Root cause analysis of recurring charging interruptions in Pylontech US2000C battery modules

Other languages German-speaking area
, , , ,
1

I have been running a Victron ESS with a total of 16 Pylontech battery modules connected in parallel since 2022. Over time, increasingly frequent charging interruptions occurred during full charging, triggered by individual cell groups with unusually high cell voltages.

What began as an initial fault investigation has since developed into a fairly extensive technical study.

Among other things, I have:

  • recorded the cell voltages, currents, temperatures and other diagnostic data from all 16 battery modules via the Pylontech serial interface,

  • analysed the behaviour of the individual cell groups during charging,

  • carried out internal resistance measurements at module, cell-block and cell-group level,

  • opened and examined affected battery modules and 5S2P cell blocks,

  • analysed the BMS, including the passive balancing circuit,

  • and examined the plausibility of the data output via the Pylontech diagnostic interface.

One significant finding is that the observed charging interruptions in the modules examined can be explained by a greatly increased internal resistance in individual cell groups.

Under charging current, the terminal voltage of these cell groups rises disproportionately. The BMS responds by limiting or interrupting the charging process. Once the charging current is removed, the cell voltage drops again and charging can resume.

The passive balancer in the US2000C units examined uses two 51-Ω resistors in parallel for each cell group. This results in a balancing current of approximately 130 mA. When the internal resistance of individual cell groups is significantly elevated, this current is too low to effectively compensate for their voltage rise during charging.

There were also some interesting findings concerning the BMS and the diagnostic interface. In particular, based on my investigations, the information provided about balancing activity should not be interpreted as the actual switching state of the balancers in all battery modules without further plausibility checks.

I have now compiled the entire investigation in a technical report and published it on GitHub together with measurement data, Excel analyses, photographs and the Delphi logger software used:

The report is in German and contains a complete description of the measurements and results.

The findings may be of interest to other operators of Victron/Pylontech systems whose batteries are ageing and are experiencing larger cell-voltage differences or recurring charging limits or interruptions.

I would be particularly interested to hear whether other operators have made similar observations and whether increased internal resistance in individual cell groups can likewise be demonstrated in their affected modules.

2

We’ve also observed extreme changes in cell voltage in larger Pylontech systems with 16 or more batteries. One customer with 24 batteries had 11 fail after four years (they had literally swollen up). In the meantime, all the battery cables have been installed in alternating directions, so that the cable runs are the same length across the entire bank. In addition, all the connections were tightened to the same torque and then checked during a load test with a thermal imaging camera. The batteries were replaced under warranty by Pylontech, but that involved quite a lot of paperwork and back-and-forth with the Chinese.

At the moment, everything looks good. But it’s only been six months or so.

3

Probably all US2000C models? So far, only the 2000-series have been affected in the posts here…

4

Yes, exactly. Especially when they have never been charged from the grid, but only using solar power. Then there are customers who never fully charge their batteries to 100% using solar power either, so as not to give the electricity to the grid operator for free.

We recommend that our customers fully charge their batteries at least once a month so that the balancing process can actually complete.

5

And many simply entered… set the maximum values from the description as well…

and that then leads to outages or errors in some cases.

6

Yes, I neglected that a bit myself too—I didn’t deliberately charge the batteries to 100% during winter. One of the cells in a US3000C did make me pay for it, although I don’t have any swollen cells. So far, this seems to be more widespread only with the US2000.

7

I don’t believe it has anything to do with an unbalanced connection. All my batteries were connected symmetrically. I also always charged them to 100%. My suspicion is that the balancer, with its 130 mA capacity, is underpowered. Because the cell pairs have different capacities, the same cell pair is then consistently overcharged and eventually develops a higher internal resistance.

As Pylontech has now stopped responding to my enquiry, I’m going to forget about the warranty and open up all the affected batteries. If I sacrifice two batteries, I can repair the remaining four affected batteries using the four intact cell modules from the two sacrificed batteries. I’ll then install capacitive active balancers with a maximum balancing current of 5 A. They’ll just about fit if I build a substructure from PCB material — non-conductive, of course.

I’ll report back once this has been implemented and then monitor how things develop over a longer period.

8

Pylontech always responds to us, but it can sometimes take several weeks between replies. However, just today they arranged a collection for one defective battery. Whether they’ll actually accept it as a warranty claim is, of course, another matter.