DIY Electric Car Forums banner

1 - 20 of 31 Posts

·
Registered
Joined
·
46 Posts
Discussion Starter #1
Was about to take it for a test run after rebuilding the speed controller, but ran into other issues. I'm needing to test out some things to figure out what initiated the whole thing. It is actually kinda satisfying to have my own experience with a catastrophic cell failure.

These are Headway 10ah cells that I got from Headway Headquarters in September. Jeffery commented about the headways in the Porche 911 having a 20% failure rate. It seems I haven't completely escaped that possibility.

I'll be receiving my own West Mountain Radio CBA3 this week and will test each cell's performance and capacity. Any thoughts as to what test scenarios I should run?

 

·
Registered
Joined
·
2,045 Posts
I have a CBA III and for testing single cells the most it thinks it can do is 29 amps on the LiFePo4 cells. What it can do reasonably is 20 amps. Mine overheats at 25A in 7-10 minutes. One problem I have had is with the Anderson power poles they used to connect the battery to the CBA III. They dont make great contact and I have to wiggle them to get decent contact at 20 amps. It doesnt have a way to remotely sense the voltage so you see the voltage drop in the wires and connectors. I used 12 gauge wire and the voltage drop across the wire and connectors is about 0.2 volts at 20A.

It makes pretty charts. You can graph the charge voltage and several options during the discharge.

I would suggest a 1C discharge and then a maximum rate discharge. Whatever you can get to work. Try the 29 amps it suggests and I would like to hear if you get it to work. That way you can compare the voltage sag and compute the actual internal resistance at different SOC.

I've considered buying the 500 watt expander which would let me test cells at up to 140 amps. Supposedly it has a way to tell the software the resistance in the wires so it compensates the voltage in the graphs.

For your headways the regular unit is probably good enough.

The video wasn't clear enough to see what happened. You may have had an internal short of that one cell which caused a thermal event and let out its magic smoke. That was a fairly tame event as these things go. But that is what is supposed to happen with LiFePo4 so you got some heat but no fire.

Several years ago when I was doing the software for the AstroFlight Inc hobby chargers I heard this story. A customer was having a problem with his charger and battery so he brought it to the factory so they could look at it. It was determined the problem was with the crummy connectors on the pack. He gave the pack to one of his guys and told him to replace the connectors. This guy picked up the wire cutters and clipped both wires at the same time shorting the pack for a fraction of a second. This was a 2AH 3S LiPo pack and that brief short of the outputs was enough to cause a thermal runaway and the pack caught fire a few seconds later. They tossed it outside the building and watched it burn.

Hope you figure out the cause!
 

·
Registered
Joined
·
1,378 Posts
Wow, thanks for posting that. I'll look forward to hearing what you figure out.

I did get another brand of 10 Ahr cell to vent like that, but that was with repeated discharges over spec, and it wasn't nearly as much steam. That cell must have been really hot. Had you been running the pack hard? I didn't see any flame, but is there any possibility whatever was venting was also burning?

I'll also ask the inevitable questions: It looks like there was no BMS, did your system do any kind of cell level monitoring? Had you top or bottomed balanced the cells? Did the charge just rely on overall voltage to determine when to terminate the charge?
Was about to take it for a test run after rebuilding the speed controller, but ran into other issues. I'm needing to test out some things to figure out what initiated the whole thing. It is actually kinda satisfying to have my own experience with a catastrophic cell failure.

These are Headway 10ah cells that I got from Headway Headquarters in September. Jeffery commented about the headways in the Porche 911 having a 20% failure rate. It seems I haven't completely escaped that possibility.

I'll be receiving my own West Mountain Radio CBA3 this week and will test each cell's performance and capacity. Any thoughts as to what test scenarios I should run?

 

·
Registered
Joined
·
46 Posts
Discussion Starter #4
Wow, thanks for posting that. I'll look forward to hearing what you figure out.

I did get another brand of 10 Ahr cell to vent like that, but that was with repeated discharges over spec, and it wasn't nearly as much steam. That cell must have been really hot. Had you been running the pack hard? I didn't see any flame, but is there any possibility whatever was venting was also burning?

I'll also ask the inevitable questions: It looks like there was no BMS, did your system do any kind of cell level monitoring? Had you top or bottomed balanced the cells? Did the charge just rely on overall voltage to determine when to terminate the charge?
I have no BMS on this pack. Frankly my intent with this whole project is to have a quick use cycle to observe the merits of the whole BMS debate for myself.

These cells have less than 5 minutes of use. I am working with original factory charge. I have not yet used the charger for them at all! My plan was to get the bike running and use the batteries up a bit to get them discharged so I could then bottom balance them. All measured voltages were between 3.3 and 3.5 before assembling the pack.

If you check out my other video from October 8, you can see the majority of the use on these cells just before I roll of camera. The mosfets popped about 10 seconds into the run and I've been spending the time since rebuilding/reevaluating the speed controller circuitry.

The cells had each been around the 3.3v range when assembled, there was practically no use on them - just spinning up the tire a few times in air. The pack sat out in my garage for a month and voltage measured at the pack was 53.8v; no notable self-discharge, so I didn't inspect further.
 

·
Registered
Joined
·
2,222 Posts
Wow, thanks for posting that. I'll look forward to hearing what you figure out.

[snip]

I'll also ask the inevitable questions: It looks like there was no BMS, did your system do any kind of cell level monitoring?
WHAT! There was a Lithium battery fire. A BMS had to be involved!

BMS = fire, no BMS = no fire -- I read it on the interwebs :D
 

·
Registered
Joined
·
4,044 Posts
The resting voltage with these doesn't accurately indicate the SOC or that they're charged. I've had cells resting at similar voltages and one would take one or two Ah more than the other.

Did you even charge them up one by one before installing? Equalize them after installing (charge groups)? Were you monitoring voltages?


How many amps were you discharging the cell when it had the failure?
 

·
Registered
Joined
·
46 Posts
Discussion Starter #7
The resting voltage with these doesn't accurately indicate the SOC or that they're charged. I've had cells resting at similar voltages and one would take one or two Ah more than the other.

Did you even charge them up one by one before installing? Equalize them after installing (charge groups)? Were you monitoring voltages?


How many amps were you discharging the cell when it had the failure?
I guess not all the details are typed out, and only in the audio of the video that I made and some is only inferred from them! Sorry. The most critical detail to outline is that as mentioned in the video's audio track, there was some sort of a short - so much for getting a credit; I can't ask Jim to eat the cost of something that was completely out of his control. - I only this evening was able to open things up and confirm that there was a short somewhere.

What i'm having a hard time with is understanding where the short started. The power and precharge switches were off! I am currently suspecting the digital voltmeter I have installed in the speedometer cluster - but here's the rub- it has only 18awg wires that are hardly adequate to carry a current beyond that of the cells rated power capacity. This leaves me to needing to rip out all my wiring and inspect every inch of every piece.

I do know I had a short; I have no idea how much current it would have been pulling. I never charged any cell. They were all above the plateau of the nominal 3.2v resting voltage - meaning they were all in the upper end of the charge curve. I do subscribe to the notion that you cannot trust the voltage of a pack in active use, charging or discharging. However, a relative SOC can be deduced from a standing voltage. Or would you recommend that I put a specified load for a specified time to knock the cob webs off before measuring voltage? And I assume you suggest equalizing them based on apparent voltage - if voltage does not correspond to SOC, what would that matter? And how would you recommend segmenting a single string of non-paralleled cells?
 

·
Registered
Joined
·
4,293 Posts
... However, a relative SOC can be deduced from a standing voltage.

no, it cannot.

you MUST either fully charge or discharge to have any real idea of balance. best to balance whole set in parallel of course, but AT LEAST take individuals cells to a specific low, or end-of-charge voltage, to balance before use. resting voltage means almost nothing w.r.t. SOC with Li cells.
 

·
Registered
Joined
·
4,044 Posts
I bet that negative wire and fuse that you run right next to that cell shorted through that heatshrink somehow. It's routed right next to that cell, and could carry full pack current. If it was low, it could have gone into overdischarge.

Is that cell shorted? or is it open circuit?

What is the voltages on all of the other cells?
 

·
Registered
Joined
·
363 Posts
Another possibility is that the cell shorted through it's casing, which are live (under the plastic wrap). This would explain why damage was limited to one part of the pack (if that is the case).

Steve
 

·
Registered
Joined
·
46 Posts
Discussion Starter #11
I bet that negative wire and fuse that you run right next to that cell shorted through that heatshrink somehow. It's routed right next to that cell, and could carry full pack current. If it was low, it could have gone into overdischarge.

Is that cell shorted? or is it open circuit?

What is the voltages on all of the other cells?
Haven't openned the pack up yet. Im letting it air out, but I plan to dig into it more tonight. The vented cell is physically adjacent to the fuse, but it is not adjacent in the circuit. So if it did short through the can to the fuse, I expect the cells on either side of the circuit to be dissimilar.
 

·
Registered
Joined
·
1,378 Posts
I bought a pack of 10 Ahr HiPower cells to test. From the factory all the cells were very close in Voltage (3.27V to 3.3V). On the first discharge, after only a 33% discharge one cell dropped 0.4 V below the rest. So even though the delivered voltages were close, the state of charge was not!

After bottom balancing all the cells delivered spec capacity and stayed closely matched, until I vented some by running over spec.

Anyway, an educated guess is one cell was lower in capacity, and it was greatly overheated by being over discharged. I'd suggest doing a bottom balance first next time. Despite what Jack Rickard says, do not assume your cells are delivered with equal states of charge!

If you'd like my test numbers:

http://explodingdinosaurs.com/9electric/celltest
 
G

·
Guest
Joined
·
0 Posts
Nabil,

Have you figured out any thing? Got more photos of the damage? We are waiting for results of any kind. :)

Pete :)
 

·
Registered
Joined
·
46 Posts
Discussion Starter #14
Another possibility is that the cell shorted through it's casing, which are live (under the plastic wrap). This would explain why damage was limited to one part of the pack (if that is the case).

Steve
Congratulations Steve, your hypothesis has been the most correct so far.

I have learned a couple things: do what you intend to do, and optimism has no place in design.

Indeed the reason that that specific cell vented was that the mounting strap for the battery box made contact with it. I had intended to put another plastic liner between them, but didn't get around to it yet...

It still seems that it all began in the super expensive hong kong special voltmeter. I had contemplated putting it on its own DC - DC isolator, but again got lazy and didn't.

I am still ripping the wiring out of the frame to discover any more burnt wires, but it seems the primary short path was through the vicor 48-12v isolated regulator. The high side is still isolated, but the 12v side is 0ohms across the outputs.

In short, I'm rewiring the whole bike to be cleaner and better isolated.

Also, I am using the WMR DBA3 to see what each cell looks like still and bottom balance them while i'm at it!

Oh, also - the electrolyte destroys acrylic. Fractures started where the fluid had pooled and spread out from there. The battery box is a complete loss - I'm actually glad for the opportunity to make a better one anyway!
 

·
Registered
Joined
·
46 Posts
Discussion Starter #16
I've been having a most exciting time with my new CBAIII.

I have been doing a two stage approach - pulling each cell to 2.0v at 28a, which is right at 3.5c. Then pulling them down to 1.8 at 1c to have an end resting voltage of 2.8 (idealy). The actual ending rest voltage has been slightly under that at 1.7 and some change... one being 1.69, I think.

Attached is a pdf with 10 of the cells analyzed. I started from the end furthest away from the burnt cell.

There are some dips in the voltage that it then recovers from - I measured with my muliti-meter at the screws on the terminal and the voltage stayed steady throughout the dip and recoveries. I'm going to revisit the solder joints on my test lead in the morning.

Also, I'm doing this in my garage with an ambient temp of 13.5c or around 55F, i think - rather chilly and I can occasionally see my breathe. There is a fairly consistent increase in voltage after about a minute or two by a few hundredths of a volt. Another trend that I noticed is that around the 4.5Ah mark on them all, the voltage declined at a gentle curve then a very linear grade nearly to the end.

Enjoy!
 

Attachments

·
Registered
Joined
·
4,044 Posts
That's good to know that the cell didn't fail on it's own and that it shorted to the case.

why on earth are you overdischarging your cells to 1.7-1.8V? That's not good for the cells. Overdischarging will damage the cell.
 

·
Registered
Joined
·
46 Posts
Discussion Starter #20
Wow, looks like those cells were way out of balance for being only 8Ah cells. You might do a proper charge on them and then do another capacity test to see how closely matched they are.
That's the plan. I have them all bottomed out now and will charge tonight. My resting bottom voltage target is 2.8. I find that those at 2.9 - 3.1 resting voltage had around .09ah draw at 1c to get them +/-.05v of the 2.8v target.

My bottoming process was a dual stage. 3.5c to 2.0v (that's where I got the spreadsheet linked above), then a second stage of 1c to 1.8v. The first stage ah counts can be read from the spreadsheet. I have not compiled the second stage data yet, but it seems an average ah drawn out was .8ah! So yeah, I'd say I'm most certainly on the face of the cliff. (I'd be interested in anyone's data on these cells in this region.) This process left most at 2.8 - 2.88v. Three were above that and were treated to the process explained in the preceding paragraph.

Since the initial state of charge was unknown, I'm not sure what conclusions can be yet deduced. The second charge/discharge cycle will be more informative, Im sure.
 
1 - 20 of 31 Posts
Top