52V battery now 20V - dead?

donn

10 kW
Joined
Aug 13, 2018
Messages
875
Location
Seattle
My LiFePO4 battery abruptly lost charge, on the road yesterday. When I got back it was 20V or so, and the charger won't light up and start charging. Dead?

The long story: I've recently had connection problems, such that everything drops - CA3, throttle etc. Sometimes comes back by itself, or I may need to work the controller start switch (KH6xx Infineon 4 clone.) I fiddled around with it last week, seemed to be working but of course it's an intermittent problem so I can't tell. Yesterday was the first ride after that. A few miles into the ride, I lost connection again, but also voltage - intermittently, it would drop below the cutoff, and then come back up to 48-50V. A couple of those, and then it went to 16-17V and stayed there, and of course was no longer available to drive the motor. But the controller was still on, at least some of the time, and CA3 was reporting 16V, which is not how I remember low voltage cutoff when I've hit it in the past. No regenerative braking, and it didn't feel like I was getting any core resistance while pedaling.

I would expect this battery to last a lot longer. Supposedly 85% after 1000 cycles, and I haven't been counting but I doubt I'm a 10th of the way there. Apparently there's something catastrophic in the wires, but it looks like discharging ca. 6 Ah in a few seconds, which seems like would have more visible consequences.
 
I figure your battery has a BMS, A working BMS will shut off a battery when one or more cells get below LVC, isolating the cells from the outside world.Whatever you see on the output is just leakage across the BMS transistors, so it probably isn't at 20V. Some BMS also shut off when cells are over voltage.

AT this point, I think it's about someone opening up the battery, getting all the cell voltages and then deciding the next step. based on that info.
 
This usually means the BMS has shut off it's output and input, so the voltage you see on there is a "ghost" voltage; any load on it would drop it to nothing.

This could be a bad cell that's gone lower than it's safety LVC, or a broken-off sense/balance wire. You'd have to open it up to check voltages. First just check the main cell-pack + and -, rather than at the BMS. This will tell you if the pack itself is really that low or not (probably isn't).

Then test all the cell group voltages. This is best done at the solder joints of the BMS for the sense/balance connector, one pair of pins at a time, and including from the last pin on each end to the main cell pack positive / negative as appropriate. Post the list of voltages here, starting from the most negative end, and we can see if there are any problematic ones.


Regarding the connection problem: which connection is causing the problem? If you're not sure, you need to find that, too; if it's internal to the battery it could be causing both problems; if it's elsewhere it's probably an additional issue.

If the problem is so intermittent that it cant' be found by the typical "wire wiggle" tests, then best is to disconnect and examine every plug, one at a time, that could cause the symptoms you see. If there are multiple symptoms, there may be multiple problems. Each connector should be checked for oxidation/corrosion or arc damage on the pins, pins that can "back out" of the shell when pushed on from the mating side or pulled on from the wire side, wires that are not properly crimped into their contacts, or broken inside their insulation at or near the back of teh connector where bending is commonly stressful, etc. Also for shell damage; distortion, crush damage, etc.


donn said:
My LiFePO4 battery abruptly lost charge, on the road yesterday. When I got back it was 20V or so, and the charger won't light up and start charging. Dead?

The long story: I've recently had connection problems, such that everything drops - CA3, throttle etc. Sometimes comes back by itself, or I may need to work the controller start switch (KH6xx Infineon 4 clone.) I fiddled around with it last week, seemed to be working but of course it's an intermittent problem so I can't tell. Yesterday was the first ride after that. A few miles into the ride, I lost connection again, but also voltage - intermittently, it would drop below the cutoff, and then come back up to 48-50V. A couple of those, and then it went to 16-17V and stayed there, and of course was no longer available to drive the motor. But the controller was still on, at least some of the time, and CA3 was reporting 16V, which is not how I remember low voltage cutoff when I've hit it in the past. No regenerative braking, and it didn't feel like I was getting any core resistance while pedaling.

I would expect this battery to last a lot longer. Supposedly 85% after 1000 cycles, and I haven't been counting but I doubt I'm a 10th of the way there. Apparently there's something catastrophic in the wires, but it looks like discharging ca. 6 Ah in a few seconds, which seems like would have more visible consequences.
 
amberwolf said:
If the problem is so intermittent that it cant' be found by the typical "wire wiggle" tests, then best is to disconnect and examine every plug, one at a time, that could cause the symptoms you see.

And those would be relatively few, I think? The problem initially, anyway, was system wide outage, so
  • "ignition"
  • power to controller

That stuff all looks good to me, but apparently there's something wrong in there. Anderson Powerpole connectors to the battery.

The problem started after I swapped in my LiCo battery, but at that point it was just disconnects, not the apparent battery outage. I don't really have any notion what would cause the LiFePO4 battery to break down, can't be just coincidence.
 
Ha, I imagined there'd be A123 cells in there, or something of that nature.
20220516_162631.jpg

The BMS has a row of wires out to the cells, and testing those I get a sequence from the 0 to ca. 48V, incrementing by ca 3V each connecting point. Starting from the pin that's towards the bottom here.

20220516_165307.jpg

So if there's a problem with any of the cells, it isn't obvious. I count 16 packs; 14 white wires, 1 black and 1 red coming out of that terminal, of which the 14 white and 1 black connect to packs. That doesn't exactly add up for me, but it's apparently as designed.
 
donn said:
The BMS has a row of wires out to the cells, and testing those I get a sequence from the 0 to ca. 48V, incrementing by ca 3V each connecting point. Starting from the pin that's towards the bottom here.

20220516_165307.jpg

So if there's a problem with any of the cells, it isn't obvious. I count 16 packs; 14 white wires, 1 black and 1 red coming out of that terminal, of which the 14 white and 1 black connect to packs. That doesn't exactly add up for me, but it's apparently as designed.
For a typical "48v" (58v full, so sometiems called 52v) LFP pack, there should be 16 total cells, so 16 monitored wires from the cells to the BMS, one for the positive end of each cell group. That's what I count on that connector, in total.

To measure the voltages of the cells with them, you will need to put your voltmeter black lead on the main thick black wire that comes from the most negative end of the cells. Then move the red lead to each of the wires on that connector.

If you get 16 total readings of about 3v, then the cells are probably ok, and if you get those readings at the BMS circuit board itself it means all the sense wires are connected properly. When you get it working and can recharge it fully, it should read around 3.6v per cell.

If the first reading is 0v, it could mean the first wire is the pack ground...if so then you should *also* measure at the main battery positive *if* it connects to the BMS board, at that point. (if it does not connect to the BMS there's no point as it wont' be being used to monitor any cells).

Otherwise it may mean either a problem with the connection to the cell group on that wire, so you should then measure that cell group at the cell + terminal. If it still reads 0V, then that's a dead cell group. :( If it reads 3v like the others at the cell, but not at the BMS, the wire is broken somewhere along the way. Either way, it would be why the pack isn't working.



If the pack is reading as it should everywhere, but the BMS is still not allowing charge or discharge, the BMS may have failed; this could happen if there is a connection fault at the battery discharge connector that allows arcing to happen (RF from the arcing can damage electronics)--but arcing leaves obvious damage you really can't miss to both halves of a connector like that.
 
amberwolf said:
For a typical "48v" (58v full, so sometiems called 52v) LFP pack, there should be 16 total cells, so 16 monitored wires from the cells to the BMS, one for the positive end of each cell group. That's what I count on that connector, in total.

If the red wire goes to a cell, yes - but I don't see that.

I think I found the 16th wire - the last cell on the other end has a big black wire that connects to the other end of the BMS board, soldered at the corner. This does seem to add up to the 16 x 3.1V full battery with all cells functioning, as that wire reads 49+V.

If the pack is reading as it should everywhere, but the BMS is still not allowing charge or discharge, the BMS may have failed; this could happen if there is a connection fault at the battery discharge connector that allows arcing to happen (RF from the arcing can damage electronics)--but arcing leaves obvious damage you really can't miss to both halves of a connector like that.

So far I haven't seen signs of stress of any kind, including heat.
 
Is the BMS on or off now? Normally it shuts or breaks the negative wire so if you measure the big black wire that’s going out from the BMS vs battery plus you’d see if it’s on. To do a simple load test you can hook up some 12V bulbs, 4 in series to the battery.

I’ve had sense wires that were loose in the bms connector sockets even though they all looked solid, this caused bms cutoffs since bms would sometimes measure zero voltage on some cells. you can try to give all sense wires a gentle pull and wiggle, one by one while connected to a load.
 
On the bright side, it doesn't look typical for a dead lifepo4. None of the pouches have swelled to twice normal size.
 
Yeah, the battery is fine. I get 50.2V at the terminals, and individual cells check out at 3.1V. All the connections look good to me, nothing loose or broken. No signs of heat. Battery out from BMS remains 21V. I'm stumped so far.

  1. Supposing the BMS shut down to 21V after detecting a low cell or something, what condition has to occur, for the BMS to come back to full battery voltage?
  2. Supposing the BMS is toast, could my controller or some other part of the system have caused this? If I put my other battery back in, will I end up with two ruined batteries?

All tests so far have been resting voltage. I get that there could be a problem that shows up under load, but ... I am only guessing, but I can't easily believe that BMS would be designed to shut down permanently, forever, in this case. If it isn't permanent, then it seems like it needs to come back on at resting voltage. Which so far hasn't happened.
 
Write down the cell voltages is the best way plus you have a starting point like. Starting at neg end as # 1
1. 3.90v
2. 3.85v
3. 2.75v. ?


16. Xxx volt



Plus voltage of your battery on paper
Voltage of your charger.
To start.
 
donn said:
I am only guessing, but I can't easily believe that BMS would be designed to shut down permanently, forever, in this case. If it isn't permanent, then it seems like it needs to come back on at resting voltage. Which so far hasn't happened.

Some bms locks at fault (like an undervoltage caused by cells diving under load) and don’t turn on until battery is charged. Kind of logical even if battery returns to higher voltage as load is removed.
 
There have been BMSs (especially in OEM complete systems, like whole bikes, etc) that do exactly that, to avoid the possibility of fire if there has been tampering, or if there has been an event where any cell went below the safe low voltage limit (not just the regular LVC), potentially making it unsafe to recharge.

Whether yours is one of those designed for that, I don't know.

There are also some that just "crash" (because they often are run by MCUs these days, so have software that can have bugs or just be plain poorly written), and require rebooting. Sometimes this requires completely disconnecting them from power and leaving them for a while; some of them require a specific reset action, which may be shorting certain pads on the board, etc. Some have been reset by connecting the B- to the P- / C- momentarily. I don't know what yours would take (if it *can* be reset).

Yours could even just have failed FET or gate circuitry, so it can't turn on the input or output FETs. (often these fail stuck on, but they can fail open, too).



donn said:
I am only guessing, but I can't easily believe that BMS would be designed to shut down permanently, forever, in this case. If it isn't permanent, then it seems like it needs to come back on at resting voltage. Which so far hasn't happened.
 
Back
Top