High Cell Delta - Bad cell / Bad Connection?

[Philip Schuchardt]

It was a cloudy day today, and I drained my batteries to 39%. 

My cell delta is really high, 200mV on discharge!! On cell 2. 

I've been noticing that cell 2 seems to have high deltas while charging and discharging. It also seems like the first one to get to 3.55V while charging. Although the SBMS shows 100% for about a hour before cutting off charging. 

When discharging from 100% to about 60% the deltas are < 10mV. 

I have 4s configuration with 4 sets of 2x150Ah BLS cells. 

Any advice on what to do to try to get the cell voltage back down to 10mV?

Thanks,

Phi|ip

[Dacian Todea]

Philip,

Your pack is not at 38% but at around 5% at most so you should stop discharging else the cell 2 will be at 2.8V very soon.

You have a hugely imbalanced pack mostly the cell 2 that seems to be at least 30% below the other 3 cells or it may also be that you have a dead cell (one of the two paralleled as cell 2)

You will need to check those two cell making the cell 2 they should not be bulged but if you say that cell 2 will not get to full charge it may be good just horribly unbalanced so you may need to manually charge that cell to catch up with the other.

The SBMS0 can balance but such a large imbalance about 100Ah or so will take many months to do as it may correct maybe at most around 1Ah per day so over 3 months to correct this huge imbalance.

[Philip Schuchardt]

So this morning I checked the voltage of the battery cells. Everything seem balance. Is this normal behavior for a bad cell?

It seem unlikely that the cell that's out of balanced, with a SOC 30% difference, would rebalance in 7 hrs? I'm currently running a load on it to try to reproduce the issue.

[Philip Schuchardt]

Here's a picture of my current setup:

Not sure if the photo helps at all. I don't see any obvious cell bulging, they all look good. I do have the batteries well secured since they live in a van. 

[Dacian Todea]

It is likely not a bad cell just very imbalanced pack. But I do not understand why the charging stopped ? 

I will like to see the battery when fully charged at 100%

Bellow is a charge discharge curve for LiFePO4 to understand what happens. Notice the voltage curve is very flat on most of the SOC range.

Notice that at 10% a cell can be 3.2V even above that while charging and the other cells can be at 30 or even 40% and will have about the same voltage.

Voltage will drop very fast only when cell is below 10% SOC 

So is likely your cell 2 is almost empty 5 to 10% and the other cells are at maybe 30% or more.

[Philip Schuchardt]

The charging stopped because I turned off the PV via a breaker to try to reproduce the problem.  I put a 150W to 250W watt load on for about 1.5 hours and drained about 8% of the battery. Cell 2 seemed to be higher end of the voltage delta during the test. I discharged from about 24% down to 24%. I would assume that if cell 2 as at 5% to 10% SOC, before the test, it would have dropped to below or close to 2.8V. Cell 2 didn't do that.

It's confusing because cell 2 was 3.089V last night with about the same load, and I only charge maybe 25Wh this morning (probably less). 

I reset the Wh sum mid test, so it's just a partial number in the screenshot.  The pack stayed around 7mV to 15mV delta during test discharge discharge. 

I'm not sure how the pack got unbalanced. I topped balanced the pack before installing it about 2 month ago with new cells. Maybe I didn't have the bus bars tightened all the way when top balancing. I guess, can I leave it be, and let the balancer do it's job? If I can limp along without manually balancing them for the next couples months that would be great. I'm currently in the middle of the woods and I need the batteries to operate to do my job. 

You can also see in the 1m graph that the inverter cut out after dropping below 25% SOC and then I turned the PV back on. 

It also seems that I can't upload images to the forum? Server says it's refused... I've been externally linking them. 

[Philip Schuchardt]

I'll try to get the pack fully charged. It's sunny but very smoky here. It'll probably get up to about 75% to 80% today. 

It would be nice to download the data directly from the SBMS0. Might not be possible with how my SBMS is currently mounted. USB port is blocked (I think). 

BTW, Dacian, you're super helpful!

Phi|ip

[Dacian Todea]

Philip,

The only anomaly is the photo you sent when SOC showed 38% and cell 2 was 3.029V. Was that 3.029V continues or was just a sporadic low reading of cell 2 ? 

As it will not make sense that cell was 3.029V at 38% but then at 24% was 3.167V

[Philip Schuchardt]

Last night it was at 3.029V and rapidly falling similar to the discharge graph you show above. I also thought I only have 5% of the battery left on the cell last night. I also observed cell 2 doing a similar behavior last week when we discharged the pack to measured ~40%.  

It's so confusing for it to go back up to 3.221V and the discharge to 3.167V this morning at 24% with a good load (200W) and with no PV input. 

I do have some beefy bus bars on the cells. 1" x 1/8" (sorry for the freedom units). I wonder if they're not making full contact with the cells. I checked the tightness this morning. They all seem pretty tight. Don't want to strip threads on the batteries so I'm being pretty gentle.

Phi|ip 

[Dacian Todea]

The cell connection can not be a problem as they will not have this behaviour. Your describe behaviour is very strange and not heard about that before.

If you have a multimeter and this thing happens again please measure that low cell to make sure that is the real voltage of the cell and is not some sort of monitoring problem.

Maybe one of the two cells is not making good contact but hard to see how it will behave like this and then be able to make good contact unless you tightened the bolts in the meantime.

[Barry Timm]

FWIW, I have a similar (but less severe) thing happening with my 16 cell 24v battery. Cell group #5 (2 x 150ah cells per group) seems to get depleted faster than any of the others under load, but when charging, it appears to be charged faster than the others, according to the SBMS0 monitoring screen. The strange part is that the cell groups ultimately all get within the 11 mV balancing threshold at the top of the charge. My only real concern with this is that when I use a high current device like an electric kettle, that cell group 5 really drops much lower, and I'm worried that that could ultimately cause my loads to be cut off if the battery gets much lower. It has not happened yet (battery loads cut off) as my overall battery SOC is generally never below 50%, and usually between 75% and 95%. 

FWIW, I raised my SOC recharge set point from my earlier 80% to 93% in order to cycle the charging between 90% and 93% after the initial 100% charge. The battery has always reached 100% each day.

I figured that I have a loose battery connector on cell group 5 or the signal wire to the SBMS is not well connected, but don't feel/see any issues.

[Dacian Todea]

Barry,

Your problem seems to be much simpler and is related to a bad connection between cells. You should find that connection and maybe screw the bolts a bit harder there.

If you have a multimeter you can measure to see with connection the one with the problem as it can be one of 3 connection points depending on where your sense wires are connected. 

[Philip Schuchardt]

Barry,

I usually try to keep the batteries at high SOC. We've been doing some late night coding / PhD paper writing. Coupled with smoke, clouds and the fall, doesn't equal great solar output or high battery SOC. At least we can always drive to more favorable conditions. 

Dacian,

After going hiking for most of the day. I produced about 2kWh of solar. Used about 1.1kWh (significant other working on PhD). It was at about 52% SOC. The deltas were good <10mV.

Several hours later, I'm now at 42% SOC and 8mV of delta. So it seems that things are working okay. 

I'll continue monitoring it. If the issue comes up again. I'll pull out my multimeter like you suggested and test the cells. My multimeter is only good down to 0.00V. But it should be good enough to detect 200mV delta. It just takes me a bit of time and effort because the batteries are buried under the van's couch. I did tighten one of the bolts maybe 1/10 of a turn this morning. 

I've also been working on a cross-platform app (desktop, macOS, Linux, iOS, android or anything that runs Qt) for directly connecting to the SBMS. It should allow me to graph cell voltage overtime. I definitely want to have a graph of cell voltage vs SOC. Would there be any other graphs that would be useful?

I'll keep you all updated. :D

[Dacian Todea]

If all works correctly you will see something like the generic graph below for the cell voltage vs SOC.

As you see LiFePO4 voltage is useless in estimating SOC thus the SBMS0 uses energy counting to calculate SOC and top charging to correct the SOC to 100%

Is easy to see if cell is full 100% SOC or empty 0% SOC base on voltage but nothing else intermediary as at 3.3V you can have 90% SOC or 40% SOC is just impossible to estimate based on voltage.

Since battery in an offgird system will be fully charged almost every day even multiple times a day is a good point where to calibrate the SOC to 100% while empty battery 0% SOC should ideally never happen so SOC correction in the SBMS is based on full charge 3.55V.

[Philip Schuchardt]

I've drain my pack down to about 25% twice now, and haven't reproduced the issue. Perhaps tightening 1/10 of a turn on one of the cells helped. I don't know for sure. Pack seems to be well balance and working as expected. Thanks for all your help Dacian!

[Dacian Todea]

Philip,

Glad to hear.

[Don Fukushima]

Most of what's been discussed applies to LFP and LIon except for the specific chart showing SOC level/voltage. This one is plausible but would like a thumbs up that it is accurate in conveying the realistic voltage range is 3.8 - 3.3v where about 80% of pack capacity.

thanks

On Monday, October 19, 2020 at 10:58:05 AM UTC-7 electr...@gmail.com wrote:

Philip,

Glad to hear.

[Dacian Todea]

Don,

Not sure where you took that graph form but no Lithium battery type has that type of discharge curve. Discharge curve is as flat as a LiFePO4 but voltage is way to high so it is not that and all the high energy density cells like LiCoO2 used in laptops and phones and the NMC types use mostly in electric vehicles have a much more linear voltage discharge curve vs SOC.

I'm fairly certain that is just a fake graph as I'm not aware of any rechargeable battery with those characteristics. It even says typical Li-ion discharge curve meaning it was likely done by someone that has no idea about Lithium-ion.

[Don Fukushima]

What is the li-ion voltage at 80% and 20%? That is the true figures I'm looking for. As you say, it's hard on the batteries to charge them too high or drain them too low but it's a balance to get the most power from them without overstressing them. 

[Dacian Todea]

Don,

You can not guess SOC based on cell voltage.

Li-ion is a very generic term that includes all Li-ion type batteries all very different in characteristics from each other.  A few example of Li-ion chemistries are LiFePO4, NMC, LiCoO2, LTO

Below is a graph of voltage vs SOC for LiFePO4 and notice how flat the voltage curve is a dependent of charge and discharge rates so cell can be 3.3V and SOC for that cell can be 90% but can also be 40% there is no way to know based on voltage. When cell is 3.55V assuming not ridiculously high charge rates or cell with very high internal resistance then you can be sure that cell is at 100% SOC so you can set the SOC counter at 100% and that is how the SBMS will calibrate the SOC based on a full charge.  From there the SBMS know that battery capacity that was introduced by user say 100Ah and based on battery discharge and charge current over time it can calculate fairly accurately the real SOC. Say battery was full and there was a constant load of 20A for two hours that means 40Ah where used form battery then SBMS knows to calculate the SOC as 60%