Fixing a (new) dud 36V10Ah battery pack.

Buk___

10 kW
Joined
Jul 28, 2017
Messages
750
I bought a 36V 10Ah rack-mount battery pack from (via) Amazon. https://www.amazon.co.uk/gp/product/B01NGUGEED

When it arrived properly packed and sealed, it turns out to be a dud.That is, there is 0 (zero) volts available at the output wires.

I've done the return thing, got a full refund, and due to problems getting a courier to carry the "faulty" battery pack back to Amazon, they have told me to take it to a recycler. I essentially have £200+ pounds of brand new samsung 18650 cells + plus case charger etc. sitting there unusable, but it offends me to scrap it. What do I do?

As the refund has been issued, I've looked inside (broken the seal voiding warranty) and I cannot see any obvious signs of broken wires or such from either end, which is what I expected to find as my understanding is that even if the cells themselves were old or worn out, I would expect to see some voltage at the output if everything was connected properly.

Internally, there are connectors at both ends that allow the end caps of the aluminium housing to be removed.

At one end this connector connects to a 4 prong slide-fit spade connector that mates to the fixed-to-the-bike part of the case which (I assume) contains the controller. There is no voltage at this connector.

At the other end the connector goes to an integral tail light and a press-to-check led power-remaining-gauge. I measure 8 volts here?

The cells themselves all look brand new and undamaged from what I can see, but the view is extremely limited as the cell mass is wrapped in (probably shrink wrapped) polythene and then wedged and/or glued inside the hollow extruded aluminium tube. I haven't attempted to remove it yet.

Advice?

Cheers, Buk.
 
Hillhater said:
See if you canfind any contacts you can get a MMeter probes on to to look for signes of life.
Idealy see if you can measure voltage on a single cell somewhere.

This is what I see looking in both ends of the case:P1100154(800x533).JPG

and the other end below (weird photo handling here?)

As you can see, there is nowhere accessible to stick a meter (other than the connectors, one gives 0V and the other 8V), without I remove the cell mass from the tube.

And looking at the glue and rubber(ish) wedges, that is likely to be a pain. I'll probably have to bite that bullet, but if I do, what then?

What can I realistically hope to do then? (I'm squeamish about completely disassembling the entire pack.)

I guess I'm looking for some guide as to how to proceed and maybe a little reassurance (non-binding :) that I might end up with a working battery pack.

Thanks, Buk.
 
Hillhater said:
See if you canfind any contacts you can get a MMeter probes on to to look for signes of life.
Idealy see if you can measure voltage on a single cell somewhere.

Okay. I tried pushing the cell pack and amazingly it moved, and with a little wriggling, I managed to remove it.

But it doesn't help much as the entire pack is encased in shrink wrap, so you cannot actually see much more.

One interesting thing is that I've disturbed something (which is itself a little disturbing) as I am now measuring 8 volts on both connectors?

P1100156(800x533).JPG
P1100155(800x533).JPG
 
I bit the bullet and split open the shrink wrap I've attached a few shots (larger available if they help anyone help me).

Is my only option to break the packaging down completely? I cannot see another way to test the individual cells, nor can I see anything obvious that would account for there only being 8 volts at the output. All the cells look intact, no signs of swelling or discolouration. I cannot see any loose wires.

P1100161.jpgP1100162.jpgP1100163.jpg
P1100164.jpgP1100165.jpgP1100166.jpg

Any thoughts on how I might proceed?
 
Disconnect the white connector with the small white sense wires, cut a way the tape if in the way.
You will then be able to check the 10 cell groups voltage with a DVM, start at the one end (gnd wire) and then move your probe to the next pin/s until you get 10 voltage readings.
 
hemo said:
Disconnect the white connector with the small white sense wires, cut a way the tape if in the way.
You will then be able to check the 10 cell groups voltage with a DVM, start at the one end (gnd wire) and then move your probe to the next pin/s until you get 10 voltage readings.

Damn. That was easier said than done. Over an hour with a razor blade, pin and magnifying glass to pick all the silicone rubber caulking from the connector.

Anyway, there are eleven wires: red, 9 whites, black. Within the limitations of my whole volts only meter, this shows the voltages between the red and each white, the black and each white, and the 9v between red & black:
Code:
        Red       9v
a      0v         8v
b      1v         7v
c      2v         6v
d      3v         5v
e      4v         4v
f      5v         3v
g      6v         2v
h      7v         1v
i      8v         0v
       9v        Black

Which doesn't tell me anything. Can anyone else draw any conclusions, suggest a next step?
 
Hillhater said:
You dont show the side view (cell ends) ..if you have no luck with the multipin connector, keep removing tape until you can see the cell ends to use the MMeter on them.

This is that view:P1100167(800x533).JPG

And after carefully parting the insulation:P1100168(800x533).JPG

The sense wire showing is the one labelled "e" in the post above.
 
If I'm interpreting your readings correctly, you have <\=1v per cell with a total pack voltage of 9v across 10cells?
Can you take a reading directly across a cell, any cell to confirm?
If this is the case I'd say the battery has self discharged somehow til this point, faulty bms? Really crappy cells?
Hard to know without much more testing. The safe thing to do is chuck it, cause it could be a fire risk now or in the future.
But some have had good results with a very slow recharge to ~3v per cell then normal charge monitoring heat and cell group voltages etc. then you would need to test self discharge and bms function. Then it would still be one you keep an eye on. (Charge it under your pillow so you know straight away if it catches fire ha!) but a slow discharge seems to harm less than a rapid hot discharge to zero volts, so you might be in luck.
Do the cells have markings like brand etc?
 
kdog said:
If I'm interpreting your readings correctly, you have <\=1v per cell with a total pack voltage of 9v across 10cells?

Not exactly. The sense wires I'm testing go to different points on the pack of 40 cells and presumably sense the voltages at different points in the pack so that it can decide the appropriate charging strategy. I'm still trying to figure out how the individual cells are arranged electrically and connected. And where in the schema the sense wires are connected. Ie. What they are actually sensing.

I was hoping I might get away with not breaking the packaging down too much as I will then have to reassemble it some how.

kdog said:
Can you take a reading directly across a cell, any cell to confirm?

That's really difficult the way they are physically packaged and electrically connected.

kdog said:
Do the cells have markings like brand etc?

Yes. The 40 cells are Samsung ICR18650-26H (pink). Per this pdf.

Everything about them say's they are the genuine article in perfect (physical) condition. (Of course, you may be right about them having self-discharged.)

Having exposed one end a little more, I can see that there are 8 cells connected 4x4. That is, a group of 4 facing one way, next to another group of 4 facing the other way, and all 8 ends (4 +ve & 4 -ve) are connected by a metal plate spot welded to the terminals.
P1199169(crop).jpg

If that schema continues throughout the pack, it would mean 10 groups of 4 cells in parallel, connected serially; which gives 36.5V @ 10.4Ah which matches the spec. And that would suggest that the sense wires would be connected at the serial connection between each group of 4 parallel.

And if that's the case, the fact that I'm getting ~1V between each stage rather than 3.65v, certainly seems to indicate a state of very deep discharge.

Is there any other explanation? Is recovery possible? Is the only solution to completely disassemble, buy a single battery charger and try recovering them individually?

Has anyone here broken a pack like this down? How easy is it to break the spot welds? How do you go about reassembling them?

(I know practically nothing about Li-on battery charging!)
 
Its likely that the BMS has failed and drained the pack right down.
If it were mine , i would remove all the wrapping and the BMS to leave only the cells.
No need to remove the cells or connections from the cells.
I would then slowly charge ( 0.1 amp) each individual group of 4 paralel cells to see if any will hold a charge when they get to 3.8 volts.
You will need a adjustable rate single cell lithium charger, and a lot of patience.!
 
Hillhater said:
I would then slowly charge ( 0.1 amp) each individual group of 4 paralel cells to see if any will hold a charge when they get to 3.8 volts.

If all the serial connections are made by spot welded plates, can I charge a parallel group without isolating it from the groups either side?

Hillhater said:
You will need a adjustable rate single cell lithium charger, and a lot of patience.!


Something like MagiDeal 5A Adjustable Step Down Lithium ion Battery Charger Module LED Voltmeter?

If so, would the output from the supplied battery charger be good as input to that module?


And assuming I managed to recover the pack, what would you do about replacing the BMS?
 
I recovered a pack showing 1v by just disconecting the bms and plugging in the charger,I did this outside in the barbecue to avoid any fire risk,after an all night charge much to my surprise the pack was recovered and all cell groups equal.This was done about 3 months ago and the pack is still good.Your bms may need replaceing though.
 
Each P group of 4 cells you can count as a single cell, so 1v across the pgroup is what I meant. You could bulk charge the whole lot without disassembly. You'll need a low amp constant current power supply of sorts. Then just monitor each cell grp voltage as you go. But it might be easier ist to get a 5v 0:5a wall wart and do the groups individually. You could charge through the sense wires at that low amp rate which would be super easy, but you'd just have to make sure you get the wiring right.
 
kdog said:
You could bulk charge the whole lot without disassembly. You'll need a low amp constant current power supply of sorts.Then just monitor each cell grp voltage as you go.

Can you elaborate, link to, (explain in simple terms) "You'll need a low amp constant current power supply of sorts"?


kdog said:
But it might be easier ist to get a 5v 0:5a wall wart and do the groups individually.

By "wall wart" you mean something like a phone charger?

I've got a old charger from a feature phone that is rated at 4.75V @ 0.55A, would that do the job for trickling individual 4P groups?

If so, how long should I leave it connected before checking the result?

Is this something I can do in my home office, or do I need to find an abandoned quarry with a power supply?

kdog said:
You could charge through the sense wires at that low amp rate which would be super easy, but you'd just have to make sure you get the wiring right.

I've worked my way through the wires and connection plates and I've come up with a graphical representation of the physical and electrical layout:LayoutDrawing.jpg

I can work out which pair of sense leads to connect to for each 4p group okay.

It is +ve to +ve & -ve to -ve to charge?
 
andy1956 said:
I recovered a pack showing 1v by just disconecting the bms

Did you physically disconnect/remove the BMS, or just hotwire past it?
 
Buk___ said:
andy1956 said:
I recovered a pack showing 1v by just disconecting the bms

Did you physically disconnect/remove the BMS, or just hotwire past it?
just pulled the plug from the bms board then connected the charger to + and - on the battery.
 
Your pack is 10 groups of 4 cells paralleled together, each of these groups is connected in series to give the final voltage of 42 volts fully charged. the sense wires (the thin ones) go to each P group so that the BMS can determine the voltage of each group when charging or discharging or balancing. if a group goes above 4.2 or below ~2.5 it turns the ouput/input off to prevent cell damage. So between neg and sense wire 1, you are measuring the first group, between sense wire 1 and 2, the second group and so on till sense wire 9&10.
if you apply a power source to these wires you can charge the individual groups one by one. an ideal power supply would be say 3-4volts at aprox .5amps. pretty close to your phone charger (which you can use), HOWEVER its output is 4.75v which is a bit too high to set and forget. you can charge each group with this phone charger but you will have to watch the voltages carefully so that the V doesn't go above 4.2 (which will take quite a long time ~20hrs)
I would pop it on each group via the sense wires, bring each group up to say exactly 3.60volts, (im guessing aprox 1hr, get a new DMM that does 1/100ths V). Do this by inserting a small wire into the correct pins on the BMS plug and use alligator clips to connect to phone charger. Don't create a short doing this, use some tape etc to insulate carefully.Be absolutely sure you have the polarity correct!
when you have finished this process, reconnect the sense wire plug into the BMS and put it on the charger that came with the pack and charge in the normal way (monitoring heat and cell V's). the BMS may/may not be toast so youll have to make sure no cell group goes over 4.2v manually
once all that is done, get back to us regarding further testing/monitoring :wink:
there are other ways to do this and im not saying this is the best... just what I would do.
 
this is fine to do at home, just be sure on your connections, polarity and monitor temp and voltages (never above 4.2 never below 2.5 (well after this episode!)) id check it like a newborn say every 10 mins (depending on your parenting style). Aim for no more than 0.05V between the cells at the end and aim for 3.6 volts.
 
kdog said:
... just what I would do.

And, thanks to your extremely patient, clear and concise instructions, the one I will be trying first as disconnecting the bms completely requires unsoldering the main power leads and they've been silicone caulked. I'd rather leave it in situ with just the sense lead connector unplugged so I can (assuming I manage to get some voltage back in the cells) at least try the bms before I conclude it is scrap.

I found a couple of my wife's tailoring pins are a good fit into the sense lead conenctor without forcing and they have large round plastic heads which serves to keep them a reasonable distance apart without shorting, and I get a good connection with the alligator clips on the new multimeter I just bought.

Probably of no interest, but I left the sense lead connector disconnected overnight (maybe irrelevant) but today, instead of measuring just under 9 volts across black/red sense wires it was upto 11.35v when I started posting this reply (which pretty accurately reflects the sum of the P group voltages) and just now(10 mins later) is upto 11.56v.

The latest(more accurate) readings are (a->j neg->pos) a:1.28v b:1.04v c:1.16v d:1.11v e:1.24v f:1.17v g:1.23v h:0.93v (crept up from 0.79v over 5 minutes as I watched) i:1.23v j:0.96v

I'm not sure if this is typical for these cells, or maybe a side effect of the (very) small current that flows when the voltmeter is connected; or something else. And I'm not sure if it has any significance.

With respect to using the 4.75@0.55A charger: Am I interpreting you correctly in saying that its okay (or I'll get away with) using greater than 4.2V so long as I don't allow the cells don't accumulate to greater than 4.2V?

Also, I've found I have another old charger that outputs 3V@1.0A. Would this be a safer option to start -- requiring less regular monitoring with for getting close to 3.6V (lower voltage differential) or is the 1.0A just too high?

And, can I connect both the voltmeter and the power supply at the same time for monitoring purposes, or would the meter reflect the nominal output of the supply and not the battery? (Ie. Do I need to disconnect the charger supply each time I check a P-group voltage? If so, is it enough to unplug the charger to take the reading, I must I disconnect the transformer brick from the p-group to take the voltage reading? (It'd be a lot simpler to connect both and just monitor the voltmeter every 10 mins or so, if that will give me an accurate reading.)

Sorry for needing to be spoon fed, and thank you greatly for your patience. Li-ion batteries are a new world to me, and this problem means a baptism of anxity and I don't want the fire to go with it :)

Buk
 
I guess you could use the 3v power supply but I'd go with the 4.75v cause it'll be a bit slower, and the 3v one will stop charging at 3v, which is fine but the power is a bit high (1a. Yes you can connect both v meter and power supply but it'll be in accurate (high). Measure whilst off the charger for the final voltage, but for progress readings sure- leave it on. Record as 999 suggested.
 
kdog said:
I guess you could use the 3v power supply but I'd go with the 4.75v cause it'll be a bit slower, and the 3v one will stop charging at 3v, which is fine but the power is a bit high (1a. Yes you can connect both v meter and power supply but it'll be in accurate (high). Measure whilst off the charger for the final voltage, but for progress readings sure- leave it on. Record as 999 suggested.


Okay. I now have a reading of 31.9V with most of the cells ~3.20 after a first round of ~15 mins per p-group. Then I did the 3 lowest for ~80 mins getting them 3.50 which falls back to 3.47 once the supply is off.

I ran out of time and patience last night. I'll bring them all up to 3.6 over the course of today. Thanks kdog and all.

Voltages on cells when I left off last night, and again this morning after 6hrs standing.
Code:
1: 3.15    3.12
2: 3.12    3.10
3: 3.06    3.04
4: 3.46    3.43
5: 3.47    3.43
6: 3.47    3.43
7: 3.07    3.05
8: 3.11    3.09
9: 3.06    3.05
10:3.09    3.08

Seems to take about~80 mins per to go from ~3.0 to ~3.5.

Update: I think I'm going to have to settle for 3.52v instead 3.6. I don't think my wall wart is up to more.
It took a little over 2 hours to from 3.12 to 3.52, but 2 hours more made no further difference.
 
Back
Top