BMS Problem = Toasted Cells?

[velowatt]

I built a 12s Headway pack (10Ah cells) for nominal 36V run through a older-style "Signalab" 12- channel BMS board (same as on the V1 Ping packs). After taking the pack out for a longish ride, I got a bit over 5Ah before the voltage (total pack) sagged to 32V, which seems low. I went to charge the pack up, and noted that the BMS board was slightly warm to the touch, but I have to climb a steep hill to my house, so I figured that maximum current draw had just happened. I took a look at the board and was surprised to find evidence of Black Smoke (soot on the protective foam loosely associated with the BMS board)! A picture:

Close up of 'burned FET:View attachment Soot.JPG


Now the pack does not want to charge. Looks like an FET in the corner may have blown, the soot is collected around the B- solder terminal on the board. The FEt package looks intact. Other tracings and components appear OK, no burned spots or shorts.

View attachment Burned!.JPG

I disconnected the battery and tried to charge directly across the + and - terminals. After a bit over an hour, the charger shut off as usual. Pack voltage off he charger was only 33.4V! (Usually settles in at about 40.4 after charging and stays there). I measured the voltage across each cell and got this (starting from positive terminal of the string)

Cell 1: 3.32
Cell 2: 0.82
Cell 3: 3.32
Cell 4: 3.32
Cell 5: 0.32
Cell 6: 3.64
Cell 7: 3.23
Cell 8: 3.76
Cell 9: 3.32
Cell 10: 0.01
Cell 11: 3.75
Cell12: 0.63

I think cell 2, 5, 10, and 11 are toast (as is BMS). The others are probably salvagable. What do forum members think? I would like to replace the BMS with one that actually works (LVC for each cell would be most important to me, since I hate trashing brand- new good cells). I also would like to not spend an excessive amount of $$, since this pack is an 'extra' pack that can be used for increased range or better performance on a couple of different bikes (I don't deend on it for commuting).

The bike I was using this pack in runs an older brushed WE motor, controller is a '35A' unit that can draw MAX or 33 amps. Max power draw from this battery is just about 1000W for a short time. The ride which 'kiiled' the BMS was 20 miles. used 5.15 Ah, and about 9.5 Wh/mile (I went slow with lots of pedaling to maximize range and get a workout- usually draw 12-20 Wh/mile depending on hills and speed).

My questions to the Form are these
-Can I assume the 8 cells tthat are still above 3V are OK?
-Suggestions for a reliable/basic BMS that won't break the bank
-How low should low voltage cutoff take Headway cells? Is 2.5V per cell about right or is that too small a safety margin?

Thanks

 

[dogman dan]

That's wierd, I thought you had cell level lvc on a ping bms. But no tellling what else went, before the fet actually shut down. Perhaps the pack was really poorly balanced when you set out, and 5 ah was all there was in the dead cells. Shoulda triped though, when just one got low.

 

[velowatt]

I don't really know when the semiconductor blew- it might have been a few days ago. I did not mention it in the original post, since it seemed too long any way, but this pack has always been hinky with charging through the charger jack- First it worked, then it didn't- I thought there might be a bad solder joint or loose wire that causes intermittent failure with vibration, but can't see anything obviously wrong, so I just kept breaking the pack in. The last few cycles, if the charger did not sense the battery and start up, I just charged across the + and - terminals using wires with alligator clips attached to the charger output, until the charger throttled back and went into constant voltage mode, then disconnected. When it did charge through the BMS, I would leave it on CV for several hours in order to balance the cells. Some of the big blue resistors would get a little warm during this time, which I would expect if some cells are shunting away current. Never got too hot to touch, tho.

All the 'sense' wires are attached to BMS with good continuity. Probly the pack has been getting more and more unbalanced with successive charges (about 2/3 of the cycles have charged through the BMS) until I ran the pack for a good long ride and drew some cells done to nothing. Weird failure mode, i guess.

I think I am a lucky dude that all 12 cells are not ruined- looks like 8 of them are still okay!

 

[amberwolf]

FWIW, I have a 16s Ping v2 BMS sent to me as failed that I've worked with a bit, and it has a couple of slots that shunt *all the time*, so they'll kill the cells attached to them to 0V when it's not on the charger. It also has failed FET(s); I don't know which failed first or if any of the problems caused the others.

From what I have seen of various posts around ES, this doesn't seem like an uncommon failure mode for these, and is probably one reason for the change to the v2.5 by Ping, though I think I read of the same problem with one of those (can't remember for sure).

I'm not sure if it's the transistor that failed to short the shunt "on", or the voltage reference failed to keep the transistor biased off, as I didn't finish troubleshooting it yet. (I'd intended to work out how to use it for LiCo/LiPo instead of LiFePO4 but after the failures started working on other ideas instead.)

 

[dnmun]

you connected the charger directly to the battery? not through the BMS?

i have had the cell level LVC fail to shut off the discharge on both of the v1 and v2.5 signalabs so i am not surprised to read this failure.

 

[recumbent]

I've had a few cells destroyed also because of a faulty BMS. This occured on two seperate cell packs, with two completely different BMS's.

Sorry to see this happen to you, but many of us had this occur, and leaves a trail of discontent towards BMS's.

If I had my CellLogs (or similar device) hooked-up this would not have happened.

The $14 CellLogs are still not complete devices unfortuanely, including: no wire harness and #7-8 plugs not hooked to power and theoreticaly drain packs down unevenly. However in real life testing this never happened to me after been left on for over 3 hours, all cells were right on.

 

[velowatt]

Thank you all for comments and suggestions. I will look at the cell log device- this looks like a cool (and inexpensive!) technology. I will post back any interesting results.

 

[dnmun]

in your first picture, it shows that you have a red wire running to the C- terminal on the BMS.

where is the other end of that red wire connected?

did you measure the voltages of the individual cells before you discharged it? had you balanced the pack already before you ran the voltage down to 32V?

 

[velowatt]

The red wire in the picture is (+) terminal for the pack- it is running 'past' the BMS and through a foam pad that covered the BMS to protect it from bumps. The C- black wire connects to charger negative. The B- black wire connects to pack negative (it has a red stripe). P- goes to "load", i.e. controller negative.

Batteries were new from factory, I had them for about 5 months before assemblng the pack. When built, all were 3.33V except one which was around 3.1V (I did not write values down); I assumed that charging would balance the cells. Never checked until there was a problem after a couple of dozen break- in rides that used 1.5-2.5 Ah. I assumed that things were fine and cells were balanced while charging, which was clearly not the case.

I have ordered 'cell-log' devices (2) from Hobby king so that in the future I can check the vltage and performance of individual cells in my packs. I have not decided whether to replace the BMS or just charge and monitor the cells and charge up low cell(s) with a single charger to balance when needed.

I do wonder if this board is repairable. All the semiconductors and IC packages have had identifying numbers scraped off. Then again, the cost of a replacement is not very high. I don't have much confidence that a replacement BMS will work any better, so I will watch the individual cell voltages more closely in the future.

I have a 10 Ah 48V Headway pack (with BMS) that seems to work great, but I have never ran it all the way down to LVC. I will wait until I can monitor the cells with cell- log devices before trying this to avoid problems.

 

[dnmun]

the FET in the corner is the charging FET. that's why i asked about the wiring since it looked like you had it wired up so there is a dead short across that FET.

i think the FET is a irf3205, i have some here if you wanna replace it.

 

[dogman dan]

If you do get it fixed, or replaced, lose the foam wrapper. Bms needs a bit of air.