"Good" IR for common 18650 cells

Nemo

Member
Joined
Apr 26, 2019
Messages
53
For those interested in IR and wanting to know what a "good" IR is for various cells I've put together the following table. This was built from a combination of Wolf, 100kWh Hunter, and my own data. All IR measurements were taken with a 4-wire Kelvin type AC 1kHz impedance tester, specifically the YR10XX series. This expands on original work by Wolf, so credit for the idea goes to him.

The values have been derived from a linear best fit line of all measuredremaining capacities greater than 70% for each type of cell. There is still noise within this data, so optimal IR has been defined as the IR which correlates 85% remaining capacity according to the trendline. Marginal Capacity is defined as the IR which correlates 75% remaining capacity according to the trendline. This means that you are likely to achieve 80% remaining capacity if you measure the IR to be equal to or less than the optimal IR. And there's a chance you will achieve it if IR falls between optimal and marginal IR. If the IR reading is above the marginal IR value you are unlikely to achieve 80% remaining capacity.


The hope is that this will save harvesters a lot of time by allowing them to focus on testing the cells with the greatest chance of success. Also, if purchasing tested secondhand cells, be sure to ask about IR measurements and ensure they are good!

Thank you very much to Wolf and 100kWh Hunter for sharing their data. If you like this information and want to contribute, I will accept data collected using reasonably accurate charges (equivalent or better than OPUS) and IR readings from a YR10XX type AC impedance meter or equivalent. The more data, the better the results.

If you'd like to view the source file it is available on my google drive under "Cell IR Database"


image_qrrejp.jpg
 
Hey Nemo,

Would love if you could add my log of cells to establish a community based benchmark for the LGDAS318650 cells.
I am using a SM8124A battery tester to measure IR and once i have my IR number i then use a Opus charge test (at 1A) to obtain the mAH of each cell as we shuck through the packs.

Cheers
 
Brute-averaging that table one could say that you lose 5% capacity for each 10mOhms. So each 2 extra mOhms from spec value would result in 1% capacity damage.
 
Overmind said:
Brute-averaging that table one could say that you lose 5% capacity for each 10mOhms. So each 2 extra mOhms from spec value would result in 1% capacity damage.
It's probably a good idea for me to include manufacturer's IR in future tables. I'd have to bring it in to double check but I recall seeing some manufacturer's IR that were way above what turns out to be the marginal IR based on this method. I don't know how or why that could happen. Wait for me to update and we can discuss based on actual values.
 
I have tested about 7000 cells and have aprox 5000 cells online. When I test (using OPUS) I accept down to 85% of nominal/speced mah with the majority around 93%. I would note 2 things:
1) The posted spreadsheet is exactly in line with my experience - although I agree this is anecdotal I have not way of measuring the accuracy of my OPUSs
2) I'm not sure that X mOhms per capacity (at a couple of %) is meaningful (to me) since my actual operations - battery bank charge/discharge daily over many days - seems to be running perfectly and I'm sure the IRs were a bit here and there regardless of 85% or 93% mah charge test results.

I think that OPUS is a blunt instrument in some ways compared detailed analysis of 1% = 10mOhms. I would put OPUS accuracy more in the range of 5% capacity, 50mOhms, fuzziness overall - something like that.

What do you folks think?
 
If you say down to 85% its more like between 70-90% that you accept :) The Opus is easy +-100mAh and also it generally is a tad high. With that said I dont think its a problem as long as you accept the variance and use it throughout.
iR is more like +-100% on the Opus since it can vary between 50-150mOhm and more just by pressing on the contact.

I think it will work fine for a powerwall as I have stated since many years now :)
 
The opus charger was used because I have one, and so do many others. If you go look at the original data there is lots of fuzziness, but generally speaking the correlation between IR and % capacity remaining is fairly linear to about 70% of capacity, which allows for the linear trendline.

@ daromer - no doubt, your method of +/-100mAh binning is proven to work. This exercise is more about reducing the time it takes to collect good cells, and the required investment in chargers. Using this kind of data you can weed out cells and increase your per charge success rate (finding a good cell) to near 100%. That means if you havethe supply to manage it a person couldtest enough cells in a year on a single 4 slot charger to build a 14s80p wall. That significantly lowers the time and gear investment for people just getting into this.
 
My experience with OPUS has been quite variable.

I'm dealing with a single type/brand (LGDAS31865), thevalues i get from the Opus vs the SM8124A meter is wildly different.

Overall the SM8124A meter seems consistent, i kinda wish i added a 'OPUS IR value' and 'SM8124A IR value' in my spreadsheet for comparison purposes. I'm sure it would have been an interesting comparison!

The datasheet (for my cells) suggests that 80 mOhm was the acceptable pass for initial testing. I do notice that when i conduct a 1A discharge on cells metered around 70-122 mOhmthat the cells heat up (and in some are scolding hot!). A majority of cells sit around the 55-62mOhm range and they seem to be really good in both capacity and overall heat during charge and discharge.

So i guess you need to lookup the datasheet to get a real feel for what your cells can do.
 
intra the IR from Opus is not even close to being interesting. As said that depends on the pressure... So it basically depends on how you insert the cells into the tester. Therefore i would say ignore them and go with the SM8124 instead.

I have done a video showing how it can differ from 70 to 250 by just going back and forth or pressing.

As said IR is direct relation towards how the cell performs in terms of current vs heat. And yes for those that have time to do IR test do it. It goes fast to do a first check .
 
daromer said:
intra the IR from Opus is not even close to being interesting. As said that depends on the pressure... So it basically depends on how you insert the cells into the tester. Therefore i would say ignore them and go with the SM8124 instead.

I have done a video showing how it can differ from 70 to 250 by just going back and forth or pressing.

As said IR is direct relation towards how the cell performs in terms of current vs heat. And yes for those that have time to do IR test do it. It goes fast to do a first check .

No one is using the IR from the Opus, we were all using a YR1030 kelvin type (or similar model).
 
Nemo Read the post from intra that i answered.
 
only use yr10** , more and more users are coming.
Why not as a mark or standard.
There is a lot of data based on this ir tester, the tester is tested a lot.
The tester is had a lot comparative research with a lot ofother testers.
Why don't you make this as a standard tester for your data?

Don't forget the 10% rule, if factory dictates 26 mohm min/max is ~24 to 28.8 mohm.
Or factory dictates 68 mohm it would be ~62.2 to 74.8 mhom for a giving cell.
But this is what I applied in my pw

thanks in advance
 
Because in a DIY community we have varying levels of commitment, time, money and risks.

I think overall we all want the same thing, a safe and happy community doing what we enjoy as a hobby, pushing a 'standard' is hard work and i think all we can do is suggest better ways to do something and let the end user decide on what fits them best.

I settled on the SM8124A by seeing a video from daromer and its helped me understand the relationship/trend in IR for my cells. I'm sure it would have been different if daromer did a YR1030 instead, but thats just how it worked out! :)

The positive thing was that i moved away from using the OPUS as it was insane some of the number i was getting back, it would have made me junk atleast half the cells i have currently.
 
At this point the standard is a 4-wire kelvin type IR tester. To push that a bit farther I'd say one that's been tested and found to be acceptably accurate. Making it any more restrictive limits the amount of data that can be collected, and I think more data (within an acceptable margin of error) is worth it. Considering the margin of error is already set by the chargers themselves being out up to 10% there's a pretty wide acceptable margin.
 
the SM8124A is pretty much same as the YR. I doubt you will have any issues with either or. They may not be on same point but they are most likely accurate enough depending on the version you have. The first one of the SM had bad design tips but the later versions was better.
 
daromer said:
the SM8124A is pretty much same as the YR. I doubt you will have any issues with either or. They may not be on same point but they are most likely accurate enough depending on the version you have. The first one of the SM had bad design tips but the later versions was better.

Agreed.
I have tested pretty much all the most common (reasonably priced) IR testers.

image_qnqqjz.jpg

Here are the results. (Skip the BVIR-2018 it is Junk)

image_dcpozp.jpg

Although my favorite is the YR1035+ (not shown)
Seehttps://secondlifestorage.com/showthread.php?tid=7742&pid=53191#pid53191

Wolf
 
I am curious if you ever tested the accuracy of the YR1035+ against any popular RC chargers which have 4 wire IR or the manual method by adding a known value resistive load and looking at the voltage drop.

I typically use an RC charger for these measurements but I have been curious about accuracy or how it might stack up to the testers you reviewed here. I can at least do the manual method for comparison and report back.
 
I compared the SM8124A with my Icharger DUO 308 with attached 4 wires and its pretty even.
 
CrimpDaddy said:
I am curious if you ever tested the accuracy of the YR1035+ against any popular RC chargers which have 4 wire IR or the manual method by adding a known value resistive load and looking at the voltage drop.

I typically use an RC charger for these measurements but I have been curious about accuracy or how it might stack up to the testers you reviewed here. I can at least do the manual method for comparison and report back.


I have not.
The only RC charger I have is an iCharger X6 and it does not do 4 wire.
I have compared it to the ZH-YU ZB206+https://secondlifestorage.com/showthread.php?tid=7742&pid=52825#pid52825
Other than that just testing 4 different IR testers and coming within 3m? of each other is a pretty good indication of the accuracy of a sub $100.00 tester. I guess the next step would be aKeithley DMM7510 at ~ $4000.00.


But there is a review of theYR1030 by this guy. (The YR1035+ is the same just bigger screen and better resolution on Voltage.)
https://lygte-info.dk/review/InternalResistanceMeterYR1030%20UK.html

H
is conclusion was:
I like this meter, it is very useful for a lot of stuff, not only batteries. It can measure the resistance in a switch, a tail spring, a atomizer and a lot of other stuff.
When used to measure resistance it will give same value as an ordinary DMM except for much better resolution at low values. On batteries the value cannot be compared to the values chargers measures, but usual the datasheet value is measured this way.
His Notes:
I got the meter from Vapcell.
As reference I used my Keithley DMM7510, it can do 4 terminal ohm measurement.


image_twjiqa.jpg



Wolf
 
Wolf,

I actually learned this from @daromer but you should be able to connect two JST balance leads from your iCharger X6 to create a 4 wire system. I have done it on both my iCharger 4010Duo and 206B


image_lhaful.jpg


Just for fun... I did a quick test and wanted to share the results. Note: This Samsung 25R cell under test is kinda beat up / fatigued. When it was newer, the IR was better. (Too many high drain stress tests)

iCharger 4010 & 206B both put the IR of this cell around 60 to 65 mOhms

Manual reading (multimeter+ resistor) yielded some varying results.

With a 2.05 ohm resistor as a load, I got a IR reading of about 131 mOhms
With a 10 ohm resistor as a load, I got an IR reading of about 63 mOhms

2 ohm resistorputs the discharge at about 1C, which isn't unreasonable for this cell, but could be considered aggressive for a laptop cell.

I guess my questions is the IR measurement can vary depending on the load. Should I be testing my IR with a standard like a 10 ohm resistor to it somewhat aligns with my chargers, or should I apply a real world load and use those IR measurements.

My smaller packs will obviously see higher discharge rates than some of the powerwall projects here.
 
Back
Top