How Do I Diagnose A Bad Cell In A Long Parallel String

Joined
Oct 17, 2009
Messages
2,245
Location
Republic of Ireland.
Hi folks.

I just test a 3.3v, 1s 16p LiFePo4 string of cells, at around 1c. The results were good, around 14.8 Ah recorded by my CBA III

I think I might have a runt in there though cannot be sure.

It seems to be that the string charges in an odd fashion, sometimes sitting at Xamps on my CC-CV power supply for an extended period.

Additionally, it seems to drop in voltage very quickly despite being fully charged.

I charged it up to 3.6v last night and around 12 hours later, it was down to 3.39v.

I have had much smaller strings keep their 3.5v and 3.4v ratings for weeks.

I am Constant Current charging it now on my desktop power supply, at around 6 amps. I will do that for a few hours then CV it until it is drawing no more current and is fully charged at 3.6v.

Then I am going to watch it over a week and see if it drops below 3.3v in that period. If it does I know I have a problem.

Question is.........can I still use the battery, seeing as it gives out good returns at 1c??? For a light or something??

Thanks.
 
most of the lifepo4 will lose the full charge resting voltage when they come off of the charger as they age with cycling.

but some of the cells may not actually reach full charge if their internal resistance is higher than the others connected in parallel.
 
dnmun said:
most of the lifepo4 will lose the full charge resting voltage when they come off of the charger as they age with cycling.

but some of the cells may not actually reach full charge if their internal resistance is higher than the others connected in parallel.


Ahhhh, that could of course be it. I made this string when I was young and inexperienced.

I was just wondering too, if anyone could help me with this:

when charging this long string, I notice that the voltage at one end of the string is usually slightly higher/lower than at the other end.

For example, right now, the voltage at one end is 3.368v and the voltage at the other end is 3.377v

Any ideas as to why this may be?

The charge lead is terminated much closer to the end of the string that is reading 3.77v than it is to the end of the string which reads 3.368v.

Reiterating, it is a 1s 16p LiFePo4 string made of supposedly genuine A123 18650's, which have passed muster at 10c discharge rates on their own.

Thanks.
 
do a binary search like AW showed before. but 3.39V is still essentially full charge for lifepo4 after it is off the charger.

to do the binary search you only have to cut one of the parallel straps, not both.

you should cut the one on the anode end since it should be easier to reach with the scissors.
 
What you're doing is a self-discharge test. To find the bad cell in a parallel group that is being held back by a runt (nice terminology) you have to separate the cells and run this test on them one at a time.

The reason your string has higher voltage at one end than the other is because of the impedance in the connections between them. If you had heavier wire joining them, or maybe better secured connections, this would be reduced or go away completely. Charge the group from the other end...does the higher voltage follow? How about if you connect the charger at the middle of the group? You can prove very quickly what is happening here with a power supply and a volt meter. With such high impedance between cells, it will take a long time for the parallel group to balance out to the same voltage (as they inevitably will, given enough time.) If the measured voltage on the charger end of the pile is so much higher than at the other end, the charger will shut down before it's job is really done, and the low side will pull everyone to an average that is somewhere in the middle of the spread. You say you've put a charger on with no auto shutoff and waited for current to drop low. That is what you should do, but if you're still seeing this problem, you haven't waited long enough OR you should try charging from the opposite end (or both at once)to see if that helps speed balancing OR you have really bad cell or two in the mix. Separating them is the only way to find the bad guy.
 
Hi, and thanks for your detailed reply.

The reason I did the discharge was to see what the overall capacity of the string was. The average returns were good, over 900mah per cell up at 1c discharge.

That the cells did not collapse totally was encouraging.

Interestingly, the cells are terminated with solid copper sheet. So no resistance problems there.

I noticed that when I first charged the cells, they dropped very fast down to the 3.3X v range.

When I discharged them, at 1c, the voltage dropped to 3.2X very fast, then stayed in that range for some time.

I have since recharged the string, and, over the last 12 hours, it has HELD strongly in the 3.5x range.

It is 3.508v as I write.

I think what may have happened is that when I recharged, I recharged up to 3.68v, really making sure they all got charged.

When I did a discharge test after that, this time at 2c, the string did not drop out of the 3.3v range despite doubling the discharge rate!!!

After 4.0 AH of discharge, my CBA III gave out, citing overheating.

I recharged the string, and it is holding steady now.

I think its just I have a cell in there that doesn't fully charge at the basic 3.6v.

I will keep you posted.

PS "RUNT" is a word I borrowed from somebody in here, not sure who.
 
Interesting. How thick is the copper sheet? Any possibilty of corrosion anywhere? Maybe the cells just needed a"wake-up cycle". I've never experienced that myself but it is a possibility if the cells were sitting unused for a long time. I'll be interested to see what you find.
 
wb9k said:
Interesting. How thick is the copper sheet? Any possibilty of corrosion anywhere? Maybe the cells just needed a"wake-up cycle". I've never experienced that myself but it is a possibility if the cells were sitting unused for a long time. I'll be interested to see what you find.

I think it is because they were sitting for a long time and because they were paralleled in differing stages of charge, back before I knew what I was doing.

The sheet is 0.1mm I believe, quite thin.
 
we have no pictures so no way to know what you have. what is a copper sheet and how can you connect a copper sheet to a cell?

if you connected the individual cells to each other in parallel then they are balanced in voltage by being connected through a common node.
 
The sheet was cut into strips and then soldered on.

Here is a photo.

Thanks.

DSC_0440_zpsfmhbyyum.jpg
 
Oh my....you DID make this before you knew what you were doing :mrgreen:

I stand by my original assessment. It only takes milliOhms to get a voltage gradient across such a long string of cells. You're probably correct that connecting these at different SOC's prolonged the agony, but there will be measurable voltage drop from the charge end of that string to the opposite side. It's really no different from the voltage drop you would observe from one side of leads going from a power supply to battery being charged. The resistance of the wire is not zero, and the voltage drop is not unmeasurable.

Are those A123 cells? If the cans are not isolated from the electrode stack, what you've done here is exquisitely dangerous, especially in a motive application.
 
They are purported to be A123 cells, yes. They have some of them at least , stood the discharge test. I discharged one of them at 20amps once, it gave up 900mah and was quite hot at the end.

Why has what I done "dangerous"....the cells are all isolated by their own heatshrink. Forgive my ignorance but I am not sure what is meant by "Electrode Stack"

As it happens, the battery is not complete, there are no series connections at all on the battery pack.

Thanks,
 
The Mighty Volt said:
They are purported to be A123 cells, yes. They have some of them at least , stood the discharge test. I discharged one of them at 20amps once, it gave up 900mah and was quite hot at the end.

Why has what I done "dangerous"....the cells are all isolated by their own heatshrink. Forgive my ignorance but I am not sure what is meant by "Electrode Stack"

As it happens, the battery is not complete, there are no series connections at all on the battery pack.

Thanks,

The cans on these cells are tied to the + electrode inside each cell. [The "electrode stack", also known as the "jelly roll" is the term for the anode and cathode set inside the cell--Li ions are shuttled from one electrode to the other during energy transfer.] Yes, the cell can is insulated by the heatshrink PET sleeve, but this is easily damaged by abrasion or high heat. On the negative terminal side of the can there is an additional white plastic disc contained by the PET that gives added insulation protection. This is here to prevent the connection to the - electrode from shorting to the can of the cell, which would result in a spectacular event. This plastic disc is pretty robust, but I wouldn't trust it to stand up to the heat it endured when all those big, globby solder joints were put in place. You apparently haven't made a hard short anywhere yet, but expose this thing to vibration, temp cycling, and time.....I wouldn't trust this to last under those conditions. Were it me, I would only use this arrangement for stationary applications, and only in locations where a nasty incident can be fully contained without your direct supervision.

If your discharge test results really are indicative of cell performance, these are some real dogs if they are A123 cells. I wouldn't expect the result you decribed unless you started at a low SOC and kept discharging until SOC was near 0% or even lower. You should be able to pull a steady 100 Amps from a single cell before the cans become "quite warm". If the connections to the cell were not very solid, that may also explain the heating. It doesn't take much series impedance in your connections to seriously impact high current performance--this would be similar to the cause of your voltage gradient across the string.
 
riba2233 said:
Good one! These cells sag like s*** at 30 A:

http://lygte-info.dk/review/batteries2012/A123%2018650%201100mAh%20%28Yellow%29%20UK.html

I don't want to imagine them at 100 A. I got a 880 mah at 5 A from new cell that wasnt used for a year and stored at 50% SOC.

I was talking about the 26650, not the 18650. Big difference. Move up to the 32113, and you can pull 200 Amps steady from a single cell. I've done this, seen it with my own two eyes on calibrated lab-quality equipment.
 
Thanks to people for the above replies.

With regards to getting 100amps from a cell, a 1Ah cell, without getting warm, I find that a bit optimistic, no :?: :?:

That is 100c. Not even the 26650 from A123, the known genuine ones, had that sort of a rating. They could give 30c burst, if I recall, and they still got warm.

As it happens I plan to use the battery for lights by splitting it into three 12v sections, each with its own balance tap, for some LED set ups.
 
The old 26650 is a 2.3 Ah cell, the new one is 2.5. "Quite warm" is a subjective description, so your definition may be different from mine. A little bit warm at lower current, sure.

I'll have to run a test here....I scaled down from what I've seen the 32113 do, but maybe the 26650 isn't quite as capable as I think. I know for a fact the 26650 can do 10 sec bursts at over 50C (over 100 Amps), but I haven't done a steady load test that high on a single cell.
 
Back
Top