How important is matching li-ion cells in a largish pack made entirely of new batteries?

KiefRichards

100 mW
Joined
Jan 14, 2022
Messages
42
Hello there. Reaching out with a question about pack assembly. Ive searched but most info I’ve found here was regarding used li ion cells and I’m using new cells.

I’m building a 20s10p pack for a small off road motorcycle with a QS138 motor and em150 controller. I bough a 210 new Moceil P42a batteries from a seller who I think is honest 🤞.

I’m getting ready to assemble the pack, but am wondering if i need to match the cells first? Is this a needed step with new cells? Are the $50 Chinese cell testers even accurate enough for this job?

Would checking the voltage of each cell be enough? I’m not not sure if it would be wasted time and money to test each one and sort them using the program ive seen used here?

I have no one to ask in real life, so i hoping to get some input here. Thanks.
 
If you link the place you bought from, people with experience with them can better tell you if they might be good or not.

If the cells are all actually new, (not merely NOS or new old stock), and are still in their original boxes as batches and are all the same batch, they are probably already well-matched.

If any of those are not true, then it might be safer to at least do basic internal resistance testing, to be sure that all the cell groups end up with about the same Ri and so they operate about the same under load and charge.

Testing for capacity is a second test you can also do, to ensure each group is about equal capacity, but the Ri is more important for equal voltage drop or rise during current flow. As long as the Ri is ok and equal among groups, then whatever BMS you use should take care of monitoring when cell groups are empty or full, and the lowest-capacity group will determine how much capacity the total pack has and how long you can ride.

There's a few people here on ES, like Pajda,
https://endless-sphere.com/forums/memberlist.php?mode=viewprofile&u=47285
that do a lot of cell testing with various equipment, and those threads would be the best ones to use as a guide for good testing equipment (and how to use it). Most of the threads are here in this subsection of ES.

If the conditions noted previously *are* true, then you can probably just do some random voltage testing of a few cells in each box to ensure they're already voltage-matched to each other, so you aren't connecting ones in parallel with wildly differnet voltages. ;) If they are actually new and same batch, etc., they should be virtually identical voltages to the hundredth of a volt or so.
 
Thanks for response, it was helpful. Hopefully the batteries are legit and close in specs. They came in a fatory-looking box in sets of 100.

The batteries are from 18650batterystore.com

https://www.18650batterystore.com/collections/21700-batteries/products/molicel-p42a


I went ahead and order this inexpensive tester
https://www.amazon.com/gp/aw/d/B00MYQ8IYS?ref=ppx_pt2_mob_b_prod_image

I’m not confident it’s accurate, but if it’s results are repeatable, I may attempt sorting them and using a program to tell me which to group together. Or maybe not 🤷‍♂️
 
If they're genuine, you should be able to decode the 2d barcode on them to verify the text markings on them, AFAIK. (probably phone apps for that). I don't know the specifcs on decoding manufacturer dates, batch numbers, etc, for those cells, but they're popular enough out there it shouldn't be hard to find. :)

(there have been posts that say there are cells whose barcodes don't decode, and/or don't match the info they should; presumably those are not genuine cells since known-genuine ones do).

Then you can see if they are indeed new (not NOS), and if they are the same batches, etc., and if they are, then you shouldn't need much random testing to verify, which should save a lot of time, as you won't need to sort them all out to match them as groups.

Regarding accuracy of equipment, it's more important that it just be reliable and repeatable, for using for cell-matching, than that it give measurements that match other people's equipment (as long as it is at least close to "real" values :lol: ).
 
My matching and testing. 1010B
 

Attachments

  • View album “Favorites”.png
    View album “Favorites”.png
    4.9 MB · Views: 813
KiefRichards said:
I went ahead and order this inexpensive tester
https://www.amazon.com/gp/aw/d/B00MYQ8IYS?ref=ppx_pt2_mob_b_prod_image
The one you posted/purchased is only good for AA & AAA size batteries ...
81mKEB88xeL._AC_UL116_SR116,116_.jpg
... cancel or return.

You'd need one that tests the IR of 21700 Lions. There are several 4-bay models under $30. All you really need it for (IMO) is for IR comparison of your 210 cells. The IR of these will probably read high (not accurate), but you're only making cell comparisons. Hopefully all the cells should be within 2-3 mΩ of each other being new cells. The IR of the 4 bays may vary slightly from one another so just double check in another bay when in doubt of IR reading of a certain bay ...

61khGPp-szL._AC_UY327_QL65_.jpg

This is the one i used and it served its purpose. After checking all 200 cells you hopefully will find that you really didn't need to check them. What you need is a DMM that reads to 3 decimal places for checking the millivoltage of each of the 'new' cells as hopefully they shouldn't vary by more than 3 mV from one another (5mV at most) assuming all cells are from the same manufacturing run (for example 3.400V ± 0.005V) ...

616ixRGpepL._AC_SS450_.jpg

azechimp AN870 19999 Counts True-RMS Auto Range Digital Multimeter Volt Amp Ohm Meter

IMO, you're wasting your time checking capacity if in fact they are 'new' cells. IR and millivoltage comparisons are more important when checking what are sold as new Grade A cells (an occasional Grade B may be found) . If it were me (curious) I'd remove say 10% of the label wraps to see if all 200 are from the same manufacturing run. Then use adhesive insulator rings on those cells with removed label wraps and separated by label wrapped cells when assembling ...

712Xa9yDR+L._AC_UL116_SR116,116_.jpg


I wouldn't mess around testing capacity. Normally you'd only do that with unused outdated cells or salvaged cells. Assuming you're spot-welding just test the cells for IR and millivoltage comparisons. You may find that you didn't need that 3 decimal place DMM, but it will eventually come in handy.

Haste makes waste so don't be in a hurry with your build; even moreso if it's your first build.

.
 
for the new cells you need an IR tester, a real IR tester, the ones on the 4 cell testers arent accurate enough. I use the yr1030, I payed about 55 dollars for it, its the best investment I made as far as building battery packs, this tool quickly finds a bad or weak cell so you don't have to waste time doing a capacity check.
I found many cells that do well on the capacity test but don't do well on the IR tester, those cells if you use them will give you balancing problems or self discharge problems.
When first building packs I didnt have an IR tester and was always running into balancing problems. Now I recommend everyone get an IR tester, I wouldnt build a pack without one.

yr1030 IR tester
yr1030 internal resistance tester.jpg
 
jonyjoe303 said:
I use the yr1030,
<snip>
yr1030 IR tester
yr1030 internal resistance tester.jpg

any thoughts on which, if any, of these might be "real" and the right one? i see ar least 3 different designs with that mdel#, one of which looks like yours.
https://www.amazon.com/s?k=yr1030&crid=1M95W4U8KSJCB&sprefix=yr1030%2Caps%2C233&ref=nb_sb_noss_1
 
amberwolf said:
If the conditions noted previously *are* true, then you can probably just do some random voltage testing of a few cells in each box to ensure they're already voltage-matched to each other, so you aren't connecting ones in parallel with wildly differnet voltages. ;) If they are actually new and same batch, etc., they should be virtually identical voltages to the hundredth of a volt or so.
IF there are no more than 10 cells that are no more than one hundredth of a volt or so below the Molicel P42a voltage shipping norm than was good that you purchased ten extra cells instead of just 200.

You don't need to buy that azechimp AN870 Digital Multimeter if the benchmark norm for those 210 new Molicel P42a 2170 cells is a variance of one hundredth of a volt "or so" as acceptable. It won't take you all that long to check all 210 cells, if you are so inclined to do so with your DMM.
amberwolf said:
just do some random voltage testing of a few cells in each box to ensure they're already voltage-matched to each other,
Sometimes one can go overboard when testing new high energy cells for Delta OCV (unless they're new/used/outdated 30Q). The means to determine which cells are sub-standard is usually after 50-100 c/d cycles. That's why it's always better/safer to use a build assembly that can be easily assembled and disassembled to check individual cell voltages, replace bad cells, and then reassemble the pack. Even moreso, if you're using cells of questionable quality and you're not proficient at building a DIY spot-welded pack.
 
I bought mine 2 years ago (same as picture) and has been excellent. I use it to check the 18650 and also lifepo4 cells and even use it on my lead acid. I tested 100's of cells with it and has been very good, all the test are repeatable, if I test a cell and 1 minute later I test the cell again I get the same results. The good thing about these testers is you can check many cells quickly.
Other people have used the yr1030 I use and they have got good results with them. Some of the other 1030's are older versions which also work good.
The 1030 I have is rechargeable, no batteries needed, and since I've had it and all the tests I ran I still haven't had to recharge it.
When you run your tests, you want all the cells to be about the same in IR, high discharge cells are in the .30 ohms or lower, cells coming from laptops are in .70 ohms, any cells that give you too high or too low readings I would set those aside as probably bad. I seen cells in the 100's ohms.
The 1030 I bought also came with a little 18650 4 wire teststand that you can slide an 18650 into it. Thats why I payed 55 dollars for it, You don't really need the teststand, you can use the 4 wire probes only to do your tests.
 
Thanks for the input, everyone.

I ordered the above IR tester and well check them when it arrives. Will be returning that AAA battery analyzer, haha.

In the meantime, I checked the voltage of each battery.
Highest voltage: 3.550
Lowest voltage: 3.541

If I throw out the 2 highest and lowest cells
Lowest : 3.542
Highest: 3.549V

So 7mV range.

The voltage seems low to me. I thought it should be close to 3.8V. Should I be concerned about the overall low voltage?

The variation seems ok to me, but I’m not exactly sure what I’m looking for. Any input would be greatly appreciated. Thanks!
 
IIRC, lithium shipping regulations require something like max of 30% charge. So they should read fairly low in voltage; you can check the molicel charge curve for that model to see where your voltage falls on it.
 

Attachments

  • 1010B.png
    1010B.png
    493.2 KB · Views: 666
KiefRichards said:
I checked the voltage of each battery.
Highest voltage: 3.550
Lowest voltage: 3.541

If I throw out the 2 highest and lowest cells
Lowest : 3.542
Highest: 3.549V

So 7mV range.

The variation seems ok to me, but I’m not exactly sure what I’m looking for. Any input would be greatly appreciated. Thanks!
After testing all 210 P42A cells with the YR1030 you may find out that both the "Lowest : 3.541" and "Highest: 3.550V" still have **acceptable** IR (and voltage being within one hundredth of a volt of each other) ...
amberwolf said:
If they are actually new and same batch, etc., they should be virtually identical voltages to the hundredth of a volt or so.
The following P42A IR Test Chart should provide some help when rating all 210 cells with the YR1030 IR Tester. After all is said and done you'll hopefully find that all 210 cells are **acceptable**. IF say 10-15% of the cells test closer to 25mΩ than 20mΩ, (or better yet closer to 15mΩ), you may still decide to rate them as **borderline acceptable** or if in doubt build a DIY 20S9P battery pack instead of a 20S10P. Hopefully (ideally) all 210 "new" cells will test closer to 15mΩ than 20mΩ.

Depending on how "new" the cells and storage conditions over half could test closer to 25mΩ, but hopefully closer to 20mΩ (or better yet closer to 15mΩ). Never-the-less attempt to arrange the cells so that each 20S parallel group has the same IR ... which you can also verify (measure) with your YR1030 before, but more importantly after spot-welding.

7ab13bd3c3b94a0999206d8ba8beee.png

https://www.sellingbattery.com/index/goods-detail/67

Others more knowledgable than me would better advise you on what YR1030 IR mΩ test is still considered **acceptable** with P42A cells for a DIY 20S10P battery pack.
 
KiefRichards said:
If I throw out the 2 highest and lowest cells
Lowest : 3.542
Highest: 3.549V

So 7mV range.
Are you confusing IR with **unacceptable** high self-discharge when you say . . . "So 7mV range" ?
IR and high self-discharge are possibly related, but each requires different testing methods. Unless you have some very expensive equipment for testing all 210 cells it would be very time consuming and labor intensive. A task no one would welcome.

Never test a P42A cell for high self-discharge unless it's voltage is at least 3.75v or preferrably charged to 4.00v followed by several days of relaxing decay stablilizing ... then a 10-day rest before testing for unacceptable self-discharge (e.g. -7mV) as explained in the following procedure ... https://endless-sphere.com/forums/viewtopic.php?f=14&t=114417 . . . was that by chance why you mentioned . . . "So 7mV range" ?

Being that no one that i'm aware of has mentioned that P42A is prone to high self-discharge don't be overly concerned about testing your P42A cells for high self-discharge. That said it's not a bad idea to test a sampling of a few. Maybe the ones having the lowest voltage (e.g. 3.542v). It's never wasted time to take on a new learning experience if you have the time.

Before spot-welding all 200 cells (or 180 or 160) the voltages should be between 3.550v and 3.540v (and arranged appropriately).

High self-discharge isn't so much of a problem with a pack that's c/d cycled every few days. Because i don't use my Vruzend 10S3P 30Q experimental pack 4 months out of the year (winter), i have the time to disassemble and replace any 30Q cells suffering from unacceptable high self-discharge. Then during 8 months out of the year i usually cycle the pack every other day, if not every day. So even if there were any cells suffering from unacceptable high self-discharge it wouldn't be a noticeable problem in just 2-3 days.
KiefRichards said:
The voltage seems low to me. I thought it should be close to 3.8V. Should I be concerned about the overall low voltage?
Being the datasheet says the shipping voltage is 3.45V i'd be a little perplexed why the voltages are between 3.542V and 3.549V. To me that would border on being a red flag as to who may have had their fingers on those supposedly "new" P42A cells before you received them. Especially if the original shipping voltage of the 210 cells from the manufacture was truly 3.45v per cell.

If you got a good deal price wise on all 210 cells they may not be as "new" as you were led to believe. That's why it's not only important to test IR, but also a sampling of the lowest voltage cells for the possibility of unacceptable self-discharge. In the end it may be best to build a 20S9P or 20S8P pack instead of a 20S10P pack. In the long run doing so could save you a fair amount of frustration; espcially if your DIY build is spot-welded and a pain to later partially disassemble to replace one or more bad cells.
 
I've spent a couple hundred bucks now at 18650batterystore.com. The guy that runs it seems honest to me. I bought some cells that arrived the day before thanksgiving missing six cells, I first tried his phone and got an answering machine. Left a note, but decided to explain myself better with an email to support. He returned my call Wednesday night, shipped the missing cells on Black Friday, and I had them on Sunday. These were Lishen cells, and everyone came in at 3,43x. I've already made them into batteries, just in time to charge them once and then run them down to 3.5V for winter. so who knows how good they will be,

Wish I had a series R checker. I've been doing it with a 1 ohm resistor, Kelvin probe. I need to switch to a 4 ohm. I've got Panasonic 22F's from 2017 that still have 2400 mah like new, but the estimated series R with a 4A draw is like 150-200 milliohms, so these are only good now to power flashlights.
 
Emark, thanks for pointing out my error in checking for high self discharge without charging then first. So the delta I measured is meaningless.
I too, wondered why the voltage is higher than the shipping voltage on the datasheet. I just rechecked it actually shows “>3.45” so apparently it’s ok.
I think I’ll just check the IR of each cell and group them so each parallel bank has the same IR, or as close as I can get. Is there a program for doing this or do I just have to figure it out on Excel?

Also, is there an ideal SOC for testing the IR on this type of battery?

Thanks
 
KiefRichards said:
In the meantime, I checked the voltage of each battery.
Highest voltage: 3.550
Lowest voltage: 3.541
Seems to be a conflict of interest between your previous post and the following post :wink:
KiefRichards said:
I too, wondered why the voltage is higher than the shipping voltage on the datasheet. I just rechecked it actually shows “>3.45” so apparently it’s ok.
Yes, always a good idea to make sure you have a fresh battery in your DMM before checking cell voltages :wink:

KiefRichards said:
I think I’ll just check the IR of each cell and group them so each parallel bank has the same IR, or as close as I can get.
Spot-weld all twenty of your 10P parallel groups first. There looks to be enuf cable length on the YR1030 probes to check the IR of each of the twenty 10P spot-welded groups before spot-welding the series bus bar connections ...

file.php


KiefRichards said:
Also, is there an ideal SOC for testing the IR on this type of battery?
Do you mean the cell or actual 20S10P battery? What's important is that the SOC is the same for all cells (P42A shipping SOC = "3.45v") when testing IR of "new" cells. Look at the previous IR chart for P42A (or any Li-ion battery). In other words not when the cell or "battery" is discharged with cells as low as 3.20v and not when the cell or "battery" is near fully charged with cells above 4.00v(IMO).

If you haven't already -- check out this youtube ... https://www.youtube.com/watch?v=ublFs7IU05c

There are other ES helpful threads if this is your first DIY build. I wouldn't start a DIY spot-weld build until i was confident i knew more than enuf and even some more before taking on the task of your first build; especially when its a 20S10P 2170 battery pack. Which means looking at a lot of youtubes and reading alot of related ES threads.
 
Thanks for the vid link. Always enjoy watching other’s builds.
I’ve built a 4s1p and a 3p4s as practice with my DIY MOT welder and 0.15mm nickel strips. I bought some 0.2 nickel but my welder couldn’t make good welds with the thicker nickel. I just received my Kweld yesterday and am waiting for the lipo to arrive for it.
I plan on using the “nickel sandwich” method with a 0.15 nickel-plated steel and 0.15 copper. Im considering making a 4s1p with the new cells with the copper/nickel method and testing it at 35-40A before I start on the big battery.
Re the actual layout of the big pack; I have and idea how Ill handle it. I’d like some opinion about it. Would it be better to start a battery build thread in this sub, or just post it here?

Thanks!
 
Wasn't sure what you meant by "battery" in previous post -- so just added the underlined portion in my previous post after your recent post.
KiefRichards said:
I’ve built a 4s1p and a 3p4s as practice with my DIY MOT welder and 0.15mm nickel strips. I bought some 0.2 nickel but my welder couldn’t make good welds with the thicker nickel. I just received my Kweld yesterday and am waiting for the lipo to arrive for it.
I plan on using the “nickel sandwich” method with a 0.15 nickel-plated steel and 0.15 copper. Im considering making a 4s1p with the new cells with the copper/nickel method and testing it at 35-40A before I start on the big battery.
Ni-plated steel is cheaper, but is it the right choice for your 20S10P P42A DIY build ??
KiefRichards said:
Re the actual layout of the big pack; I have and idea how Ill handle it. I’d like some opinion about it. Would it be better to start a battery build thread in this sub, or just post it here?

Thanks!
Probably best to start another thread. However, most of your DIY build questions can be resolved by first reading other ES threads and watching youtubes. When you've done your due diligence and exhausted all other resources then it may be time to pick the brains of the top ES crème de la crème by starting (IMO) another thread. Members that are much more knowledgeable and experienced than me.
 
Regarding the ni-plated steel, I’ve read here that it works better for the copper/nickel sandwich method than using pure nickel. I’ve got both pure and coated strips to try out. If both have the same results, I’ll use the pure nickel on top the copper. Thanks again!
 
Got my 1030 IR tester.
Of the 210 batteries, all were between 8.74-9.17m ohms.
After removing the 5 batteries with the highest resistance, all are between 8.74 and 9.09m ohms. Most are around 8.85m. Now I just need to sort them into parallel groups to get 10 p groups with matching IR values.
This has been an eye opener for me. I never realized these cylindrical lithium cells have such high IR. The lipo for my kweld, for example, has only 3.1m ohms of resistance at just 5ah. My assembled 10p20s should have about 19m ohms when completed.

If my math is correct, a 19m ohm battery at 300A will produce 1700W of heat. Crazy. I see why some people use lithium pouch cells. 170W of heat beats the hell out of 1700.

Also assembled a 1p4s test pack with copper/nickel sandwich method and tested at 33A load. Batteries reached 50c in 1min and 50 seconds. That’s pretty damn fast. Im starting to worry i might have heat issues once its done. Especially since its over 100f here for months in the summer
 
Back
Top