Looking for opinions: Charge capacity test vs Discharge test


Solar Batteries, Panels, Chargers, Inverters & Mounts at www.batteryhookup.com Save 5% with coupon code "Powerwalls"

harrisonpatm

Member
Joined
Jan 5, 2022
Messages
420
Typically when we test a cell, we'd perform a charge-discharge-charge test, in which capacity is recorded during the discharge cycle. This makes sense as it's possible that the cell in question may have minor-to-major self-discharge issues, so we'll get a better idea of the cell's capacity if we can be sure that the cell is 100% full voltage, and perform a full discharge immediately afterwards.

Here's my hypothetical. Let's say I have a cell at 2.5V. The manufacturer's specs list this as a 3200mah cell. I charge this cell. It took 3200 mah to reach 4.2V. Can this be taken as a capacity test as well as if I had done a discharge test instead? Can I treat this as a 95-100% SOH cell? (Spoiler alert: it's not a hypothetical, I have acquired a large batch of the same manufacturer's cells and am wondering if I should alter my normal testing procedure for just this batch of several hundred cells)

Assume that I will also perform an IR test and check that against original specs as well, and assume that I will hold my 4.2v cell for 10-30 days and check voltage again after that time, to check for self-discharge issues. Because of course I will. My inquiry is wondering whether I can trust capacity recorded during a charge cycle rather than the usual habit of doing it on a discharge cycle. Looking forward to opinions.
 
The core issue (to me) is not whether you test by discharging or charging but to have a *consistent* test so that tested capacities are apples to apples.

My goal is always to classify / sort the cells by capacity so I can evenly distribute them thru the battery to 1) reach the battery designed AH by adding up the capacities of the member cells and 2) so that each cell/pack in series has a very similar makeup to eliminate operational issues such as balancing.

Therefore, if every single cell is exactly the same 2.5v then a charge capacity test to 4.2v makes perfect sense - e.g. each capacity test is identical so you get apples to apples capacity measurements. For me, all my cells have varied in their initial voltage when I'm ready to test so I have no choice but to charge, discharge to do the test.
 
Last edited:
For me, all my cells have varied in their initial voltage when I'm ready to test so I have no choice but to charge, discharge to do the test.
In this case, for this particular batch, every single one is resting at 2.5v.

not whether you test by discharging or charging but to have a *consistent* test
That's why I'm considering a different procedure for just this batch and this batch only. My normal procedure is charge-discharge-charge tests, because same as you, I normally get cells in all kinds of voltages. So my normal haul cells will still follow the same procedure as always. Just think I'm going to go ahead and do this different procedure and compare the 800 cells against each other.
 
compare the 800 cells against each other.
Sure but as mentioned above, this won't be apples to apples with future cell acquisitions.

No judgment! just discussing 'the why' charge/discharge became the common procedure for capacity tests to get consistent results over time and different batches of cells.
 
Alright then, here's an extension to this premise that you're really gonna hate and disagree with!

I have hundreds of these cells. They're all totally discharged, so they need to be brought up slowly. I have an 80p rig. I slot in 80 of these cells, and charge them all at once, 200ma per cell, using power supply that tracks mah.

80 cells x 3.2 rated amp-hours per cell = 256ah. My 80p rig charged all the cells from 2.5v to 4.18v It took 258.5ah. Therefore, all 80 cells are at or very close to their original capacity. And don't need to be individually capacity tested. Right?!?

I know you guys will disagree with this concept of not testing cells individually. My thinking is, if the first charge came back with 240, 230, 225ah, then yeah, I would need to break down and test them individually. But I have already individually tested a batch of 312 cells previously, all from the same batch, and they all came back above 95% SOH. So I'm thinking I might roll the dice with this testing method.

The cells will be separate from my main powerwall, this is for another motorcycle conversion project. If the cells fail, it's on me to rebuild the pack and I'm willing to take that risk.
 
To me, it's not about right and wrong :)

It's a matter of risk level to achieving an operational battery that meets short/long-term goals and what you do about it. If you're willing to fix/replace trouble packs till you get a well behaved battery per you're needs - then there's nothing wrong safety wise etc. It's just work at the back end vs work at the front end type of DIY decision.

Look forward to reading how it works out.
 
Last edited:
So far it's just one group of 80p tested at 100% estimated capacity. If i get through all 11 groups with this testing procedure and they all test at the same capacity, that should be its own indicator that the whole lot is in decent health. We'll see.
 
Just my 2cents.....
Capacity testing during "charging" event has a lot of variables that can come into play to throw the values off. Resistance in the contacts, Resistance in the wires, Resistance in the cell causing slight heating, can all throw the values off as "more" energy is delivered than is actually stored in the cell.

Capacity testing during "discharging" event has the least amount of variables as the cell will only discharge until it reaches ending voltage (either dead or set disconnect). Sure the resistance in the wires/contacts can still throw the values off, but nowhere near as much as charging.

Let's say I have a cell at 2.5V. The manufacturer's specs list this as a 3200mah cell
IIRC, most, if not all, manufacturers will get their rating during a "discharge" test. If the cells are new, or new-ish, you should be able to get the same value during a charge test as long as the other factors don't interfere with delivery. The other key point is that you need to know what the manufacturer has decided is the cutoff voltage for their testing. Usually this should be what they designated as the bottom voltage.

Me personally, I'd do the Charge/Discharge/Charge method for capacity testing. Charge to full, then Discharge to get capacity value, then Charge to get 2nd capacity value and compare with the Discharge.
With a good tester, you should be able to accomplish this all automatically and log the data. Or even make one from arduino's, TP units, other bits and lines of code.

Ok, maybe 4 cents instead of 2 :p
 
Just my 2cents.....
Capacity testing during "charging" event has a lot of variables that can come into play to throw the values off. Resistance in the contacts, Resistance in the wires, Resistance in the cell causing slight heating, can all throw the values off as "more" energy is delivered than is actually stored in the cell.

Capacity testing during "discharging" event has the least amount of variables as the cell will only discharge until it reaches ending voltage (either dead or set disconnect). Sure the resistance in the wires/contacts can still throw the values off, but nowhere near as much as charging.


IIRC, most, if not all, manufacturers will get their rating during a "discharge" test. If the cells are new, or new-ish, you should be able to get the same value during a charge test as long as the other factors don't interfere with delivery. The other key point is that you need to know what the manufacturer has decided is the cutoff voltage for their testing. Usually this should be what they designated as the bottom voltage.

Me personally, I'd do the Charge/Discharge/Charge method for capacity testing. Charge to full, then Discharge to get capacity value, then Charge to get 2nd capacity value and compare with the Discharge.
With a good tester, you should be able to accomplish this all automatically and log the data. Or even make one from arduino's, TP units, other bits and lines of code.

Ok, maybe 4 cents instead of 2 :p
No, that's all fine, I came looking for opinions, not just the ones that I agree with.

The other key point is that you need to know what the manufacturer has decided is the cutoff voltage for their testing

All the other spec sheets for this cell look like this one, so I'm working on this being a respectable source for specs.

Discharge is down to 2.5v according to manufacturer.

I do C/D/C for all my other cells. My hypothesis for this group is based on my testing of the first 312 cells of the same source, which I did individually, cell by cell, C/D/C. And for those first 312, every single cell tested within 5% of each other in all categories, as well as within 5% of their original manufacturer specs.

So, when getting another 832 cells from the same source, I can still test for individual cell data. But if I charge them in groups of 80p, and get 254-256 amp hours recorded, and that happens for 10ea 80p groups in a row... previous experience with only this source of cells tells me that i might not need to do individual capacity tests. Especially since I'm also doing IR and self-discharge checks. I don't plan on adopting this method for my normal cell testing.
 
Yeah, I can see how that would help narrow down the issue of any bogus cells. If you're already doing C/D/C, I don't see any reason why doing batches wouldn't yield similar results, just divided by the number of cells.
The only issue is making sure your charger can handle the extra amps required to charge them.

Meaning, if your charger can only do 10A max, than doing 80p means that each cell is only getting 0.8A anyways. However, if you have only a few small chargers, and you have a beefier charger, the beefier one can off load a lot of the work required when doing in bulk.

The biggest issue here though is to make absolutely sure that your two, or more, measuring devices are almost identical in their readings. Otherwise the results would be skewed anyways.
 
Yeah, I can see how that would help narrow down the issue of any bogus cells. If you're already doing C/D/C, I don't see any reason why doing batches wouldn't yield similar results, just divided by the number of cells.
The only issue is making sure your charger can handle the extra amps required to charge them.

Meaning, if your charger can only do 10A max, than doing 80p means that each cell is only getting 0.8A anyways. However, if you have only a few small chargers, and you have a beefier charger, the beefier one can off load a lot of the work required when doing in bulk.

The biggest issue here though is to make absolutely sure that your two, or more, measuring devices are almost identical in their readings. Otherwise the results would be skewed anyways.
Very correct, I'm in agreement. I actually only use 1 charger in this case, 20 amp capable (max 250ma pet cell in this case, which is just fine with me). And I won't switch chargers mid process. Consistent methods for consistent results.




if your charger can only do 10A max, than doing 80p means that each cell is only getting 0.8A anyways.
.125A each actually
 
Interesting charging method. I started a thread a while back about charging multiple cells in parallel -- but it focused on energy flow between the cells (rather than between each cell and the bus bar). My issue / assumption was that all of the cells were NOT at the same voltage prior to charging -- fairly close to one another, but not at all identical. My position was that as long as the power supply to the bus bar could maintain a voltage higher than any individual cell on the bus, energy flow would necessarily be from the bus bar to the individual cells, rather than any energy flowing between the individual cells. As a simple example, imagine a battery at 3.0v, another battery at 3.4v, and a bus bar capable of being energized to 4.2v. If just the two batteries were connected in parallel to an (unpowered) bus, everyone agrees that energy would flow from the higher-voltage pack to the lower-voltage cell, until voltage equalized -- and everyone agrees that this could be a very bad result, depending upon the size of the cells and the amperage that might transfer very quickly. But if the bus was energized to 4.2v (with sufficient amps), the discussion was over whether all flow would be from the bus to the two parallel cells, or whether any energy would still transfer cell-to-cell across the bus.

I got a lot of interesting opinions and data. If I recall correctly, I believe that the consensus was that I was "probably correct in principle, but likely to fail in practical implementation." I realize that this is somewhat different from what you're dealing with, but your 80p charging circuit is the closest thing I've seen to a real-world build of what I was talking about.

Cheers, John
 
A few weeks later, and currently discharging the last batch of my test group today. 80 cells in parallel, all rated for 3200mah, means I would expect charge results of 256,000mah. My lowest group came back with 257,895mah, while my highest group was 262,054mah. So everything above expectation. I then rested all cells at 4.17v (this was my chosen top charge voltage), and 10 days later checked voltage for self discharge. All cells stayed at 4.15v... with the exceptions of a handful of cells that I'm classifying as self-discharging for my purpose. These I tested at 4.11v. So not even heavily self-discharging, just noticeable enough that I want to seperate them and use them in a different project. I found 8 of these in total, no more than 1 per 80p group. When I found these sel-dischargers, I put them in an Opus for an individual capacity test, and they all tested above 3050mah. So still a good cell, just for something else.

All 80p groups went back in the 80p rig for a storage discharge down to 3.6v. I recorded this result as well, and got 167,500-169,000 mah for each group. Again, a good sign that these are all consistent, healthy cells. For my last group, which ended up as just 31p, I plan on testing those with individual slot chargers. Not only because their results will be inconsistent compared to the 80p groups, but it will also act as my random sampling of individual capacity testing.
 
Hello all ! I am no exxpert, just learning a bit a day , but I find lately a video of an electrication of a small cyclomotor.

THe guy did an admirable work, trying to be scioentific at every level. For his battery he used a Litokala 600 to test capacities, and then discover that the manufacturer procedure where not respected (temperature, charge and discharge current, 100% charged = lowest level of charge = ? LItokala, for instance , provided him a 3000mA capacity, while when he did the test again respected the factory method ( datasheet) the result was 2500+- quite a difference I find, and he said it was due to the computation during charge. Voila ! hope it gives some ideas ! I also discovered
lygte-info.dk/review/batteries2012/Common18650comparator. (look on google) a very compete database of cells with comparative tools , all mesurements done by the owner since 2012 - Have a good sunday
 
Last edited:
lygte-info.dk/review/batteries2012/Common18650comparator
URL Not Found.

Just paste the full URL instead of trying to get around any URL blocking algorithms. If it's a safe site, then it'll be just fine. If it's not, we'll remove the URL.
 
funny that it appears as "flashlight information but the guy have an interest in good flashlights, may be due to his location ...

, I hope the link will work that way .It is a "blof" from a Danish guy that make no compromises, he eveb offer a test ptogram to test equipments, the one he used to tes hundred of chargers of different brands (mostly for 1860 seem to me, each one is teste really in deep . That Danish guy is amazing !

I have also access to a Drive page with manuals and softwares from a Chinese firm spmecialized in battery tools , M... craft if someone is interested please write me privately . There are not so many manuals but I think the seller made that drive file just for the products I inquiry about , mostly Balancing tools and software, but also battery srveillance tools (seem useful fro different actions as solar panels, as it is a Coulometer if gives all the informations ,you need to follow wear at charge then use times Android application ( KG-F series). I will add it to my scooter, I use it a lot

example :!
This is software to control and log data from many different devices (Like: DMM, power supplies, electronic load). It is not designed to use devices to their limit, but to run test where different brands of devices is used together and make a common logfile in CSV format from all of them. It can also be used for some data analyses and make charts. There is a versatile toolbox with the popups and the possibility to build custom readouts and controls with the "Grid panel".
It has a command line, but that is only necessary to use for more advanced stuff, for simple setup, logging and looking at data the line is not needed.
TC support scripting, it can automatic generate scripts to restore setups and includes scripts for some jobs.

Regards
 
Last edited:
https://lygte-info.dk/info/batteryIndex.html index of tested LiIon batteries the data sheet gives "computed RI" so I don't know if those are the battery maker data or something else numbers between 6mOhmsd and 300 mOhms
 
Last edited:
hello,
I have now a RC3563 "battery tester" . that (not expensive , I paid about 37$ for the tool and 2 sets of probles , pliers and é retractables pins type.
You can measure the RI of assembled or unique cells and even a whole battery .
I noticed that most of decent but old 18650 cells are between 15 30 milli Ohms.
I will check and compare capacity and RI particularily with cells that have 25-30 mOhms . Is there a sensible capacity difference in that case ? Do some made such testing ? I noticed that when testing // ensemble, RI should be low always , when there is weak cells or cell the RI rise a lot as 300 mOhms. On good batteries 5 3.6V cells monted in // have easily 5mOhms . It does not relate to the computation obtained with Ohlow which would give way lower resistance . Any links or readings , experiences on those aspects would be appreciated. Thank you
 
Last edited:
Is there a sensible capacity difference in that case ?
sure, IR and capacity go together, when IR rises you can be sure the cell is losing capacity. Older cells/more used cells will have a higher IR respect to a newer cell of the same producer/model, and a lower capacity. What's important for you is to set your own standards for your projects. My cells, all used 18650 ICR cells, are acceptable for powerwall applications when in 40-70mOhm range. Cells with higher IR are good for less-critical applications, like flashlight, power backup packs, etc.

Do some made such testing ?
Oh, you'll find quite a few threads with tests. First to come in my mind are Wolf's tests, on capacity/performance/IR...

Here there's a link to Wolf's IR cheatsheet: https://secondlifestorage.com/index...cgr18650-cell-specifications.1855/#post-79108
 
Back
Top