing self discharge rate (SDR)
I agree that there is a definite correlation between AC IR and the SoH of a Li-Ion cell. SoH being defined by the percentage of capacity loss from the original manufacturer designed / claimed mAh ratings of the cell.
By definition a cell that is a SD would self discharge at an abnormally fast rate to an unacceptable voltage (i.e. 4.2V to 4.0V) over a short (30 days) period of time whereas SDR would be the calculation of a daily voltage loss over an extended period of time.
Taking my test results from my last 14s80p battery build and at an average of 79 days for the SD V drop test the average V drop of 1120 cells was
0.0335V and the average daily "SDR" was calculated to 0.00043V.
1639067170098.png
One thing that is paramount to remember, if you are going to do a SD/Vdrop test, is to measure the cells voltage after it has finished its C/D/C cycle with a proper meter and definitely not count on the Charger /Analyzer for an accurate representation of said voltage.

Now since my powerwall is virtually done, all I have to do is hook up my 4th and final 14s80p battery, and I am now delving into the investigation of AC IR, DC IR, SoH, SD (Self Discharge) and SDR (Self Discharge Rate) and other non destructive cell testing analytical matters as time permits.
I have a lot of mixed tested cells left over from my first build (The Frankenstein Pack) and I am running them through 2 chargers again. The 2 MegaCellChargers that I have and I have just purchased 4 more SKYRC MC3000 for a total of 5. One reason is just to see how the 2 best (in my opinion) charger/analyzers match each other.
I will have a fire sale on all my other charger/analyzers soon.
As I am digging through all my "extra" cells I am trying to gather same manufacturer and model numbers together to test.
Since I first tested these 6000+ cells starting Dec 2018 to August 2019 and after a V drop test, chose just 1120 of the best cells, I have a bunch of cells that were just put aside with a "full" charge.
So I am now measuring IR and V again and testing these cells.
This brings me to several points.
High IR does not necessarily indicate a "SD" or a "SDR" cell. However it will most likely indicate that the SoH is certainly not up to par.
These cells have been sitting for at least a year at "full" charge and in this example are all LGDAS31865 or the ICR18650S3 series from LG
cells. These cells have a high IR from the get go. Manufacturer spec ≤80mΩ as it is a low drain workhorse cell for many applications.
This is an unusual cell as it can have a quite high IR yet perform well.
So looking at the sheet you will see I sorted by highest IR and the first cell you see has an IR of 203.56mΩ yet the voltage is 4.096V (cell has been sitting for at least a year) and tested just now with a 68% SoH.
The second cell with an IR of 177.35mΩ still has a voltage of 4.08V yet falls flat on the capacity test.
From there you can just indulge yourself into the data. Also if you want to see this work in progress here is the link to view as I will continue to update it as I run these old cells through the charger/analyzers.
1639070245990.png
One thing is irrefutable though IR does influence SoH and it can not be any plainer than this graph of ≈100 LG S3 2200mAh cells.
The trend lines cannot be argued with.
Wolf
1639070985863.png
 
Last edited:
Thanks a lot Wolf, for this detailed description - much appreciated! :)

What is "MCC IR" and "MCC Voltage"?
What is interesting in your Excel sheet is there are "Ausreisser" (non-typical cells) with a "bad" IR but very (or quite) good SoH - or vice versa. Your second picture shows the overall trend, which i perfectly agree, but there are still these cells which dont really fit into the schema - and this - unfortunatly - biases the predictability of Soh just by the IR. And you also can't predict the SD-rate by just watching the IR, this is probably even more non-corelated? But in general these situations are exceptions from the rule of thumb - i agree :)

BTW: not 100% sure if i understand what you concider SD and SDR cells. For my understanding "SD" just means Self-Discharge and SDR means the rate of Self-discharge. A SD-cell then is a cell with an abnormally high SDR, so what is a SDR-cell?

Not 100% sure if i got the quintessence of your findings completely to your understanding? :rolleyes:
 
In my small set of 1120 cells I didn't find a relation between high IR cells and self discharging cells. My (limited) knowledge is that old cells have higher IR while bad cells self discharge.
 
What is "MCC IR" and "MCC Voltage"?
What is interesting in your Excel sheet is there are "Ausreisser" (non-typical cells) with a "bad" IR but very (or quite) good SoH - or vice versa.
BTW: not 100% sure if i understand what you concider SD and SDR cells. For my understanding "SD" just means Self-Discharge and SDR means the rate of Self-discharge. A SD-cell then is a cell with an abnormally high SDR, so what is a SDR-cell?
MCC = Mega Cell Charger
I take the MCC IR which is DC IR and the MCC voltage as a comparison to my RC3563 which is by far more accurate. The DC IR I record to compare it to the SKYRC MC3000 results which also does a DC IR measurement.
1639140251693.png

By definition a cell that is a SD would self discharge at an abnormally fast rate to an unacceptable voltage (i.e. 4.2V to 4.0V) over a short (30 days) period of time whereas SDR would be the calculation of a daily voltage loss over an extended period of time.
SD and SDR my explanation above^^^
My understanding of SD is a cell that rapidly "Self Discharges" to an unacceptable voltage level after a short period of time, usually <30 days.
SDR on the other hand is a naturally occuring self-discharge typically stated by manufacturers to be 1.5–2% per month.

What is interesting in your Excel sheet is there are "Ausreisser" (non-typical cells) with a "bad" IR but very (or quite) good SoH
Yes those are some headscratchers, but I may have a possible answer to this. I am taking the liberty to patch together some interesting snips from an ICR18650-26F spec sheet. As you can see initial IR should be ≤100mΩ. In my experience on this particular cell anything much over 55mΩ is a crap shoot as far as SoH is concerned. Now take the PTC into consideration. Initial resistance is 9-18mΩ and the resistance spec (Max 33mΩ) after a trip raises that resistance much higher in essence giving the cell a much higher "IR". Does this mean the cell is no good?
No not really. So in theory a good cell with normally a 55mΩ AC IR could theoretically have an AC IR of 88mΩ and still be good. Would that resistance increase cause a little temperature increase? The PTC specs seem to indicate 2.5W. Nothing drastic but nevertheless there.
1639146336269.png
Additionally if we take that data and apply it to the LG cells in my sheet especially the "Ausreisser" (non-typical cells) we can theorise the same way. The AC IR of theses LG- ICR18650 S3 is supposed to be ≤80mΩ so we add the resistance of 33mΩ for a tripped PTC and our AC IR can now be in the 110mΩ range and still give us acceptable results.
1639146941407.png
These are some of my theories and quite yet unproven but worth exploring. I may be completely wrong, Lord knows I have been wrong many times before.
Nevertheless I would not use high IR cells (>55mΩ) combined with "normal" IR (30mΩ to 50mΩ) in a powerwall or a critical battery build no matter how "good " the SoH results are.
IR has a very interesting affect on cells in parallel as my study on this has pointed out.
Wolf
 
Thanks for sharing these interesting findings!

So you think the PTC's resistance is the most important factor that biases the corelation of IR and SDrate or Soh?
I didn't know that the resistance of the PTC changes (permanently?) after it has tripped once? I have to test my PTC-fuses if they change their resistance at the same (low) temp. after tripping once. I always thaught they come back to their initial resistance after cooling down again.
Maybe there is a slight degradation when tripping but i was thinking they can be used many (100...1000) times before they quit or become unusable due to increasing resistance. Maybe the material of PTC-fuses is different from the PTCs in 18650-cells?

For the "definition" of SD or SDR cells i simply have an other understanding: There is no fixed value when you can concider a cell SD - its more a fluent relation. For me these are just abbreviations for "Self Discharge" and "Self Discharge Rate" and a cell with (too) high SDR can be called a Self-discharger (SD). When you say 1,5 - 2% SDR per month is "normal" (from the datasheet) i think these are already quite high values! 2% of 4V means 80 mV and this in my experience is already a quite high value of SDR. I store all my cells around 3.8 ... 3,7 V and measure their voltage normally 2 times a year. I have many cells which discharge only 1...5 mV in 6 months! And i even have cells that discharged only 2 mV in 2 years!
80 mV/month would mean nearly 0,5V in half a year - thats terribly bad! I would throw these cells out immediatly and dont use them for any diy packs.
On the other hand i also have some bad cells which discharge 100mV and more in a month - these are already very bad imho and i would call these SDs.
We also have to take into consideration that from a level of 4.2V the SDR will be much higher than from a level of 3.8V when sitting there for awihle. I never measure the voltage right after charging a cell fully up, but 1 or 2 hours later when the cell has calmed down and the voltage settled. Still from 4.15V you will have much higher discharge (in mV per day) than from 3.8V. So the SDR is depending on the initial level you start your measurement.
One point i am not sure about: should we call a cell bad when it discharges from 4.2 to say 4.1 V in 14 or 28 days or should we better test it in the range of 3.9V to 3.8V? Many of my recycled cells tend to drop quickly at high SoC and keep their voltage very well in the midrange of SoC.
But there are some cells that dont keep the 4.2 for long and go down to 4.15 or even 4.1 in hours - but than are stable. They probably have some kind of issue...

BTW: you have quite an impressive Lab :)
 
So you think the PTC's resistance is the most important factor that biases the corelation of IR and SDrate or Soh?
I do not think it is the most important factor as I do not know if it has anything at all to do with SoH.
It certainly should not influence the SD of the cell as it is not a resistor tied to pos and neg of the cell but in series with the pos terminal.
It is a theory based on what the (many) spec sheets say that after trip IR is expected to go up by a certain amount.
These are some of my theories and quite yet unproven but worth exploring. I may be completely wrong, Lord knows I have been wrong many times before.
As stated above I have no idea if this is even plausible but it is a theory.

For the "definition" of SD or SDR cells i simply have an other understanding

I say potato you say potáto SD, SDR your choice in how you define it. Manufacturers say from full charge 4.2V expect a 1.5–2% per month self discharge. In my experience it is nowhere near that. It is what is considered a safety statement by the manufacturer. Additionally most cells will stabilize around 4.16V and the self discharge at that voltage, of a "good" cell is already greatly reduced. If you store cells at least at <4.0V or the manufacturer recommended storage voltage (usually 3.6V to 3.8V) then there is a very limited amount of "SDR" or "SD" or whatever you want to call it. In essence a very limited amount of self discharge. I test all my cells at full charge Voltage which ends up being ≈ 4.14V ±0.04V or whatever the voltage is they come out of the charger at for SD (Self Discharge) within usually 30 to 60 days. Any voltage drop over 0.05V would cast a shadow of suspicion on that cell and would more than likely be set aside.
Yes most all "good" cells when charged at 4.2V at the manufactures charging mA cutoff and left to sit for 60+ min will settle between 4.13V and 4.17V.

BTW: you have quite an impressive Lab :)
Thank you!
Here is the full picture.
20211215_224751.jpg
 
Last edited:
Yes those are some headscratchers, but I may have a possible answer to this. I am taking the liberty to patch together some interesting snips from an ICR18650-26F spec sheet. As you can see initial IR should be ≤100mΩ. In my experience on this particular cell anything much over 55mΩ is a crap shoot as far as SoH is concerned. Now take the PTC into consideration. Initial resistance is 9-18mΩ and the resistance spec (Max 33mΩ) after a trip raises that resistance much higher in essence giving the cell a much higher "IR". Does this mean the cell is no good?
No not really. So in theory a good cell with normally a 55mΩ AC IR could theoretically have an AC IR of 88mΩ and still be good. Would that resistance increase cause a little temperature increase? The PTC specs seem to indicate 2.5W. Nothing drastic but nevertheless there.
View attachment 26605
These are some of my theories and quite yet unproven but worth exploring. I may be completely wrong, Lord knows I have been wrong many times before.
Nevertheless I would not use high IR cells (>55mΩ) combined with "normal" IR (30mΩ to 50mΩ) in a powerwall or a critical battery build no matter how "good " the SoH results are.
Interesting. I'm currently processing 2600 Samsung ICR18650-26F(s) with 10yr old date codes from "Modem" packs.
1639676249971.png
After about 100, some patterns are emerging. The OPUSs are showing 110-120mOhm which 'could' be close to <100mOhm. 30-40% of the cells are 0v but the rest are 0.7v'ish and running in the 85% - 95% range in the C/D/C capacity test. Self-discharge doesn't seem to be a problem yet - we'll see.

I think for me, I'll proceed building one of my batteries - e.g. 14s110p in this case - but will be 100% of same cell type. Once I get it online (eta this summer) I can report some results after a while.

One key thing is that my battery bank is large enough this will take it down to 130ma/cell average charge/discharge range. I think that's a 130ma/2600mah capacity = 0.05C range. (do I have that right?) This low C may be a mitigating factor to elevated IR in this case.
 
I have some cells, that were shown as 100+ mOhm in my LiitoKala Li-500 but were around 70-50 mOhm in my 1kHz IR meter. I really suggest you to get one.
 
I have some cells, that were shown as 100+ mOhm in my LiitoKala Li-500 but were around 70-50 mOhm in my 1kHz IR meter. I really suggest you to get one.
I know what you mean, but after several years and 10,000+ cells on my same 3 OPUSs - I'm OK with the aproximate mOhm reading from the OPUSs.

I even have them 'calibrated' one of them in relation to the other. This one, I add "50ma" to the discharge tests for example :)
1639685237151.png
I think one of the things that works for me to add successful packs to the powerall is to use the same process / equipment for all of the cells which tends to get me consistent results in a relative way.
 
Last edited:
I even have them 'calibrated' one of them in relation to the other. This one, I add "50ma" to the discharge tests for example :)
I'm doing the same with my Liitokala li-500's and a DL24 discharge tester. But the mOhm reading from those liitos are still very inaccurate and i'd rather not use them at all.
 
30-40% of the cells are 0v but the rest are 0.7v'ish
Ouch. Well I wish you the best. The only enlightenment I can give you is of the 49 Samsung ICR18650-26F cells that I originally tested with just the OPUS is this chart. It has the YR1035 and OPUS IR on the primary axis and the capacity correlated to the secondary axis.
1639684928023.png
Here is the chart of 290 ICR186450-26X as in 26A, C, D, F, FU, & H. with the YR1035 and OPUS IR correlated to capacity.
Kinda a convoluted mess with the OPUS IR nevertheless there is a trend. The AC IR of course is far more smooth.
1639687783022.png
Oh and my lonely OPUS just because.
1639687668753.png
Wolf
 
Just for fun, I plotted (approximately) the current batch of 12 Samsung ICR18650-26F(s) I'm currently testing on your chart. I understand this is just anecdotal.

However, the overall experience I'm having is consistent with these 12 and seems in bounds with the chart. Plus, the OPUS -> YR1035 info seems like icing on the cake. All this encourages me to proceed to build the battery. Good research/documentation and a thank you for sharing! :)
1639755334304.png

1639755963391.png
1639756076321.png
P.S. Yesterday I was able to order what I hope are 600 more modem pack Samsung ICR18650-26F cells - giving me a 40% loss rate leeway. This will remove the temptation to attempt salvage of some borderline cells.
 
Last edited:
I've processed about 600 of these modem pack / Samsung ICR18650-26F cells so far and its interesting. 55% are coming out of the modem case at 0.5v - 0.9v - and these are testing OK on capacity and IR (as shows above).

However, out of the 45% coming out at 0v - 0.004v... I've tried to help along about 40 of these with the 0.004v (as apposed to 0.001v) and they ALL are either self-dischargers (never reach 4.20v charge) or test dismally as in 40-60% original capacity.

This confirms that reaching 0v = death knell for these cells.
 
Last edited:
I've processed about 600 of these modem pack / Samsung ICR18650-26F cells so far and its interesting. 55% are coming out of the modem case at 0.5v - 0.9v - and these are testing OK on capacity and IR (as shows above).

However, out of the 45% coming out at 0v - 0.004v... I've tried to help along about 40 of these with the 0.004v (as apposed to 0.001v) and they ALL are either self-dischargers (never reach 4.20v charge) or test dismally as in 40-60% original capacity.

This confirms that reaching 0v = death knell for these cells.
Try to put them aside for a week and charge them again, some can turn out just fine.
 
Back
Top