Help diagnose cut-out please!

[bjm522]

Setup: 9c motor, infineon controller, cycle analyst, 48V16AH headway lifepo4.

Chain of events:

Break fork because i'm a heavy dude
bike sits 2 months while waiting for money for new fork, installed.
won't hold charge/stay on, find one dead cell (0 volts), rest of the pack has consistent voltage.
order new cell, charge to same voltage as rest of pack.
charge pack overnight on 15minute cycle wall timer. (always worked before, going to do the transistor modification to charger soon)
go on ride, dies after a mile.
check voltages, all at 3.35V, except one at 3.34V.

I would appreciate any procedure suggestions to help me diagnose the problem. ideas so far:

ride again without charging, then with charging, watch for voltage at which it cuts out
obtain 16ch DAQ module and gorilla tape laptop to bike to monitor individual cell voltages at cutout (overkill but possible, any ideas for something smaller?)

Thanks

 

[dogman dan]

I don't understand the reference to 15 min timer.

What exactly are you charging with?

Got a bms?

Have you fully charged and balanced the battery?

3.35v shouldn't have triggered lvc on a bms or controller, but it might have been lower under load.

Perhaps it's just the ol, loose pin in the halls plug trick or similar bad connection on the bike somewhere issue.

 

[bjm522]

the timer to which I'm referring is a daily timer that can be set in 15 minute intervals on or off, with a rotating dial. cheap $10 outlet timer.

Battery was fully charged and balanced

There is a red BMS. This is a kit from evassemble. I can read the cell voltages straight from the contacts on the BMS.

It's more likely there's a loose connection outside the battery, so I'll check the whole system thoroughly. I'm also suspecting that a cell might be going low under load, but it's hard to keep track individual cells while the pack is under load.

Will an open circuit in one of the hall sensor wires cause the BMS to cut power? I'm not sure if it causes the controller to trigger the BMS. Turning the controller on/off does not restore power. I'll research this.

 

[John in CR]

Did it work correctly before the broken fork?

With those at rest voltages the battery has plenty of juice. Does the cutout happen under load? I ask because it sounds more like the BMS tripping from too much current rather than low voltage on a cell.

John

 

[bjm522]

yeah, worked well before the break. I'll go on a few rides and check out the current draw and voltage at cutout.

 

[fechter]

Will an open circuit in one of the hall sensor wires cause the BMS to cut power? I'm not sure if it causes the controller to trigger the BMS. Turning the controller on/off does not restore power. I'll research this.

If a bad hall sensor caused the controller to draw way more current than normal, it could possibly trip the BMS overcurrent safety. If cycling power on the controller does not restore operation, it points to the BMS being tripped for whatever reason (over current or undervoltage cell).

Once it cuts out, what do you have to do to get it going again?

 

[bjm522]

apply voltage to charge port, even though the voltages are all ~3.35. In fact, it died once and I didn't bother getting it back on- took it apart right away to check cell voltages. Usually slapping the charger on it will bring it back to life, assuming from some logic in the BMS.

 

[dogman dan]

Yeah, that does sound like a bms trip. My best guess is the timer is not letting the charger run long enough to fully balance it. 3.3v could be 80% charged, or only 40% perhaps even less. You must have one cell sagging like hell under load, the one that's more discharged.

Get all the cells charged to 3.65v and then see what happens.

Possibly a bms wire is disconnecting intermittently? That would make the bms think 0v, and shut off.

 

[dnmun]

15 minutes is not long to charge. if you measure the voltages on the BMS for each channel, that will tell us what each cell is doing. just put it on the charger and charge it long enuff to get one of the cells up to 3.9V where the BMS will cut off the charge, and measure each cell and post up the voltages here. measure to 1/100 volt. count from the bottom of the pack, 1-16.

 

[bjm522]

Will do. The charger's only 58.4V, which would be 3.65v/cell charging voltage, which is the spec for this battery pack. Where are you getting 3.9V? also, my cheap voltmeter only has .01 precision at 0-20v range. 15 minute cycles are enough time to reach 58.4V. the charger cuts out before the timer does when i go out in the morning to ride. keep in mind that the timer basically causes it to charge for half the time (15 on/15 off). I know that there's at least 10AH on the cells, because i had to charge the new cell for at least that long @ 1A from a PSU. I think it's safe to assume full charge on pack, as long as full = 3.65V for lifepo4. Seemed to work OK before.

 

[dnmun]

you have to measure the voltage of each cell. pack voltage is not relevant here. use the 20V scale, for cell #1 you have to put the black probe on the shunt wire which is directly connected to B-. and the red probe goes to the first wire on the sense wire plug. then measure between each pin on the sense wire plug and post up the numbers so we can diagnose which one is the problem.

when one cell goes to 3.9V during charging, it shuts off the charge before the other cells are full so you have a pack with cells which are partially discharged when you think it is charged. that's why the BMS shuts off for LVC early.

 

[bjm522]

It's charging now, and i'll post all cell voltages at cutout. as in the original post. cell voltages were 15@3.35, 1@3.34.

 

[bjm522]

3.45 3.67 3.67 3.67 3.7 3.69 3.69 3.70 3.67 3.71 3.70 3.68 3.68 3.68 3.65 3.65

The first cell is the one that was replaced. I charged it up to the same voltage as the rest of the pack, but I guess that wasn't quite up to capacity. I'm going to slap the PS on it to bring it to 3.67v.

As I said, there's at least 10 Ah on there, but maybe this is enough of a difference to flip the BMS.

 

[dnmun]

yes, 3.45 is only about 70% charged. you can use a single cell charger on it or since all the other cells are high, you can put a big resistor or light bulb across all of them using a jumper wire with alligator clips except for the low one while it is charging and the load will drain the charge off the high ones so that the charger will finish up the low one as you drain the high ones. just be sure to take the load off when the voltages become even. then you can let the BMS finish balancing it. should solve your problem.

 

[bjm522]

Okay, that cell took an extra .5Ah charge to bring it up to the rest of the pack. It's all balanced now, and I took it for a ride, and works for about 100 feet then cuts out. voltage was 44.5V at cutout, under 600W load. This is 2.78V on each cell under load, assuming they're all balanced in that condition.

bad controller? bad bms? bad battery pack? Wish I could measure each cell under load.

 

[bjm522]

Additional info from a few more rides today:
limiting to low amp on cycleanalyst works fine, and if i lay on the throttle extremely gently, i can get to decent speeds (15-20mph if careful).

If I put the throttle on too fast, it seems as if there's a power spike. current jumps to at least 15A (~700W) for a moment, even when limiting to 3A. Maximum current was 15.80A according to CA. This 9C motor is supposedly 600W max.

The hall sensor appears to be okay. Each wire was high/low at nice even intervals all the way around.

I turned the IntAGain on CA up to the max to see if I could eliminate that spike, but it didn't work. I checked the grounds and everything is nice and tight, and same with the BMS cell wires. I really hope I don't have to baby this thing around town- that's no fun!

 

[bjm522]

I'll assume everyone's stumped now, and I'll just start throwing parts at this til it works unless anyone else has an idea.

 

[dnmun]

"This is 2.78V on each cell under load, assuming they're all balanced in that condition."

you cannot measure the cells? how can you make such an assumption?

 

[www.recumbents.com]

Either there is another cell that is damaged, which is dropping below LVC under load, or the BMS was somehow damaged and is now cutting off at a very low current. Have you tried just leaving the battery plugged in on the charger overnight? LifePo4 pack are fine to leave plugged in for an extended period, and that is sometimes the only way to get them to balance properly.

-Warren.

 

[999zip999]

Bring thta low cel to 3.8v and not over so watch and let sit a little and loss surface charge. then you now it's full. I would put some sense wires with plugs for now and later.

 

[dnmun]

just leave him be. he decided it is best to just throw parts at it so lets see how that turns out. offering free advice is an expensive waste of time.

 

[bjm522]

oh, feisty in the forums today. in response to your first post, no, I can't measure the cells under load with the bike going. My arduino doesn't have 16 inputs. I specifically said 2.78V on each cell by just taking the average, and stating the premise for that conclusion. If you had just said "that's not a safe assumption to make", then I could really investigate that path by getting a microcontroller with more inputs and do some on-the-fly datalogging.

If you're going to be a douche, don't bother responding.

To the others:

the pack's balanced now, just sitting around. I left it on the charger overnight before the recent rides. The pack is charged to max charger voltage, and has been over several surface charge/discharge cycles. Yes, this was verified several times. The only possible problem with the cells is one or more going below 2V under load because of degradation. One or more cells being damaged is a reasonable idea, considering the pack sat off the charger for a few months. once again, without the datalogging, I can't see a way to check this under load. I was considering wiring up a D-sub for the sense wires anyways, but with current equipment I can only log half the batteries at a time. I suppose it will have to do.

I appreciate your help, and I wouldn't be asking it if it were an easy thing to fix. I don't want to throw a new controller/BMS at it, but that would eliminate those as a source of the problem.

 

[fechter]

The BMS has an overcurrent sensing circuit that is supposed to protect against excessive discharge rate. The motor controller also has a current limiter, so the two are a bit redundant.

It may be possible to bypass the current sensor on the BMS and see if it still trips under load. If it still trips, you could assume it's due to a cell voltage sagging enough to trigger it. The current sensor is ususally comprised of multiple resistance wires on the board (shunt). This is usually on the negative side of the pack connection. If you reroute the main pack negative wire directly to the cells, you could bypass the shunt.

If this clears the problem, you may have a bad resistance wire or bad solder joint on one, or some kind of fault in the detector circuit.

 

[bjm522]

Excellent, thank you for that idea. I'll check out that path. I'm not sure what you mean by that re-route ("directly to the cells") as there is already a wire from ground on cell 1 to terminal 0 on the BMS (there are 17 wires total from 16 cells, ground on first cell and every positive terminal, BMS terminal 16 is pack positive wire).

I'm going to remove the BMS and ride straight from the cells for a lap around the block. If a cell really is dropping below 2V, it's already dead and I don't mind damaging it. Should not cause any damage since the pack's charged and balanced. If this fixes the problem, I'll do your check once you clear up the above confusion to see if I can fix the BMS. If not, I'll get a new one.

I'll post results from the test tomorrow. Thanks again for the suggestion.

 

[fechter]

On the BMS board, there should be a heavy wire from the pack negative (from the cells) and a heavy wire leading out to the load. Somewhere between these should be the shunt.

Can you post a picture of the BMS board?

If you bypass the BMS completely and run it pretty hard, any bad cell should be pretty obvious after the discharge.