Here is a graph of my 3 batteries. They're all Pylontech UP5000, bought about 6 months apart.

I'm curious why the yellow "battery 2" doesn't follow the others during the final charge phase.
This pattern occurs every day as the batteries reach 100%. The weird thing is that yellow battery doesn't like to climb up the voltage curve with the other two. But only after they've hit 100% and dropped their voltages back down, then the yellow allows itself to climb back up again, and soon after that, all three are showing 100% SOC, and their voltages are back in sync until this same point, 24h later. That dip you see after nearly reaching 100% was just a 5kw load that drew on the batteries a bit, it doesn't have any relevance to the issue.

The lowest I allow the pack to go is 25%, but I'm considering taking it down further once-off for calibration. Not sure how worried I need to be?

That's normal behavior for different age batteries , the packs balance at 89-90%. I see the same thing with mine.

I run them down to 5% every day for the last 5 years, currently they are at 93% SOH . There is also a lower balancing point between the packs at 11% SOC. 

43 minutes ago, Ben Harper said:

The lowest I allow the pack to go is 25%, but I'm considering taking it down further once-off for calibration. Not sure how worried I need to be?

I think the only explanation is that battery 2 disconnects because of cell overvoltage. Otherwise it's not possible to see different voltage on batteries connected in parallel.

Differences in internal resistance of cells might be the reason, causing balancer to operate the wrong way during charging.

8 minutes ago, kuba.cz said:

I think the only explanation is that battery 2 disconnects because of cell overvoltage. Otherwise it's not possible to see different voltage on batteries connected in parallel.

Other way around , Nr 1and 3 disconnects because they are full first there is no overvoltage, then Nr 2 catches up,the 89-90% pause to balance packs is a pylontech thing , i see the same behavior with the middle battery in my stack as the top and bottom battery receive slightly more current. A graph showing the current flow into each battery would confirm this.

Edited January 21Jan 21 by Nexuss

26 minutes ago, Nexuss said:

Other way around , Nr 1and 3 disconnects because they are full first there is no overvoltage, then Nr 2 catches up,the 89-90% pause to balance packs is a pylontech thing , i see the same behavior with the middle battery in my stack as the top and bottom battery receive slightly more current. A graph showing the current flow into each battery would confirm this.

Very interesting that you see the same with your middle battery. Also, I thought 5% was a low SOC to hit every day, but interesting to hear your SOH.

32 minutes ago, Nexuss said:

Other way around , Nr 1and 3 disconnects because they are full first there is no overvoltage, then Nr 2 catches up,the 89-90% pause to balance packs is a pylontech thing , i see the same behavior with the middle battery in my stack as the top and bottom battery receive slightly more current. A graph showing the current flow into each battery would confirm this.

If you look at the SOC graph, you'll notice that my middle battery's SOC goes down slower than the other two. Could this be because the effective cable length is further for it?

1 hour ago, Ben Harper said:

If you look at the SOC graph, you'll notice that my middle battery's SOC goes down slower than the other two. Could this be because the effective cable length is further for it?

The top and bottom battery in a stack will have the least resistance(because of the daisy chaining) and thus provide a little more current than the middle batteries. that along with different age /cycles on batteries will either cause the pause at 89% for balancing or not , i hardly every see the 89% pause anymore as i cycle past 2 balancing points every day . Atleast thats my theory haha!

@Nexuss I've just taken a careful look at the battery cable lengths, and if you sum up the total length of black + orange cable, then all 3 batteries have equal cable length. For example, if you say that the long cables to the inverter are length "10", and the short connector cables are length "1", then the top and bottom batteries get 10 + 12 = 22 (counting all the cable length in aggregate). The middle battery gets 11 + 11 = 22. In other words, the top and bottom batteries have shorter orange/black, but longer black/orange. So total resistance of the two halves would be the same. Surely that's what matters?

Anyway - I'm still puzzled and vaguely worried about the middle battery's "underperformance". It seems like it's getting worse over time - i.e. the pause between batteries 1&3 reaching 100% and then waiting for battery 2, seems to be increasing. Next week I'm going to try flashing them all with the latest firmware, and then swapping them around to see if it makes any difference.

On 2026/01/21 at 4:34 PM, kuba.cz said:

Otherwise it's not possible to see different voltage on batteries connected in parallel.

That's what I thought too. But on my bank of 6 packs I observe the following: During charge and discharge all packs show quit coherent voltages with the inverter, taking measuring inaccuracy and voltage drop in the cables in account. However at full charge with no current some of the packs voltage deviate from others and from the inverter up to 1V. I have no explanation for this other than weird BMS. What I don't know and wonder is whether the terminal voltage readings are measured at the terminals or just the mathematically adding up of the cell voltages.

Edited February 11Feb 11 by Beat

Where is the voltage data coming from? What software?

For me, the voltages are coming out of this: https://esphome.io/components/pylontech/, so they're emitted by the batteries.

1 hour ago, Ben Harper said:

For me, the voltages are coming out of this: https://esphome.io/components/pylontech/, so they're emitted by the batteries.

Great, can you pull the cell high and low values and the current for the battery 2 at the same time as the voltages.

@Tinbum Yes - VMin/VMax is very interesting:

What does it mean though !?

1 minute ago, Ben Harper said:

@Tinbum Yes - VMin/VMax is very interesting:

What does it mean though !?

I would use battery view to pull the logs for the second battery and post here please. Redact serial number.

I think the second battery is disconnecting due to high cell voltages. Looks as if it may have a bad cell. The balancer does seem to equalize it though eventually, but not normal behaviour.

Edited February 12Feb 12 by Tinbum

Hi @Tinbum

This is batteryview shortly after it reached 100% SOC (which took 4 hours after the other batteries reached 100%).

I don't see anything in the logs. Would I only see alarms in the logs if batteryview was connected and recording while the alarm happened?

Also, why do you say to erase the serial number?

21 minutes ago, Ben Harper said:

Hi @Tinbum

This is batteryview shortly after it reached 100% SOC (which took 4 hours after the other batteries reached 100%).

I don't see anything in the logs. Would I only see alarms in the logs if batteryview was connected and recording while the alarm happened?

Also, why do you say to erase the serial number?

4 hours ! that's very long , what is your charging current like? if there is plenty of solar available that is not normal. Paste the history log here so we can check it ,erase serial number for internet weirdos . Also check the firmware version on each battery they should all be the same version ideally. Might be worth upgrading them all .

Edited February 12Feb 12 by Nexuss

41 minutes ago, Ben Harper said:

Hi @Tinbum

This is batteryview shortly after it reached 100% SOC (which took 4 hours after the other batteries reached 100%).

I don't see anything in the logs. Would I only see alarms in the logs if batteryview was connected and recording while the alarm happened?

Also, why do you say to erase the serial number?

The battery will record logs all the time, you just need to download them with BatteyView.

I suspect you have a cell in that battery that is going over voltage very easily/quickly. The battery disconnects from the bus (inverter/other batteries). The bms then re-balances the cell, the battery connects again, probably cell goes over voltage again and cycle continues until all cells are balanced. That's when it then reaches 100% SOC.

Serial number erased just as a security measure. I would expect you to be making a claim for a replacement as it's not normal behaviour. (Edit Cell 8 looks to be a bit different to the rest so it could actually be that that cell is always low causing the other cells to be higher).

Was this the first battery you bought in your setup?

Edited February 12Feb 12 by Tinbum

Yes it is indeed cell 8. I've checked now as it's discharging, and cell 8 drops much faster than all the others.

What are my chances for getting a warranty claim? I installed all my kit myself, so I don't have certificates etc.

Any place I should look around in the cape town area for somebody who can replace a cell?

That yellow line is cell 8. It's the culprit when charging, and also when discharging.

This battery was the 2nd one I bought, so I don't think it's just an age thing.

Your chances are good as long as the battery hasn't gone overvoltage ie above 54v. How old is it though?

This battery was installed in 2022. I guess I contact the store where I bought it?

4 hours ago, Ben Harper said:

This battery was installed in 2022. I guess I contact the store where I bought it?

I would try that first and if they aren't helpful then go to Pylontech themselves.

Did you register for the extended warranty?

I've contacted thepowerstore.co.za where I bought it from, and they're going to assist me getting the RMA done with the distributor. Thank you thepowerstore.co.za!

I did not register for the extended warranty unfortunately.

One more interesting plot. This is at night when the house is running off batteries. It's interesting how low the voltage of the bad cell drops.

My low SOC limit of 40% kicked in there at 4:30 AM, which is why the voltage stopped dropping. I normally run down to 25%, but since I've been getting this trouble lately, I've raised my lower limit.