one group keeps drawing down

[cloudy]

I have a home built 16s 9p ebike battery made of cells from Queenbattery 18650 as reviewed by Thunderheart here:

Queen Battery QB18650 2500mAh capacity test + comparison with QB18650 2600mAh

18650, 2500mAh, 2600mAh Queen Battery, QB18650, 3.6V, 4.20V, 3.7V, capacity test, comparison, datasheet pdf, DIY battery, DIY ebike, Li-ion, review,

www.thunderheartreviews.com

I use a JBD 10-17s smart BMS with three wires, both charge and discharge happens through the same lead on this BMS:
JBD

I recently went away for three months, I decided to disable bms balance while away in an effort to reduce quiescent BMS usage. I returned to find group 16 had discharged down to 0.7V over three months. The other groups were healthy around 3.6 V.

BMS temperature showed fairly close to ambient (11C), but when I charged it and re-enabled balancing BMS temperature has risen to 26.5C, so I know the BMS is doing something.

What I would like some help with:

1. Why did group 16 discharge, simply a bad cell in this group, could it be a problem with the BMS?
2. The Bluetooth BMS remained on while away (no way to turn off), does anyone know how these get powered? Does bms power come from one group, or from the whole battery in a balanced way?
3. How to troubleshoot? Can I simply cut one balance wire between G15/G16, and then cut the series connection for G16? Or what is the correct order to disconnect this group and prevent damage to the BMS? I have fried one of these smart BMS already while changing connections so would like to avoid the magic smoke please.
4. My plan was to simply replace all 9 cells in this group with new ones and spot weld into the existing pack. It is the last group that is affected so will need to solder the main Positive terminal onto this also. Has anyone had the same problem, where just the final group discharges? It could be that the soldering of this group is what killed a cell in the first place, but not sure. I was fairly careful but it was my first large build.

Thanks, I'll add some screenshots of BMS shortly.

 

[cloudy]

After charging and balancing a few days:

 

[Chalo]

You chose a battery made out of rinky dink cells, but without the QR quality control required to wrangle a squabble of vape pen foolishness. The results speak for themselves.

Next time if you want 22+ Ah, you use cells of 22+ Ah. It's not a hard problem.

You don't use LEGO to build a house. Why would you use LEGO cells to build a full scale pack?

 

[cloudy]

This is the positive group 16 in question. I soldered the positive leads onto copper strip before spot welding these into place. This bike uses a 45A controller.

 

[cloudy]

You chose a battery made out of rinky dink cells, but without the QR quality control required to wrangle a squabble of vape pen foolishness. The results speak for themselves.

Next time if you want 22+ Ah, you use cells of 22+ Ah. It's not a hard problem.

You don't use LEGO to build a house. Why would you use LEGO cells to build a full scale pack?

Hey Chalo, so you think it's not worth working on at all, build a whole new pack with better cells instead?

 

[gromike]

Probably building another battery, with the knowledge you have gained, would be a good thing, but your question is what drained that one group.
Off hand, with what you have said about your bms having an always on bluetooth. Well, I would hazard to guess that that group powers the bluetooth.

 

[cloudy]

Probably building another battery, with the knowledge you have gained, would be a good thing, but your question is what drained that one group.
Off hand, with what you have said about your bms having an always on bluetooth. Well, I would hazard to guess that that group powers the bluetooth.

Thanks, I can confirm that my question is about what discharged that group.

If this group 16 powered the bluetooth BMS, wouldn't it have stopped working when group voltage dropped to 0.7 V? I was still able to connect and the BMS was still running fine.

It is for this reason I suspect there is a bad cell. Do you think the BMS could still power itself off 0.7 V?

 

[neptronix]

I take it you didn't test & then match the cells & you've got a dud cell or two. You don't know until you disassemble the pack and then measure & find out which one is responsible.

 

[cloudy]

I take it you didn't test & then match the cells & you've got a dud cell or two. You don't know until you disassemble the pack and then measure & find out which one is responsible.

Thanks Neptronix, beyond testing voltage after a one month storage period I did not check anything else prior to building . I'm assuming you mean testing cell Internal Resistance to compare cells?

I don't have a four wire device, can this be done with two multimeters and a resistor load + cell holder, to measure current and voltage with four wires simultaneously?

Or, do I need to buy a specific four wire device to measure IR accurately?

 

[neptronix]

Internal resistance and capacity are what to check.

The thread i wrote about RC Lipo might be helpful, it can be adapted to any kind of battery, you just need cylindrical cell holders hooked up into the balance leads, then with the right charger you could graph 10 cells at once. Not bad!
RC Lipo pack inspection, series/parallel charging, and lifespan/safety tips

RC Lipo chargers are also relatively cheap. iCharger brand is what i prefer because they are pretty accurate considering the cost vs a professional device.

 

[Gruesome]

But to answer the other question, the welding does look pretty bad to me; this looks like nickel over copper, and it seems there are holes burned through both strips, so how do you know they are not burned right into the cell?

 

[cloudy]

But to answer the other question, the welding does look pretty bad to me; this looks like nickel over copper, and it seems there are holes burned through both strips, so how do you know they are not burned right into the cell?

I tested on old cells using copper nickel sandwich. Turning up weld power eventually did puncture cells, and I could smell it when it happened. Sounds like I should have used less power? I found it took quite a bit to weld the copper sandwich.

 

[cloudy]

Does anyone know the process to disconnect this JBD BMS from Group 16 without damage to BMS?

 

[rafalg]

Unplug the whole connector from the BMS
Anyway, disassembling the battery will be a mess, be careful not to damage any cells or short-circuit the sections. Better do it outside with some plan where to toss the battery if it catches fire.

 

[docw009]

To take cells out of a holder, I remove the nickel on both sides. Then I cut the plastic retaining tabs in the holder on the negative side. The cells can then be pushed out with a wooden tool. Of course, your results may vary, At 3.5V/cell, that's like 40% charge so you do need to be careful. Instead of soldering your wires directly to the replacement cells, you can run nine nickel straps off the ends, and solder the wires to the ends.

Cells shouldn't be re-used after they've sat at low voltage like that. The concern is that they will develop into leakers and heaters when recharged. So you're right in wanting to replace the group.

I think the probability is high that one of the nine cells in the group had already turned into a leaker. When you remove them, set them aside somewhere safe, Bet that one of them will continue its self discharging ways and wind up at zero volts later. This is what I've seen in two cases when I removed a bad group.

Unfortunately, if one cell in a battery went bad, odds are others might too. That has also been my experience and I no longer try to replace cells. You do have a smart BMS though, and it's saying that the other 15 groups are holding voltage, so it's probably worth a try. Can't speak to the quality of your cells. They are obviously not top quality from Panasonic, LG, etc, or they would be marketed as such. I have not seen problems with quality cells. It's always the no name cells that went bad.

As for big batteries with 18650's, builders have been dong that for years, but now that we have 21700's, you could do it with less cells. I'm dubious when people try to pull +20A out of a 21700 and expect it to last.

 

[lnanek]

The manual of the active balancer I use does explicitly call out that it uses one of the p-groups (4th in this case) to power the logic chips:

https://a.co/d/9eThktF

So it wouldn't surprise me if a BMS worked similarly. It's cheaper and easier to build if these devices never have to convert pack voltage down to the 3-5V the chips run on.

Some might be able to boost voltage up without being able to buck it down too, since, again, it's cheaper to only be able to go one direction.

The low voltage cut off I use on my kid's Powerwheels takes another approach where it has a separate 12V input that's different from the pack input and protected output:

https://www.amazon.com/dp/B07RSXQHXS

So there are other approaches out there, at least.

 

[docw009]

I pulled up an arbitrary JBD BMS spec and it said 200 ua in sleep mode. For an hour, that would be .0002AH, and for 3 months it's less than .5AH. Say one group does power the BMS. Your battery at 3.6V/group had around 8AH in each groups when you left, so it's probably not a case of the BMS draining the battery,

 

[A-DamW]

Off hand, with what you have said about your bms having an always on bluetooth. Well, I would hazard to guess that that group powers the bluetooth.

I own this exact model of BMS that has been attached to a 13s4p storage test pack(no charging)for 5 months of testing so far.
After initial balancing, all cells are currently @3.9v, and no perceptible imbalance.
The whole BMS/Bluetooth goes to sleep when there is no active BT connection after ~30sec.
I connect to this test setup about once a month to check,
using either the JBD app, Overkillsolar app, or this esp32 BT display I am developing;
Post in thread 'Making bms's VESC compatible' https://endless-sphere.com/sphere/threads/making-bmss-vesc-compatible.124742/post-1815273

 

[cloudy]

I own this exact model of BMS that has been attached to a 13s4p storage test pack(no charging)for 5 months of testing so far.
After initial balancing, all cells are currently @3.9v, and no perceptible imbalance.
The whole BMS/Bluetooth goes to sleep when there is no active BT connection after ~30sec.
I connect to this test setup about once a month to check,
using either the JBD app, Overkillsolar app, or this esp32 BT display I am developing;
Post in thread 'Making bms's VESC compatible' https://endless-sphere.com/sphere/threads/making-bmss-vesc-compatible.124742/post-1815273

Thanks for your reply, I had a feeling that someone must have done some testing on this particular JBD BMS (BMS- SP17S005 80A). I only tested pack/group over and under-voltage protection, and temperature protection before installing it. I did not look closely at how it did balancing.

What you have said indicates that it is most likely a cell in group 16 of my pack that is bad, not the BMS drawing down one group. This helps narrow things down for me,

Have you had a look at how this BMS does balancing? Does it simply draw down the highest groups toward the lowest group voltage and burn this energy resistively, or does it actually transfer (charge up) the lowest group also? I have noticed the BMS warms up when balancing is active.

When you say the bluetooth goes to sleep, does the small surface mount blue LED on the BMS remain on while it's asleep? Mine seemed to remain on all the time, before I wrapped the battery and put it into the bike. I had some concerns about this during the build but opted not to install any shut down switch for the BMS.

How would you go about disconnecting this BMS, simply unplug the whole balance connector as suggested by rafalg above?

I had to look up VESC, looks a good way to protect the motor and battery.

 

[A-DamW]

Have you had a look at how this BMS does balancing? Does it simply draw down the highest groups toward the lowest group voltage and burn this energy resistively, or does it actually transfer (charge up) the lowest group also? I have noticed the BMS warms up when balancing is active.

Resistive balancing.

When you say the bluetooth goes to sleep, does the small surface mount blue LED on the BMS remain on while it's asleep? Mine seemed to remain on all the time, before I wrapped the battery and put it into the bike. I had some concerns about this during the build but opted not to install any shut down switch for the BMS.

I have only seen the blue led off when BT disconnected and balance not active.

How would you go about disconnecting this BMS, simply unplug the whole balance connector as suggested by rafalg above?

I would disconnect whole balance connector.

 

[cloudy]

Resistive balancing.

Must be hard work for the BMS to resistive balance the 15 good groups down to meet the one bad one. No wonder the BMS is running warm. I'll attempt to take out the bad group and replace, thanks for the warnings and suggestions.

 

[cloudy]

I was able to remove the cells from the bad group and replace with spares of the same batch. Was fairly slow going but got there without any fireworks or shorts.

With the dodgy group out I can see that one of the cells has drawn 0.1V below the others already, in only two days. So definitely a leaker in there as predicted by docw009. The replacement cells have not been used, so are in better condition than those in the pack that's seen close to 12 months of use, but they should otherwise be a good match.

The leaker cell has some signs of being punctured on the negative end, this probably happened during welding on the tabs for the positive battery terminals. I don't think it was poor cell quality that caused the issues, rather it was user error during welding. I've turned down weld pulse duration a bit, and it is still seems to create strong welds.

BMS working fine after being plugged back in. Thanks for the tips.