Excessive battery solar charge current when big AC load switches off

Question

question

beat asked Sep 5, '23

Excessive battery solar charge current when big AC load switches off

My system is an ESS with Cerbo GX, Solar MPPT 450/100, MultiPlus II 48/5000 (internal power measurements) and 2x Pylontech US300C, connected to grid, feed in excess DC PV only, with a limit of 3700 Watts (230V 16 amps). Grid code is Germany (for Switzerland).

I have following wierd bug happening, reproducible:

Starting situation: Battery almost full at 98%, BMS current limit 14 Amps, only one 3400 Watts load on AC out 1 (electric car charger), solar producing 3300-3600 Watts (around 60-65 amps at 52 volts), battery is balancing the remaining or excess part, within BMS current limit.

When the car is charged at 80%, it suddenly interrupts the load.

Then the bug happens:

1. The full current amperage of the Smart Solar MPPT 450/100 drops only slightly to 60 amps, but

2. The MultiPlus II does not feed-in the excess to the grid, but stays at the grid setpoint for a long time, overloading the battery massively with 60 amps while BMS authorizes only 14 Amps with 98% SOC.

It stays so for minutes, before picking up grid feed-in.

This can lead to shortened battery life and could void the Pylontech warranty for those 2 batteries!

I have seen a similar situation when switching on the system: There is a large delay between MPPT 450/100 increasing its amperage and the MultiPlus II feeding-in the excess to the grid. Both are remotely controlled by the Cerbo GX, so not sure if bug is in Venus GX or in MultiPlus.

Interestingly, when I quickly go edit the grid setpoint, the feed-in reacts instantly, within seconds, but the excess PV DC feed-in does not until sometimes a minute of few mintues later. I'm even surprised the Pylontechs don't safety-switch off!

There are no coms errors on the CAN busses for BMS.Can and VE.Can, nothing in the error logs either.

The Cerbo GX firmware and MPPT RS too, the VE.Bus firmware in the MP II is also up-to-date (except for the very latest release of August 1st). Solar MPPT 450/100 has a date-code 2222, from within the range with the extended warranty, but I doubt it is related to that.

This seems to me like a big issue, probably to be taken seriously, I can give VRM access to Victron, please just PM the request.

I searched through the forum but didn't find a similar issue. Thanks in advance for any insight. :-)

charge current limit

2 |3000

Attachments: Up to 8 attachments (including images) can be used with a maximum of 190.8 MiB each and 286.6 MiB total.

Modifications Space

Node Red Space

Answer 1 · 2023-09-05T13:11:19Z

nickdb answered · Sep 05, 2023 at 01:11 PM

It isn't a bug, the battery has to be capable of sinking the "excess" transient power when a large load turns off, while the system balances itself.

These are low frequency inverters so they have a ramp time to adjust to change.

This is why there are a minimum number of battery modules required for an inverter.

In your case you have half the required batteries for a 5k inverter, you should have 4 which would cope with this "excess" power.

https://www.victronenergy.com/live/battery_compatibility:pylontech_phantom

When a system is mismatched it will not perform optimally and, yes, you could damage the batteries.

2

2 |3000

Attachments: Up to 8 attachments (including images) can be used with a maximum of 190.8 MiB each and 286.6 MiB total.

beat commented · Sep 05, 2023 at 02:58 PM

Thank you for your answer! I respectfully only partially agree. The document referred to above is for US2000 batteries. I use US3000C batteries, so is 3 and not 4. ;-) Also the document is specific for off-grid (i've ESS with grid feed-in, which is very different: The grid can absorb excess instead of batteries ;-) ).

You're right, batteries do accept temporarily excess currents for discharging, and also for charging, when not full or almost full. But when already "full", excess charging currents can be damageable.

You're also right, this system is slightly "underpowered" when looking at the "suggested" battery size, and I'll add a 3rd battery to the system tonight when all batteries are fully charged to safely connect them and to redo the experiment (I have them in stock), but I have zero doubt that the result will be similar (way too high charging current for a long time).

Even if I had 3 batteries set up, the BMS maximum charge current would not have been respected.

Even wierder: The same happened once when the batteries were fully charged at 100% and sunshine of 5000 Watts on the MPPT: The system started up with BMS allowing 0 Amps charging current, but the MPPT RS provided over 90 amps, overcharging the batteries for over a minute, until I switched off the MPPT RS to avoid battery damage !

When fully charged, the BMS gives a maximum charge current of 0 Amps, whatever the number of batteries. And this maximum charge current is not respected by my Victron gear for several minutes. That is way beyond "balancing" times and battery tolerance, and represents several percents charge (over 100% battery capacity).

To partly mitigate this situation when batteries are charged at 100%, I have limited the float voltage to 52.4 volts (instead of the BMS-provided 53.2 volts), with a custom setting in the MPPT RS 450/100 using Victron Connect with bluetooth (and keeping "External control" set). That way when batteries are fully charged, only a few amps are going into the battery.

An excess charge current of a few seconds is certainly ok and can be understandable to balance DC loads (even though MultiPlus IIs are able to balance much much quicker (in ms-range) any AC variations when inverting). But a balancing time of a of a few minutes is not ok. It's way easier and quicker measure and to balance DC currents than AC. Proof: when there is sudden additional load on AC out 1 of my MultiPlus, the MPPT RS 450/100 generate additional current within seconds. I don't see any technical reasons beyond a bug that the MPPT RS 450/100 can't spin down as quickly as ramp up (or perhaps the Cerbo GX missing to give quickly a new setpoint to the MPPT, which would be a bug too).

So I still strongly believe that this is a bug in the Cerbo GX or in the MultiPlus II (maybe in the grid code?)

0 ·

nickdb ♦♦ beat commented · Sep 05, 2023 at 03:21 PM

The docs could use an update as they have grown organically over the years.

The US3000C has a DCL of 37A vs 25A on the 2000 model. Two packs is still too small for a 5kVA inverter which can pull well over 100A, not including transients.

3 x US3000C is a minimum.

People forget that it isn't just about discharging, that the batteries must be able to sink transients as well, which can be more problematic, particularly on a transformer-based inverter with slower reaction times.

Surges on a bad grid failure can also push up discharge requirements (not that it should be a problem in your region).

In any case, your system is unfortunately unbalanced.

This is the problem with the 2000's and 3000's and why the 5000C with it's improved performance/capacity ratio is more popular.

Pylon/Victon installs are very popular, there are huge numbers in the field, and it just works.

If there were issues, it would have been found by now, this just seems to be normal behaviour given the use case and installation.

0 ·

Answer 2 · 2023-09-29T10:51:30Z

beat answered · Sep 29, 2023 at 10:51 AM

I still think that this is an algorithm or timing or protocol bug. There is no reasons to not respect a DC charging current by a factor 4 for minutes!

Having read this comment https://community.victronenergy.com/questions/214372/dvcc-and-mppt-450-rs-extcontrol.html?childToView=234093#answer-234093 from @eham this morning, I'm wondering if it is not a bug in the MPPT RS 450/100 CAN-bus protocol ?

2 |3000

Attachments: Up to 8 attachments (including images) can be used with a maximum of 190.8 MiB each and 286.6 MiB total.

question