cell resistance matching

Joined
Feb 26, 2018
Messages
154
It seems when building a lithium battery the cells needs to be match according to their internal resistance.

I would like to see some guides, web pages, videos etc. that shows how this is done.

Seems like i can't measure it with a regular multimeter. Or can it?

I see some chargers/testers can are able to hold 4 cells and show the capacity and resistance. I guess that is what is needed but would like to know if it can be done without such tools.

Do people use such a tester what is able to hold 4 cells if they build a battery with 200 cells?
Guess it must take a long time getting the measurements.
 
For IR testing you need a yr1030 or something similar (which I own). Those testers produce repeatable results everytime. The opus 4 cell tester can do IR testing but the result are all over the place. I wouldnt trust it.
With the yr1030 you can test 200 cells in less than 30 minutes, it goes quick. You just put the probes on the positive/negative of the cells and it gives you a quick reading and move to the next cell.

The price of the yr1030 is about 50 dollars but its well worth it, without one your practically building a pack blind. I didnt want to spend the money but it has save me alot of time in weeding out bad cells. Some cells have good MAh and look in good condition but have high IR, if you used them in a pack you will end up with balancing issues and self discharge packs, all it takes is one bad cell.
Always test the IR first (quick test) before you check for mah (time consuming), that way you don't waste time on high IR cells.

My 4 must have tools for building packs
1. IR tester
2. small side snips to remove tabs from cells
3. opus btc3100 4 cell tester
4. malectrics 12 volt tab welder

IR tester is number 1, the cell has to pass that test first before I do anything with it.

You want to have all the IR close to each other. High performance cells (high discharge are in the 20milliohm or less, laptop cells are in 50 milliohm range but I seen some up to the 80 miiliohm range for old cells. Bad cells will be 90 milliohm or higher and you will find some in the 100's those are automatically used for flashlights if used at all. After testing IR you get an idea which cells are good and which not use in a pack. If the majority of cells you testing are between 50 and 70 but 1 or 2 are in the 80 and 90's, you keep the good ones and not use the high IR ones.

yr1030 internal resistance tester.jpg
 
I use a very good charger, tells me IR to the fourth decimal point. I can test up to 32 cells ( or groups of cells) at once, up to 40A each,.


IR must ONLY be taken UPON CHARGE from 30% SOC.

IR must ONLY be taken at ROOM TEMPERATURE.

IR of the cells must be WITHIN 10% of each other to make a good pack.

That means if you have a 1mOh cell ( 0.001mOh), the ones in the pack must not be greater than 0.0009 Ohm-0.0012 Ohm ( 1.2mOh - 0.9mOh)...

If you have a 10mOh cell... match on 10%... use cells that are matching that ( 12mOh max to 9mOh min)...

I aim to use a median average to group the IR in the packs. NOT a mean average.
 
larsb said:

I’d like to know why?

Cause it is the instruction booklet of my 1200$, 4 channel, 32 cell, PID loop, 48v input, 5700w output, C-Rate determining, charging system.

Ask the experts at FMA DIRECT, for an accurate answer. I suspect it is something to do with repeatability. Accur5acy. Congruent reproducibility.
 
larsb said:

I’d like to know why?

Cause it is the instruction booklet of my 1200$, 4 channel, 32 cell, PID loop, 48v input, 5700w output, C-Rate determining, charging system.

Ask the experts at FMA DIRECT, for an accurate answer. I suspect it is something to do with repeatability. Accur5acy. Congruent reproducibility.
 
larsb said:
That’s a hefty amount of $ for a kit, do you build packs for a living?

Not a full time occupation,.... but.... I do have a waiting list of 10 people, waiting, for the chance to buy and run a pack i build.... and I am building one right now. For a customer/friend. Full Super73 build with a 5kW hub motor and a Chevrolet Volt cell pack. Streetbike tires on 20" rims, and highest quality parts everywhere. New cell, 1800wh total on the bike. Very blingy, with expensive trinkets everywhere. Recaro seat, Pro taper bars, Super73 frame, DNM double disk with Magura MT5e. Total cost of the 1800wh 72vn pack of new lipo cells will be about 600$ with BMS. 6AWG lines and a QS8-S for the contact of the main discharge.

Yes I do build packs for money. However, such a small scale that I become close to the buyer and they are my friends in the end. Proballly one every three months. Yes, for cost, I do offer a used EV cell to build with, and to do so, I must have a good, fast, powerful well accepted datalogger to datalog every cell in teh pack I build, both preassy baselines and post assy test logs. I cannot sell a friend a pack that has not proven itself.

I learned I needed a good one the first time I tried to datalog a 30kWh Solar storage bank for a customer/friend. YOu cannot, in reasonable time, datalog huge cells in any sort of timely manner without good datalogging dischargers. they are fully regen capable, I do not waste the power, and they themselves are powered by the cells I datalog. I run about 4kWh usually to power the 1-2kWh discharge log.. shuffling power, not wasting it. I can datalog up to 999aH and the log can be a maximum of 99 days long.

This is my little datalog station. 5700w maximum.

I do have the equipment to datalog ( to .xls) upwards of 800?v and 1000?A ( only not at the same time... (up to 30kW max...)

I am more like a custom ebike builder/mechanic. The battery is just part of the equation. I must be well equipped if I tell people " I know what I am doing"... good tools. I am an excellent logic troubleshooter.
 

Attachments

  • 305956630_537824234773985_503813225817425851_n.jpg
    305956630_537824234773985_503813225817425851_n.jpg
    647.2 KB · Views: 2,335
  • 307382551_793452611988526_1645496436669313484_n.jpg
    307382551_793452611988526_1645496436669313484_n.jpg
    474.9 KB · Views: 2,335
The YR tester mentioned above is perfectly capable of measuring accurate IR. As for the 30 degrees measurement, we know that resistance is affected by temperature. It's not so much that it needs to be a certain temperature, but rather, measuring a cell twice at 2 different temperatures will give you different results. The tester that DogDipstick uses may very well get the most accurate measurements at 30 degrees, but whatever tester you do use, as long as you're measuring all your cells consistently, and grouping them as such, you'll be in the clear.

Check out Wolf's google drive here: https://secondlifestorage.com/index.php?members/wolf.11978/#about, he's a madman when it comes to IR and its relation to cell health.

I also saw this good idea to make a little caddy for measuring 50 cells at once. Pretty easy to build if you have scrap wood laying around.
index.php

Then when you test a cell that has a higher IR than what you want, poke it off to the side and remove all the bad ones when you finished the whole set.
index.php
 
When building a battery from salvaged cells IR to the nearest isn't possible. An IR range of 5mΩ (i.e. 30mΩ to 35mΩ), as was the case with the thirty Samsung 33G unused cells from BatteryHookup at $2. So far no need to bottom balance as all thirty cells (10s3p) are within 5mV of each other after charging (41.0V), and after discharge (35.0V).
 
Are you sure about the scale of the measurements? Seems like pretty high cell resistances if you're measuring whole ohms or tens or more ohms of resistance per cell.

I don't think I could live with a typically sized pack of cells that bad, the voltage sag and wasted power would be impossible to deal with; I'd have to overbuild the packs by a couple of factors of ten to get decent performance. :(

The EIG C020 cells I use are rated at something like <3milliohm each, so my entire 14s2p pack resistance would be only about 21milliohm for new cells (which mine are far from). I think the Cycle Analyst v3 I'm using estimates around 45milliohm for the pack (but I don't have test equipment to verify this), and I expect that also includes the resistance of my cabling, shunt, battery disconnect, breaker, and connectors, but I also don't know that either.

A quick google of 18650 cells shows a wide-ish range of values but generally seem to be in the range of 50-100milliohm each. A similar 14s2p pack of those would then be 700milliohm, assuming that 100milliohm per cell value. A more realistic 14s 4p or even 8p would be half or 1/4 of that resistance, so 350 or 175 milliohm. (not counting interconnects/etc, just cells).

An image from Lgyte.dk for a bunch of different cells:
https://lygte-info.dk/pic/Batteries2012/common/Ri.png
Ri[1].png



SOmething else to note: The number of decimal places of a measurement's resolution depends on the selected range. If you have a 2-decimal place resolution measurement and are set to ohms, you can read differences down to the tens of milliohms. If you are set to milliohms, you can read differences down to the tens of microohms, sometimes single microohms. If in the a 200ohm range, it can only read down to single ohms at best. (each meter should have documentation on it's specific limitations in each range for accuracy and precision, and how much resolution to expect, which can increase the lower the range is set to).

A useful page
https://www.fluke.com/en-us/learn/blog/digital-multimeters/accuracy-precision



eMark said:
When building an 1865 or 2170 grade AA powerpack one would think an IR tolerance of just 1.0Ω would be sufficient. In other words the umteen cells are all within +- 0.5Ω of each other. Are you inferring a 72V DIY powerpack of umteen AA cells are all within 0.5Ω (500mΩ) of each other not to mention within 0.05Ω (50mΩ) as possible.

With your high quality standard (4th decimal place) what % out of original umteen cells in your DIY 1865 or 2170 were within even 500mΩ (1 decimal place) of each other, let alone within 50mΩ (2 decimal places) of each other? Can even the best 2170 TESLA cells have that tight of an IR tolerance within 50mΩ (2 decimal places) of each other ?

This begs the question what brand cells (e.g. Samsung, Sony/Murata, Molicell, Panasonic) do you consider capable of achieving your benchmark of such tight IR tolerance? I ask this because more than once you have inferred that cylindrical cells are junk in comparison to your VOLT pouch cells. Are you in affect saying that the quality of todays 1865 & 2170 best grade A cells are or aren't capable of such tight IR mΩ tolerances (to second or even third decimal place) in an antomotive A grade manufacturing run?

Take it jj's standard is for DIY builds using salvaged cells costing no more than $2 (at most). Will need 65 good enuf cells for a 13s5p build with IR tolerance between say ... 35Ω to 45Ω. So buy 100 assuming 35 are over 45Ω, and you may need to extend the IR range from 40Ω to 50Ω for 65 good enuf cells.

When building a battery from salvaged cells IR to the nearest Ω isn't possible. An IR range of 5Ω (i.e. 30Ω to 35Ω), as was the case with the thirty Samsung 33G unused cells from BatteryHookup at $2. So far no need to bottom balance as all thirty cells (10s3p) are within 5mV of each other after charging (41.0V), and after discharge (35.0V). Only ten c/d cycles so far with my experimental 10s3p 33G. All 30 cells from same manufacturing run ... https://batteryhookup.com/products/30-100-new-samsung-inr18650-33g-3150mah-18650-cells ... time will tell, but so far everything looks favorable going forward.
 
jonyjoe303 said:
You want to have all the IR close to each other. High performance cells (high discharge are in the 20milliohm or less, laptop cells are in 50 milliohm range but I seen some up to the 80 miiliohm range for old cells. Bad cells will be 90 milliohm or higher and you will find some in the 100's those are automatically used for flashlights if used at all. After testing IR you get an idea which cells are good and which not use in a pack. If the majority of cells you testing are between 50 milliohm and 70 milliohm, but 1 or 2 are in the 80 and 90's, you keep the good ones and not use the high IR ones.
amberwolf said:
I don't think I could live with a typically sized pack of cells that bad, the voltage sag and wasted power would be impossible to deal with; I'd have to overbuild the packs by a couple of factors of ten to get decent performance. :(
Agree, wouldn't build a pack with cells' IR ranging from 50mΩ to 70mΩ. That said you're well aware (and then some) it depends on the battery application ... from a flashlight to a power packed ebike speed demon 8)

My unused $2 Samsung 33G cells ranged from 30mΩ to 35mΩ, but that seems too low as the Samsung 33G datasheet lists IR at ≤ 45mΩ ...
7.4 Initial internal impedance
Initial internal impedance measured at AC 1kHz after Standard charge.
Initial internal impedance ≤ 45mΩ (bottom of p. 4) ... https://www.imrbatteries.com/content/samsung_33G.pdf
My application is about halfway between a flashlight and a powerpacked ebike speed demon :bigthumb:
 
eMark said:
amberwolf said:
I don't think I could live with a typically sized pack of cells that bad, the voltage sag and wasted power would be impossible to deal with; I'd have to overbuild the packs by a couple of factors of ten to get decent performance. :(
Agree, wouldn't build a pack with cells' IR ranging from 50mΩ to 70mΩ. That said you're well aware (and then some) it depends on the battery application ... from a flashlight to a power packed ebike speed demon 8)
Sure, but in context of this thread (and generally ES as a whole, considering battery usage here overall) the battery packs will need to sustain loads of say, 10A-50A (with some needs outside that range by varying amounts).

So any cells that are in the actual whole ohms range as your post was discussing would not be usable (without massive paralleling of hundreds to thousands of cells per group), in those applications.

(and I've so far never run across cells that have such high resistance--not even examples of defective cells that were tested here on ES, for instance, much less still-operable ones).

Also, I wasn't referring to the *range* of resistance, but the *amount* of resistance, which in the post I quoted was factors of ten (or much more) beyond what is actually seen in typical cells. (whole ohms to tens of ohms vs milliohms)


My unused $2 Samsung 33G cells ranged from 30mΩ to 35mΩ, but that seems too low as the Samsung 33G datasheet lists IR at ≤ 45mΩ .
..
is important. It means that that value must be less than or equal to the specification.

So there's nothing wrong with a cell that's better than the specification, especially if the results from a group of them are in a fairly narrow range (narrower, the better).

To verify the tested value, comparison with groups of other (new) cells tested vs their specifications is useful, since a trend toward values all significantly less than the spec value (about 2/3 in this case) indicates a testing methodology or equipment problem.

Also depends on the testing methodology used on the spec sheet vs that used in a user-level test. Any condition that is different may affect the results, including temperature, test equipment type or connection types, etc.
 
amberwolf said:
So any cells that are in the actual whole ohms range as your post was discussing would not be usable (without massive paralleling of hundreds to thousands of cells per group), in those applications.
amberwolf said:
Are you sure about the scale of the measurements? Seems like pretty high cell resistances if you're measuring whole ohms or tens or more ohms of resistance per cell.
Back checked and noticed a few times i posted only ohms (Ω) instead of milliohms (mΩ). Have corrected the typos so it now reads as follows (see underlined corrections) ...
When building a battery from salvaged cells IR to the nearest isn't possible. An IR range of 5mΩ (i.e. 30mΩ to 35mΩ), as was the case with the thirty Samsung 33G unused cells from BatteryHookup at $2. So far no need to bottom balance as all thirty cells (10s3p) are within 5mV of each other after charging (41.0V), and after discharge (35.0V).
My mistake so decided best to delete confusing post leaving just the above correction (previously was Ω instead of as now shown). Makes a big difference (Ω vs mΩ) ... thank you for calling me out :thumb:

Thank you for pointing out my error ... hopefully you surmised i meant milliohms instead of ohms.
amberwolf said:
So there's nothing wrong with a cell that's better than the specification, especially if the results from a group of them are in a fairly narrow range (narrower, the better).
Chances are that my IR measurement method (30mΩ to 35mΩ) was too kind to me :D Maybe IR decreases with unused stock that's in storage for a few years :wink: (just kidding)
jonyjoe303 said:
The opus 4 cell tester can do IR testing but the result are all over the place. I wouldnt trust it.
Likewise for the 4 cell MiBOXER that i used for testing IR. I wanted to believe it even though i knew better.
 
So i get the tool to measure the internal resistance of each cell.
Then what do i do?

Would like to see it done in videos as the first part of building a lithium battery but seems like they never perform the cell resistance matching part.
 
scootergrisen said:
So i get the tool to measure the internal resistance of each cell.
Then what do i do?

Would like to see it done in videos as the first part of building a lithium battery but seems like they never perform the cell resistance matching part.
Checkout this youtube ... https://www.youtube.com/watch?v=1lJSJWCFQl0 ... it doesn't take a rocket scientist to see that testing each cells' IR is just as important as testing each cells' voltage and capacity when building a DIY battery using salvaged cells of questionable quality. When using salvaged cells it's mandatory to use an IR Tester as well as a good BMS. One's particular application (i.e. degree of battery drain MCD demand) will determine the allowable IR, IR range as well as voltage and capacity range.

https://www.aliexpress.us/item/2251832669676318.html?gatewayAdapt=glo2usa4itemAdapt&_randl_shipto=US
https://www.amazon.com/Jieotwice-YR1030-Battery-Internal-Resistance/dp/B09LQZ95SD?th=1
https://www.amazon.com/dp/B07R8VCJPC/ref=sspa_dk_detail_2?ie=UTF8&psc=1&pd_rd_i=&pd_rd_i=B07R8VCJPCp13NParams&smid=ABMWZP7NSODS0&s=industrial&sp_csd=d2lkZ2V0TmFtZT1zcF9kZXRhaWwy

IMO an IR tester (as shown above) is really only mandatory for a DIY build using salvaged cells as explained in previous post by jonyjoe303 (e.g. 50mΩ to 70mΩ) ... and when using Chinese cells and unused name brand overstocked cells a few years old costing $3 or less.

With new A grade brand name cells (Samsung, LG, Molicel, Panasonic, Sanyo, Sony/Murata, Lishen, BAK) IR testing probably isn't even necessary other than to see if your IR tester reads approximately the same IR as the cells' datasheet IR rating. If not chances are it's your IR tester that's slightly off rather than the datasheet of all the A grade reputable brand name cells.

With salvaged cells you've got to figure that the cells' original MCD rating is less than when new. That's why most DIY builds using salvaged cells are at least 5p. Controller amp rating should be less than the Maximum Continuous Discharge (MCD) of the cells' datasheet rating with a DIY build using salvaged cells. For example with a 20A Controller each parallel group should be rated at less than the cells original MCD. For example if original datasheet cell is 10amp MCD figure only 30amp MCD with a 4p parallel group.

Bottomline: Would never build an ebike battery out of no name Chinese cells or Chinese manufactered knock-off brand name cells priced to good to be true.
 
scootergrisen said:
Would like to see it done in videos as the first part of building a lithium battery but seems like they never perform the cell resistance matching part.

Depending on application some YouTubers are okay with either not testing IR at all or using high IR secondhand cells. For example if you build a powerwall big enough, with enough cells in parallel, perhaps each 18650 is only being asked for .5-1A each. Compared to EV builds, where amp draw is often in the 10s, high IR might be acceptable. For my stationary DIY powerwall using salvaged free laptop cells, I accept any IR under 80miliohms.

Which is to say, even on my powerwall build, I still am checking IR, I still use a BMS with balance, and I still toss or set aside any cells above a certain IR.
 
How do i figure out what the internal resistance should be if i want to make a battery for an e-bike with:
48 volt system
1200W controller with peak current 140A
original battery has 13S10P cells
2000W motor

Thinking about making a battery with twice the amout of cells. So 13S20P if i can fit them in the original battery box.
 
DogDipstick said:
IR must ONLY be taken at ROOM TEMPERATURE.

IR of the cells must be WITHIN 10% of each other to make a good pack.


if they are below 50 do you still need to match them within 10pourcent of each other?
imean there is already the capacity and voltage to be matched if we add the 10 pourcent IR it become quite a puzzle

if they are all below 50 can we just forget the within 10poucent and consider they are all matchable leaving us with just having to manage the voltage and capacity to do the matching?


Also i cannot find a translation sufficient to understand the expression to be taken upon charge from.
in IR must ONLY be taken UPON CHARGE from 30% SOC.
could you express it in other words please?

if it means that the mesurement of the ir have to be made at 30pourcent of the total charge, what about if the mesurement is taken at 80 pourcent of charge or full or something like that ? does it make such a big different cause that would mean i have to discharge the cells .
How is someone able to bring the cells down to 30 pourcent anyway( ido have a opus3100 but dont know how to use it if that is the equipment needed to bring them down.
 
sharinginfos said:
if they are below 50 do you still need to match them within 10pourcent of each other?
imean there is already the capacity and voltage to be matched if we add the 10 pourcent IR it become quite a puzzle

if they are all below 50 can we just forget the within 10poucent and consider they are all matchable leaving us with just having to manage the voltage and capacity to do the matching?


Also i cannot find a translation sufficient to understand the expression to be taken upon charge from.
in IR must ONLY be taken UPON CHARGE from 30% SOC.
could you express it in other words please?

if it means that the mesurement of the ir have to be made at 30pourcent of the total charge, what about if the mesurement is taken at 80 pourcent of charge or full or something like that ? does it make such a big different cause that would mean i have to discharge the cells .
How is someone able to bring the cells down to 30 pourcent anyway( ido have a opus3100 but dont know how to use it if that is the equipment needed to bring them down.


If a cell ( the average cell in the pack) shows 50mOh.. the acceptble range is 45mOh to 55mOh. The pack will last a long time with a cell choice of 45mOh to 55mOh . (10%)

If the cell ( the average cell int eh pack) shows 5mOh... the acceptable range is 4.5mOh to 5.5mOh... (10%)

IF the cell ( the average cell in the pack) shows 1 mOh... the acceptable range (of the other cells in the pack) is 1,1mOh to 0.9mOh..... (10%).

IR, when measured on a full pack, full State Of Charge ( SOC) the IR may be higher than the real IR. The resistance of a battery cell gets really high at the TOP of the charge and the BOTTOM of the charge.. hence they dont take any more energy at the TOP (high resistance) and bottom they cannot give any more energy ( low end of the scale... ) .. so we know that resistance increased at the beginning and end of the charge of the cell... So we use the middle of the SOC to make measurements because the SOC will not make a difference in reading here.. IF you do it at the top or bottom, the natural resistance of the cell design will show skewed reading. However, the differences may not be that big, unless you want an absolute reliability. It is just a matter of making repeatable measurements on a congruent platform.

I have very complicated charging / discharging / cycling battery testing equipment that costs ($) about 1000x more than the Opus.

oPUS = 12$... A very poor discharge tester .
mY DATALOGGER = 1200$ An excellently powerful and smart datalogging 4 channel 5700w cell cycleing system......

SO I USE IT FOR VERY ACCURATE, RELIABLE, REPEATABLE TESTING. It can auto mate a charge on a group of cells and calculate the energy and % SOC the cells are at by software. I set a %, it discharges the cell to there.
 
scootergrisen said:
How do i figure out what the internal resistance should be if i want to make a battery for an e-bike with:
48 volt system
1200W controller with peak current 140A
original battery has 13S10P cells
2000W motor

Thinking about making a battery with twice the amout of cells. So 13S20P if i can fit them in the original battery box.

The more powerful a pack ( the lower IR of the pack) the more power you can make. The higher IR directly translates into heat ( watts of waste dissipate energy). Lower IR directly translates to more power availible.

Look at compliance voltage.
" Compliance voltage is the range of output voltage of a constant current power supply (battery), over which the load regulation (controller) is within certain limits. It represents the maximum voltage a current source ( battery) will reach as it attempts to produce the desired current ( in the controller)."

AKA " Sag".

You will have a much higher compliance voltage per load current with the lesser IR pack ( cause the electricity can flow, ie, is not impeded, by resistance).

IF you take a 10p pack and make it a 20p pack, you essentially have a much lower IR of that 20p group vs the 10p... and the cells can output less per cell, to make the same power, and live longer because they are making less heat.

It really depends on if you are after power or range. With a 1200w controller, the 20p would be more adding capacity and grange ( and be much more punchy off the line) butif the controller is only 1200w, the 20p of power in a group is not necessary.. overkill. A 20p pack will laugh at most anything you throw at it... ebike size. Sag will be very little for any output current. Much less than the 10p pack.
 
scootergrisen said:
I'm all about the range.

Then, while IR is still important, range is going to be determined by how many cells in parallel you can fit into your space. You already mentioned that you want 140a max discharge, original battery was 10p, you want to fit 20p in the same space. If you can, great; 7a max draw per cell is very doable for a lot of brands.
 
I'm just saying what i have. The vehicle's controller is "1200W controller with peak current 140A".
 
A cheaper alternative is to just use a 1-5 ohm resistor as a load over a cell and calculate using ohms law, if you have a multimeter (not necessary to calculate if using 1 ohm load) and then calculate resistance using the voltage potential difference using Kirchhoffs law.

There's a good explanation in this video: https://www.youtube.com/watch?v=av38iBxcOgQ
 
Back
Top