EDIT: Before reading the (very long) post below the first quote, keep in mind that all "balance" is, is a measure of how well-performing the pack is; how capable each cell(group) is of doing the same job at the same time.
Balance itself at any one instant in time really only tells you the specific state of charge difference between cell(groups), and doesn't tell you anything specific about those cell(groups) beyond that, other than that they are different for some reason that you should either look into if you need the pack to do everything it's capable of, or let the BMS do whatever it can to maintain balance until the pack becomes so unbalanced (or aged) that it is not usable and has to be replaced (which happens to all of them eventually).
There are some subtleties about balance and what balancing can do for performance of an unbalanced pack not covered here, but if it's truly important to keep one balanced, it's more important to start with well-matched cells used nowhere near their extremes of capacity or capability than to have a good balancing system.
If it's all well-matched, you won't even *need* a balancing system (though a monitoring BMS doesn't hurt).
Nearchos said:
I do not assemble the battery pack myself, someone in China does this for me. So he might assure me that he checks the cells for consistency but ...how would I know...
You wouldn't, unless you disassemble the pack and do the tests yourself, which is impractical at best, impossible in most cases.
Or test all your cells yourself first, then send them to a reliable builder to have them assembled into a pack (but not one you don't know for certain you can trust to use *your* cells and not just do whatever...).
But I would feel safe in assuming that the cells are *not* tested at all, or that if they are tested they are not finely-graded, only coarsely, so there can still be significant variation between cells in a pack.
a) which test is the most important to ask the battery pack builder to run to ensure cell consistency?
Charge test
Discharge test
Internal resistance test
Capacity grading and matching test?
Well, the first two test results will only really tell you stuff you can calculate based on the last two.
The spec sheet for a cell will tell you what specific characteristics are tested by the factory, under what conditions, and what range of variation is acceptable for each characteristic under those conditions.
Internal resistance can be tested in several ways, so to know what is acceptable you have to look at the spec sheet to see how they tested it, and what they expect from that test. If it doesn't say, then you would need to check with that manufacturer to see if they have a test-specification of how they test all their cells, to know which method they used. But there are at least two, DC Ri and AC Ri (which can be done at different frequencies, 1kHz is probably common).
DC Ri is what most of the end-user testers you can buy will approximately test for you, but they vary widely in accuracy and repeatability. (if you don't get the same answer from the same cell on the same test every time, the test isn't useful).
There are people on ES that do cell testing, like Pajda, etc., and sites like Lgyte.dk (?), that have info on how they do their testing if you wanted to repeat this kind of testing.
b) what voltage and ohm/impedance Variance per cell group is desirable for a Reention DP-9C at 52V 21AH containing 18650 3500mah LG lithium-ion cells arranged at 13S 6P LG?
Zero variation is desirable.
The more variation, the worse the imbalance will be at the beginning, and the faster it will become worse over time as the pack ages.
Battery pack builder says each cell is tested to ensure their
-voltage is between 3.5V and 3.7V and
-internal resistance (sic) less than 35mΩ.
Neither of those is useful, regarding making a pack of well-matched cells.
For example, if internal resistance is 17 on one cell, and 34 on another, they are both within that limit, but they will react very differently in use, and will not remain balanced, as that resistance is a "symptom" of multiple internal characteristics.
All the voltage measurement tells you is each cells' state of charge (SoC) at the instant that test is done.
It doesn't say anything at all about the cells' characteristics relative to each other, other than that they are all somewhere within the median of their SoC (and not dead, failed shorted, overcharged, etc).
0,2V voltage variance and no cell more than 35mΩ...
Is this enough for a brand new pack or should I ask for sth more ambitious? I have understood that with time variance will increase and neither the nickel strips nor the BMS will save me. A good starting point will help. So what is a good starting point?...
To have a "perfectly matched" set of cells for a pack, you'd need to actually *match* cell characteristics (resistance and capacity at minimum; see cell-testing threads for more characteristics) with as little variation as possible, as close to zero as can be tested for. So say, test for all at (random choice) 20milliohm, plus or minus 0.01milliohm--then all cells would be the *same* resistance and behave the same (regarding this characteristic) during usage and remain balanced.
Neither of the things being tested for above will do any of that.
Say you have a hundred cells all of which are less than 35milliohm, but which vary by say, up to a dozen milliohm from each other; they are not even matched, much less well-matched.
Within a milliohm might be better matched, but keep this in mind: The percent of variation is still high. Say you have those same hundred cells, and all are at 30-31 milliohm. 1 milliohm out of 30 is about 3%. So they can still be up to 3% different from each other, which is ok, but will still be different in capacity, etc., and continue to drift in balance (actual charge state of each cell/group) on every charge/discharge cycle.
It will be easier for the BMS to keep a pack with this much difference balanced, but it still means the lowest capacity group (if the Ri is indicative also of capacity in this case) determines the entire pack capacity, since it can't discharge any further than the lowest capacity group without damaging that one. If the capacity difference is also 3%, then if you have 21Ah in each group, you could (roughly) have as much as 630mAh difference between groups (there are ways to minimize this in building a pack with multiple parallel cells by matching total group capacities with each other, but over time variation will still increase).
It may not be critical to your application or pack lifespan, etc., for how well-matched the cells have to be, but if you do want them to remain balanced, to be able to use maximum capacity of each cell, etc., the better matched they are the more usable the pack will be, for a longer time.
Nearchos said:
is 5mV maximum voltage difference and no cell above 25 mΩ reasonable for LG cells?
Voltage difference beween cells at assembly makes no difference at all to cell matching.
Every cell *at* 25milliohm would make a matched-cell pack, but cells just having to be between 0 and 25 makes no difference to a matched-cell pack, and won't result in a pack staying balanced over time.
For an assembled pack, 0.01 (1mv) maximum voltage difference would be balanced for SoC between groups. Doesn't say anything about matching, though, or how balance will be at any other instant in time for the pack, unless they remain within that range during the entire charge/discharge cycle empty to full and full to empty under no load to maximum load. (That *would* indicate a well-matched set of cells in a pack).
ps. Btw I have heard two of your music tracks already, not bad at all!!
Thanks--the most recent I've spent significant time/effort on is
The Tomorrow Option
https://amberwolf.bandcamp.com/track/the-tomorrow-option
The other recent-ish ones spent a lot of time on are
Convocation of Lies
https://amberwolf.bandcamp.com/track/convocation-of-lies-feat-lauscho-remix
and
Just Give Me a Voice
https://amberwolf.bandcamp.com/track/just-give-me-a-voice