Parallel arrangement of power tool battery packs for E-bike (72V 20Ah) (Photos and diagram included)

[adrianbravom]

I have been gifted a set of x10 Power Tool 10S 144Wh Li-Ion Battery packs, generic w/ no brand (tried to find on internet without success, the same for brand uses the pack connector), being honest quality seems 'OK', i have disassembled a couple and cells indicates Samsung INR21700-40T (4Ah).

• The battery packs all come at 37.0V (50%ish), and in theory are brand new with no use, so the cells should be pretty well balanced.
• I have also performed 1C charge/discharge tests on all batteries using a coulomb-meter and data-logging Temperature, Voltage, currents, Capacity and Energy on Home Assistant. All packs seem to be within expected range.
• The battery packs come with a BMS board, i have not been able to 'reverse engineer' it to give a sense how good it may be, i haven't been able to find that much info for the FETs or IC.

I am from Chile, and always wanted to build an e-bike, but importing Li-Ion cells here is very complicated, so i decided to finally start the project after getting these. Since the battery packs are 10S (x10 36V packs, 100 cells in total), i plan to rearrange these into 20S5P (72V, 20Ah, 1440Wh).

Already purchased a QS165 Motor, Fardriver ND72450B. I know that might be a overkill for the battery but it was a good deal, and also wanted to have 'spare power' for the future when i get experience riding, i will only need to upgrade the battery. I know it will get poor range, this is not a concern for me, as this is a fun/learning project.

Been researching and educating myself about what are my options for approaching these packs, the biggest problem i see is that most alternatives are series-first configurations so a lot of wiring is needed. There is a lot of info about paralleling batteries in this forum, still, i don't have a solid conclusion of what is the best solution, sorry if this has been asked endlessly.

Honestly i find myself in a rabbit hole about the convenience of doing so, specially regarding to safety and economic feasibility. So here i am asking for your help and opinions.

The short questions:
- Do i need to remove their original BMS's? or can i just bypass their FET outputs?
- Regarding arrangement, what option should i go for?:
0.- Use BMS's on the packs
1.- x5 BMS (1 per 20S string).
2.- x1 BMS, put wiring (w/ PTC fuses?) between battery pack cells balance taps.
3.- Complete disassembly, then connect parallel first and add new BMS.

Detailed question - Alternatives i have been evaluating:

• 0 - Use BMS's on the packs:
  • Pros:
    • Cheapest option, no balancing wiring. (But i guess packs will get unbalanced eventually and this will become a problem).
  • Concerns:
    • I have no clue if the FETs and other BMS components will be able to handle 72V once each x2 packs connected in series for 20S (Specially during some BEMF from the motor or if/when a pack goes out).
    • Will not get monitoring or configuration.
    • Quality of BMS is not known.
• New BMS (or BMS's):
  • Basically soldering wires directly to B- and B+ Cell of the packs, so the current will not flow through actual BMS's FET's. This alternative splits in two configurations:
    • 1.- x5 BMS, x1 per 20S string (5 BMS)
    • I see this option a lot on this forum.
      • Pros:
        • Protect all cells individually
        • Monitor all cells
      • Concerns:
        • How the charging will work here?
        • Can i keep the original BMS on the packs and just bypass it? If not, i guess i can cut it from balance taps.
        • I am not sure if new BMS and original BMS's (which will work only as balancers?) will interfere each other and cause unexpected problems or situations.
    • 2.- x1 BMS for complete Battery
      • Pros:
        • Less expensive (?)
      • Concerns:
        • I guess i will need to put balance wires between parallel cells in order to have one BMS, so a lot of wiring, complexity and failure points. In some situation high current may flow between balance wires so i might need to fuse them, i was thinking about PTC self-resettable fuses.
        • I might use a PCB for intermediate cells balancing w/ PTC fuses board if needed, but again, adding a lot of complexity, failure points and costs.
  • Concerns:
    • May cause cascade protection/failure, not a big concern.
    • Bulk, uses more space since i have to keep BMS and power tool connectors (minor issue, as i said i don't expect maximum efficiency for the project).
• 3 - Completely disassemble packs removing BMS and cutting Nickel plates:
  • Pros:
    • I can arrange in parallel-first configuration, so most of my concerns for other alternatives are solved.
    • It might be cheapest option.
  • Concerns:
    • It scares me a little bit about screw it and cause some short or damage the cells when cutting the BMS's and nickel plates. (Maybe i'm too paranoic?)
    • I will need to spot weld new nickel plates, which i have no experience with.

General comments (Alternatives 0 to 2):

• At the least, each pack output will be fused.
• I will make a busbar for - and + if going with 5 strings solution.
• I am trying to find a way to import LiFePo4 prismatic cells without success at this moment ( If i find a way to do it, ironically i think it would be the safest, simplest, and cheapest option, still i want to find some use for these packs).

Probably i'll have additional concerns and comments, but i would appreciate your opinions first, as my post is too long already.
_________________________________

Posts & Links i have as reference:

• Need help with designing unique 72v battery
• Can Lipo balancing wires be paralleled? (diagram incl.)
• https://www.eevblog.com/forum/renew...-seriesparallel-connect-a-large-battery-bank/
• Parallel'ing balance leads???
• LiPo batteries in parallel
• Li-Ion BMS - White Paper - Parallel cells vs. strings in parallel
• Li-Ion BMS - White Paper - Notation for Series / parallel arrangement of battery cells

 

[j bjork]

The best way is probably to get rid of the included bms:es and use one smart bms that can handle the load. Take all the cells and build a pack that fits.
You can always try to use the boxes and see if you can design something that works.. It seems to be risky to use bms:es for half the series, like two 10s bms:es on a 20s pack. Seems like things can go wrong if one of them trips. Very tempting though, if you have power tools that use the same battery pack.

 

[adrianbravom]

The best way is probably to get rid of the included bms:es and use one smart bms that can handle the load. Take all the cells and build a pack that fits.
You can always try to use the boxes and see if you can design something that works.. It seems to be risky to use bms:es for half the series, like two 10s bms:es on a 20s pack. Seems like things can go wrong if one of them trips. Very tempting though, if you have power tools that use the same battery pack.

Yes, seems the most logical, however I don’t know if I like the idea of cutting nickel strips for 100 cells, seems a little risky to me.

I am liking the idea of removing pack BMS and installing 5 new jk BMS to monitor all cells performance. Seems a balanced solution. Charging is kinda a concern though.

 

[AnthonyC]

1.- x5 BMS (1 per 20S string).
2.- x1 BMS, put wiring (w/ PTC fuses?) between battery pack cells balance taps.

I'd do 1 or try 2, only with the PTCs wired in star (=parallel) to the new BMS rather than chained. Both give the opportunity to parallel the 1P strings using ideal diodes for dis/charging.

When one cell in a P group ages more the load sharing varies causing 'pumping' of current between cells during discharge, charge and recovery off-load, which ages the whole P group more quickly.

Separating the strings gives better control over individual cell currents even when the strings are unbalanced, giving the battery slightly higher peak current and it can be charged significantly faster (starting slowly so as to balance the strings).

If a cell fails abruptly (impact / dendrite / corrosion etc.) the string goes off-load or if charging it's protected by the BMS. It may even vent one cell's worth of heat but it doesn't have the rest of the group contributing at the same spot (until its resistance goes high) so there should be less risk of igniting adjacent groups.