1
\$\begingroup\$

I'd like to build a power bank with hot-swappable 18650 cells. A "retirement home for salvaged cells" device.
It would be expected to keep providing power on cell failure and replacement (similar charge level first assured)

The cells have different capacities and resistances, and possibly slightly different chemistries.

The answer here explains much re: mismatched cells in parallel.

I wonder if the combination of multiple protection circuits and PTCs in a 4x4 array doesn't perhaps lead to some dangerous feedback loops, deadlocks, etc - e.g. reaching a level where components cause each other to repeatedly swith on and off, or require special circumstances to reset.

Is that a noteworthy danger?

\$\endgroup\$
4
  • 1
    \$\begingroup\$ Personally, I would not use salvaged cells for anything. There is no way to know their history or how much life they have left. You can put a bunch of work into your project and get very little out of it. Also, cells which are used after their series resistance goes up or their capacity has diminished are more likely to vent or expand or even catch on fire. I am not sure what you mean when you say "4x4." If you mean 4 series and 4 parallel, then I DEFINITELY don't recommend it. You never want to put cells with different capacities in series. \$\endgroup\$
    – mkeith
    Oct 2 '16 at 23:41
  • \$\begingroup\$ "...combination of multiple protection..." Depends on the protection circuits. To answer your question, we would need to know the schematics of the protection circuits, or their spec sheets. \$\endgroup\$ Oct 2 '16 at 23:43
  • \$\begingroup\$ @mkeith Hadn't thought to write it, but yes, the series-connected cells would need to be ensured similar capacity. \$\endgroup\$
    – kaay
    Oct 3 '16 at 0:20
  • \$\begingroup\$ @NickAlexeev I would love to know such things too, but here all I have is "$0.7/pc thin board that gets mounted on protected batteries from China plus potentially more circuits itegrated into some of the cells". The ones I've tested seem to offer overharge/overdischarge/reverse charge protection. But I'm asking more about the principle. \$\endgroup\$
    – kaay
    Oct 3 '16 at 0:30
3
\$\begingroup\$

From the end:

I DO think that recycling cells from old packs is potentially viable. Often one or a few cells fail and 'take the whole pack down' and people do reuse cells on an amateur basis and are planning to do so commercially at large scale. But for good results a reasonable degree of cell matching or of individual cell management seems almost mandatory.

_________________________________

A 4S4P hot swappable 18650 battery bank is doable with sufficient* design.
BUT I'd expect the chances of it working safely and reliably with multiple swaps over an extended period, using just original individual cell protection, would be minimal. A system with tailored individual cell protection would have greater chance.
Best of all would be a system that worked on nested single cell, per individual string and multiple string protection. Whether that would be economic or viable is TBD. I know that there are people attempting to do this commercially - whether they can succeed is also TBD.

*The value of "sufficient" depends on what the end product is to be used for and in what environment and how well you are willing or able to characterise cells before use. Original individual cell protectors ("ICP") aim at fairly coarse "don't let it die spectacularly" protection. So while "vent with flame is a failure, "be treated roughly and lose capacity after too few cycles" is beyond its purview.

Individual cells should NEVER be charged to above 4.2V/cell. How accurate the individual 4.2V of a given ICP is is unknown and may vary with quality / brand/temperature/POSSIBLY aging and Murphy. Charging a 4S string to 16.8V MAY allow no individual cell to charge beyond 4.2V and MAY allow all cells to charge to within say 0.25 to 0.5V of 4.2V, and may not. If cells are mixed indiscriminately then worst case may conceivably see 3 x conservative voltage limiters and one no longer really working one mixed in a 4S string so you get say 3 x 4.1V and 1 x 4.5V. You MAY survive that. You may not.

"Real people" seem to think you need to balance 4S strings in normal use, so there seems little chance that semi randomly selected strings of 4S unbalanced would work well. Best case the individual string voltage limiters would work, but if termination is based on I chg at CV tailing down to some % of CC max then overall string capacity could be significantly reduced by one low capacity cell. This then tends to "pump" the bad cell down over multiple cycles as it is as say 90% when others are at 100% on charge, so that on the following discharge it reaches Vminimum before the others and as Vstring is not yet at Vininun_string the low cell gets driven into over-discharge, has its capacity further reduced thereby and then trails the others on the next charging cycle.

I could keep on making up various realistic but also semi-contrived "could happen" scenarios but it should be enough to say that the variability in normal use with same brand and model cells in similar condition is enough that active balancing is deemed necessary for commercial use SO in this case with mismatch of just about everything it seems an utter necessity.

I DO think that recycling cells from old packs is potentially viable. Often one or a few cells fail and 'take the whole pack down' and people do reuse cells on an amateur basis and are planning to do so commercially at large scale. But for good results a reasonable degree of cell matching or of individual cell management seems almost mandatory.

\$\endgroup\$
7
  • \$\begingroup\$ Thank you. I was thinking specifically of just the problems with interaction of the protection components - something perhaps like with connecting regulated power supplies in parallel - But with the strong "not worth the trouble" vibe, I'll look into those commercial attempts instead. \$\endgroup\$
    – kaay
    Oct 3 '16 at 9:02
  • \$\begingroup\$ My sense is the way to make it viable is a more than usually intelligent charger/BMS/protection framework that re-evaluates the performance of every individual cell on every charge cycle, to actively participate in maintaining the pack ("cell 3 in the second chain is limiting the capacity, change it next" \$\endgroup\$ Oct 3 '16 at 10:28
  • \$\begingroup\$ @kaay I didn't mean to put you off! Really :-). Just give you a feel for what may be in store. Just adding balancing connections per cell and looking at how people do balancing systems would go a long way towards making it less likely to immolate itself. | The commercial systems I mentioned are not yet on the retail/domestic market. There may be ones that are that I am unaware of but the ones I had in mind are doing development work in Australia. I've talked to one of the two main developers.... \$\endgroup\$
    – Russell McMahon
    Oct 3 '16 at 12:22
  • 1
    \$\begingroup\$ ... I won't comment directly on things I was told BUT they have won a number of publicised prizes for innovation and there is material on web about it so the general concept and some detail is not secret. Based solely on what I was able to find mentioned publicly they are using cells from dead packs and establishing means of running them in modules which deliver constant output voltage regardless of individual cell performance. I have no details of how they do what they do but reading their claims it sounds by no means impossible but annoyingly complex. ... \$\endgroup\$
    – Russell McMahon
    Oct 3 '16 at 12:24
  • \$\begingroup\$ ... ie if you have a mix of capacities, variations in remaining capacity, differing impedances etc then balancing and maintaining Vout and a minimum Iout performance level sounds demanding - again, not in concept which is "just a matter of doing it" but in hardware required. \$\endgroup\$
    – Russell McMahon
    Oct 3 '16 at 12:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.