18650 DIY powerwall 48v

[Wolf]

@kje
ICR and INR cells of old were generally quite easy to differentiate. Not so much any more.
ICR chemistry does have a greater energy density compared to other chemistries but at a higher risk rate of "explosion" so it has been said.
The shift then to INR/IMR "safer" chemistries is a noticeable one
The amount of chemical doping that goes into today's batteries is phenomenal
Some manufactures will identify the ICR and INR chemistries in the part number. Some of them in there spec sheets.
Some of them not at all.
Notably any very low IR cells as in the Samsung INR18650-20S which measures at ≈ 8mΩ,. or the Sanyo UR18650RX 25mΩ factory spec, but consistently measures at ≈11mΩ. are defiantly INR
What I find is that if you have at least 20 cells of the same manufacturer and part number, when you check IR and the 20 cells have an IR of 35mΩ to 40mΩ and C/D/C at 1A results being close to 100% SoH that would be a good IR for that particular cell. Chances are its an INR.
My powerwall is basically built with LGBBM261865, LGEBM261865, LGFBM261865, and LGGBM261865 considered INR18650M26 or
for short LG M26.
The LG LGGBM261865 and LGGBM261865 had a min of 32.61mΩ to a max of 37.58mΩ out of 1172 cells and that is according to LG an INR cell
The LGFBM261865 also a "M26" had a min IR of 35.06mΩ and a max of 51.64 out of 1225 cells
Manufacturer Spec for the LG M26 is ≤60mΩ
The Samsung ICR18650-26F cells obviously claim ICR and the IR was min 50.3mΩ and max 55.59mΩ out of 1120 cells
Manufacturer spec for the Samsung is ≤100mΩ.
So you see there is a fine line between ICR and INR chemistries.
My excel database allows me to easily compare "know" manufactures cells so I can attempt to understand a bit more.
It is in excel format a work in progress but if you want to download it it is here.

Cell DBA.xlsx

1drv.ms

Wolf

 

[100kwh-hunter]

It is in excel format a work in progress

Now this is what i call data hording and keeping track of a DataBase......Thumbs up for the wolfman, please

 

[kje]

What I find is that if you have at least 20 cells of the same manufacturer and part number, when you check IR and the 20 cells have an IR of 35mΩ to 40mΩ and C/D/C at 1A results being close to 100% SoH that would be a good IR for that particular cell. Chances are its an INR.
Wolf
View attachment 31417

Ok, so 40mΩ on a INR cell can be good if C/D/C at 1A results being close to 100% SoH? And this has to happen to at least 20 cells with the same manufacturer and part number?

I have a lot of this INR cells in the picture. I assume this is the same manufacturer and part number? They are rated for 2500mah and all is tested as almost 100% capacity, from 2385 to 2638mah. I measured IR between 19mΩ and 40mΩ.
Should I use all these in my powerwall or should I set a IR threshold at some point?

My excel database allows me to easily compare "know" manufactures cells so I can attempt to understand a bit more.
It is in excel format a work in progress but if you want to download it it is here.

Cell DBA.xlsx

1drv.ms

Wolf
View attachment 31417

Impressive work Wolf. Are you still building or expanding your powerwall?

 

[Wolf]

Ok, so 40mΩ on a INR cell can be good if C/D/C at 1A results being close to 100% SoH? And this has to happen to at least 20 cells with the same manufacturer and part number?

That was sort of an example of how to get a baseline of IR and SoH on really any cell.
The Idea is to find an IR range that gives a good returning SoH on any specific part number.
That way if a particular cell is outside (generally higher IR) of that range you can expect poorer results.

The example I gave was even though the LGFBM261865 and the LGGBM261865 are "M26" cells they both have a slight different IR range for a good SoH
it is a small difference but it's there. I had many more than 20 cells to make that determination but consider 20 a bare minimum to get that baseline. In other words you may have 100 cells of the same manufacturer and part number and IR varies lets say from 55mΩ to 35mΩ and you find out any cells over 43mΩ tend to have a lower SoH mabe 95% and an IR of 45mΩ gives a 90% SoH and 55mΩ produces a 80% result.
However any cell with a 35mΩ to 40mΩ hits the 100% or so mark. So if 20 cells give the same results within a window of 35mΩ to 40mΩ you now have a baseline and know that if you have a cell of that type at 55mΩ no need to pursue it any further.
That logic is the same for any chemistry not just INR.
I have made my cell testing sheets available so here you can explore. my testing. https://1drv.ms/f/s!AmNMFw8cEOSHhvoulDNiU6KXmwolqw?e=tECcDT
The 1st_Battery was my frankenstein build with over 6000 recorded cells of all kinds of part numbers and manufactures.
This is where I started to find out the correlation between IR and SoH.
I hope this explains a bit more

Are you still building or expanding your powerwall?

Not at this time as I have my 40kWh, conservatively speaking, more like 50kWh if needed power wall in place.;
However I am always looking for some good 18650 deals where I can buy 1500 or so cells of the same kind.
so can build a spare replacement battery.

I have a lot of this INR cells in the picture. I assume this is the same manufacturer and part number? They are rated for 2500mah and all is tested as almost 100% capacity, from 2385 to 2638mah. I measured IR between 19mΩ and 40mΩ.

Ah I see the picture now. Well the more you can accumulate of the 19mΩ ones which I assume are the 2638mAh results cells the better.
I would also recommend to stick to no more than a 10mΩ difference between the low and high IR cells as in min 19mΩ and max 29mΩ.
If you are going to parallel them.
Personally I like to stick to 5mΩ difference but that may not be attainable with china cells.
Wolf

 

[Wolf]

Update I found out where they come from and the spec for IR.

Battery 3.6v 2500mah 18650 Lithium Battery Inr18650s, Lithium Battery, Lifepo4 Battery, 18650 Battery - Buy China Wholesale Battery $1.42 | Globalsources.com

Shenzhen Victpower Technology Co. Ltd is devoted to the manufacturing and supplying of China wholesale battery and related battery 3.6v 2500mah 18650 lithium battery inr18650s across the world. Our team strives to provide quality battery 3.6v 2500mah 18650 lithium battery inr18650s, always...

www.globalsources.com

Wolf

 

[kje]

Update I found out where they come from and the spec for IR.

Battery 3.6v 2500mah 18650 Lithium Battery Inr18650s, Lithium Battery, Lifepo4 Battery, 18650 Battery - Buy China Wholesale Battery $1.42 | Globalsources.com

Shenzhen Victpower Technology Co. Ltd is devoted to the manufacturing and supplying of China wholesale battery and related battery 3.6v 2500mah 18650 lithium battery inr18650s across the world. Our team strives to provide quality battery 3.6v 2500mah 18650 lithium battery inr18650s, always...

www.globalsources.com

View attachment 31435
Wolf

Yes I found exactly the same as well. Does that mean anything over 25mΩ is bad?

I'm grateful for all the good info you've gathered Wolf!

 

[Wolf]

Does that mean anything over 25mΩ is bad?

The long answer. Yes
The short answer...... bad sounds well bad. More like the cells IR has exceeded the manufacturer specifications and has reached its EOL. (End of Life).
Wolf

 

[kje]

That was sort of an example of how to get a baseline of IR and SoH on really any cell.
Wolf

What is SoH?
Is 100% SoH the same as 100% capacity?

I'm finished with the capacity testing of all the 14 120p packs. I charged to storage voltage 3.7v and let them rest some weeks. Some packs went down to 3.64v and one even 3.56v, the rest was holding 3.69v. I desoldered all the fusewires and after a day I found 2-3 self discharger cells.

I have a lot of other cells resting I can choose from but had to discharge them down to 3.7v. I used the regenerative discharge with iCharger x6 for this.

The cell holder is really tight so I had to carefully knock the old cells out with a hammer and a nail punch. Before putting the new cells inside I sanded sharp edges and lubricated inside the cell holder (with Lithium grease ). If I didn't the heat shrink would go off. Except Panasonic cells wich is a little bit slimmer.

I'm going through all the packs and see If I have to change some cells. Maybe I wasn't conservative enough when I was choosing cells in the beginning: After sitting months up to a year I choose cells over 4.09v. Well this should at least help.

 

[OffGridInTheCity]

I desoldered all the fusewires and after a day I found 2-3 self discharger cells.

FYI - Nice looking pack!

Obviously the best time to find these is before locking in a pack. When working with cells my test process ends with them fully charged - e.g. ~4.15v using an OPUS.

THEN, I let them sit 3-4 weeks (or more depending on how busy I am) and test voltages before fixing in the pack. Anything below ~4.11v is a self discharger after 3-4 weeks. This step adds a few weeks in time, but is simple to do and weeds out self-dischargers with minimum effort.

A 4.10v vs 4.11v vs 4.12v minimum is a judgement call - it will become clearer as you test 100s of cells vs the weeks sitting. You'll see a pattern of slight variances between good cells such as 4.102 to 4.109 kind of thing but the self-dischargers will noticeably deviate - maybe 4.090v or 4.070v or even worse.

Example: I happen to have some left-over cells sitting for 8 months 'resting from full charge' and just measured them. They are showing 4.102v, 4.098v, and 4.097v respectively. Since it's been 8 months, I'm OK with 4.098v instead of 4.11v minimum - e.g. judgement call based on extreme length of resting.

 

[Wolf]

What is SoH?
Is 100% SoH the same as 100% capacity?

Yes SoH is State of Health.

Wolf

 

[kje]

Yes SoH is State of Health.

Wolf

And 100% State of Health is the same as just 100% capacity or is there more involved?

FYI - Nice looking pack!

Thanks.
Have you ever tired resting all or some of your packs at storage voltage 3.7v and see if any cells is discharging while resting?

I wonder how it's possible that cells rated for 2500mah is testing a lot higher capacity? Like 27000mah? Edit: 2700mah

My liitokala testers is also not exact. I've tested the same cells at different liitokala testers and found I have to write 200mah lower capacity on some them.

 

[OffGridInTheCity]

And 100% State of Health is the same as just 100% capacity or is there more involved?


Thanks.
Have you ever tired resting all or some of your packs at storage voltage 3.7v and see if any cells is discharging while resting?

Yes, some for several years and they're still reasonable voltage such as 3.6v'is when stored at 3.7v'ish. I'm sure about retaining reasonable voltage but not sure exactly how much of a drop as I didn't keep detailed records. I generally mass discharge left-over cells to 3.7v'ish and store.

Last year, I collected the left over cells from the last 4years and built a Frankenstein pack - 14s107p

and it's doing just fine.... the last 14 blue bars ....

The brand-new cell packs I just installed in the trailer sat for over 9 months and dropped from ~4.11v to ~4.098v and now that their installed appear to be A-OK as well.

I wonder how it's possible that cells rated for 2500mah is testing a lot higher capacity? Like 27000mah?

Easy.... (I presume you mean 2500mah vs 2700mah rather than 27,000mah?)
1) My OPUSs are variable - 100mah + / - is not unusual
2) Cells of the same type vary as well - especially if they have a history - so another 100mah variance (or more) is not unusual in my experience.

*This is why I group cells for the entire 14s battery by 100mah categories and evenly distribute them thru the 14 packs - to leverage averages/statistics to get more even packs.

I have an Electrical Engineer friend who works with processors and the effects of high-speed electrical distortions on mother board tracing as in routers and processor <-> memory as examples. The math is complex and PRECISE. He just shakes his head w/respect to 'batteries' (of any kind) and say's they are MESSY!!! and doesn't want anything to do with them!

I've taken that to heart - batteries are indeed messy / non-exact but yet in a core range of metrics - deliver incredibly useful power.

 

[Wolf]

The brand-new cell packs I just installed in the trailer sat for over 9 months and dropped from ~4.11v to ~4.98v and now that their installed appear to be A-OK as well.

I know you meant 4.098v or 4.08v. I'm sure you do that just to make sure someone is actually reading your posts.
Wolf

 

[Wolf]

And 100% State of Health is the same as just 100% capacity or is there more involved?

No no difference. But using the SoH acronym makes us sound and look a lot smarter than we really are.

I wonder how it's possible that cells rated for 2500mah is testing a lot higher capacity?

Well as you stated the LiitoKala Engineer Lii-500 is not quite the most accurate tester on the market.
It is a step up from the Zanflare but not by much.
I did a test on 4 NCR18650A cells with 4 different "testers" and the LiitoKala was consistently high.

Here is an independent review of the LiitoKala

Review of Charger LiitoKala Engineer Lii-500

Wolf

 

[OffGridInTheCity]

I know you meant 4.098v or 4.08v. I'm sure you do that just to make sure someone is actually reading your posts.
Wolf

Thank you sir - corrected

 

[kje]

Easy.... (I presume you mean 2500mah vs 2700mah rather than 27,000mah?)

Yes, Thank you sir - corrected

What I do now is resting all of my packs at storage voltage 3.7v and see if any cells is discharging while resting. Packs with voltage under 3.65v after 1-2 months I desolder all the fusewires and after a day or two I find one or more cells with lower voltage than the others wich I change.

Have you tried this?

 

[OffGridInTheCity]

Yes, Thank you sir - corrected

What I do now is resting all of my packs at storage voltage 3.7v and see if any cells is discharging while resting. Packs with voltage under 3.65v after 1-2 months I desolder all the fusewires and after a day or two I find one or more cells with lower voltage than the others wich I change.

Have you tried this?

I check for self-discharge before building my packs. The natural end of the capacity test is full-charge - and I just let the full-charge cells sit for a few weeks and then measure their voltage.

 

[kje]

I check for self-discharge before building my packs. The natural end of the capacity test is full-charge - and I just let the full-charge cells sit for a few weeks and then measure their voltage.

Yes, I did that as well but still there is some sag in some of the packs. Maybe I've fixed it now with the cells I changed.

What about the sagging pack #31 of yours you added 18ah of charge to? Wouldn't that benefit from this method? Desolder - rest - change cell(s)?

 

[OffGridInTheCity]

Yes, I did that as well but still there is some sag in some of the packs. Maybe I've fixed it now with the cells I changed.

Let's hope!

What about the sagging pack #31 of yours you added 18ah of charge to? Wouldn't that benefit from this method? Desolder - rest - change cell(s)?

This was one of my early batteries where I didn't check for self-discharging before assembling - I just did capacity tests and soldered up the cells directly

As I posted, the self-discharge is soooooo slow, and a boost once a year is easy - so it's not worth it to me to take it apart.

Full disclosure - in my opinion, even if one does everything perfectly, there's still a chance down the road of getting a slightly errant pack, especially if it's a daily cycle powerwall in operation for many years.

 

[kje]

Should I have one or two copper plates between each cell?

One:

Two: