Coulomb

I believe it is related.
I have seen this exact problem and it was caused by the card the Wifi dongle plugs into. You can test this by unplugging the cable that feeds this internal card. Does the inverter start normally?
I have repaired two of these cards. One had a shorted capacitor and the other a faulty MAX232. Simple fixes if you know someone cable of working on SMD components.

Feedback here, I have removed the RS232 card and the inverter has been working as should ever since.
I use Home Assistant to monitor the inverter, so the WiFi card is not needed. Thank you for the suggestion.

I'd say that this is your main problem. If you can't find a setting that keeps communications up to the BMS, then resort to setting USEr. That means that the BMS won't be able to intervene to save a weak cell that goes over- or under-voltage, but at least it should work.
Yes, definitely change this setting higher. I suggest 150A, for C/2. Your battery should be able to sustain this continuously, though perhaps find the actual recommendation from the battery documentation.
If none of this helps with the PV power, then unless your inverter is made later than December 2025, I'd say give firmware 75.85 or one of the x75.85 patched firmwares a try. Unfortunately, you will have no way back to 75.08, or any other suitable 75.xx firmware, unless one happens to turn up in future.

Usually there is a small sticker over one of the screws holding the front cover on. It will have a year and one month blacked out. For example, the one below has a manufacture date some time in May 2018:

That depends. If your manufacture date is considerably later (say 6 months or more) than the firmware, then you might decide it's not worth regressing that far.
Alas, that information has not been published for many years, ever since the appearance of the clones. The best you can hope for is a vague clue in the file name of the firmware update file, e.g. "fixes fan speed at low power". But we don't have a firmware update file for 75.08 (except for one that is for an Axpert King rack mounted; a totally different firmware with confusingly the same version number).

Unfortunately Max II 11kW Twin firmwares are rare; I only have 75.85 for the '28066 processor.
So your only option is x75.85 patched firmware, that works around the "stuck at 90V" issue. It includes factory firmware version 75.85 as well. Unfortunately I have no way of knowing how old factory firmware version 75.08 is; 75.85 is from June 2025. You can compare this with the manufacture date of your inverter, to decide if it's worth updating with this firmware.

Having sorted out the mess that is the 43.xx firmware number range, I came across three newer (to me) 43.xx firmwares. The latest of these is 43.29, two days later than 43.28. Perhaps they noticed a whoopsie with the charging and updated the firmware quickly as a result.
This is main (U1) factory firmware version 43.29 for the Axpert VM II Premium 3kW and 2.5kW '28062 non-Twin. Do not use with any other model or version.
Use at your own risk. The hex file is dated 16/1/2025.
Firmware upload instructions for models with a removable display.
This firmware processes the QDOP command, so with suitably recent display firmware, it should support SoC% for settings 12,13, and 29 when connected to a suitable battery's BMS.
The rar file password is "flash" without the quotes.
VMII_Premium_2K5_3K_MAIN_Reflash_Ver43.29 pw flash.rar

Possible clue regarding Error 80 (CAN communication failed) on parallel Voltronic inverters
Hello everyone,
I would like to share my experience regarding the infamous Error 80 (CAN communication failed) on two Voltronic inverters operating in single-phase parallel mode.
For several weeks I experienced random Error 80 faults. Sometimes the system would run for more than 30 hours without any issue, while other times it would fail after only one hour. The failures occurred under very different load conditions (from almost no load to more than 6 kW), so I could not correlate the problem with output power.
To investigate the issue, I developed a crash recorder that continuously stores the previous minute of operating data and automatically saves it whenever the AC output is lost. I analyzed several crash files.
The most common sequence was:
both inverters operating normally in Battery mode (QMOD = B),
one inverter suddenly switching to Standby (QMOD = S),
the second inverter remaining in Battery mode and supplying the load alone,
a few seconds later both inverters entering Fault mode (QMOD = F) with Error 80 (CAN communication failed).
Interestingly, the crashes often occurred when the two PV arrays were producing very different power levels.
My installation has two independent PV arrays:
same number of panels,
but different physical locations,
one array becomes partially shaded before the other because of a nearby building.
As a result, during the morning and evening transitions, one inverter can produce significantly less PV power than the other.
Before changing any inverter settings, I also implemented all the recommendations suggested by Coulomb regarding the inverter configuration. These changes did not eliminate Error 80, but they clearly improved the overall quality of the installation.
For example, before applying those recommendations, I had an audible high-frequency buzz in my computer speakers whenever the inverter was operating. After applying the recommended settings, that noise disappeared completely, suggesting that the inverter output waveform became cleaner, with fewer harmonics or less electrical interference. Although this did not solve Error 80, it definitely improved the electrical behaviour of the system.
While reviewing the documentation, I noticed something that caught my attention.
Both inverters were configured with:
Program 28 = PAL
Program 30 = ONE
The official parallel installation guide recommends Program 30 = ALL for parallel operation.
I therefore changed only one parameter:
Program 30: ONE → ALL
Nothing else was modified:
same firmware,
same batteries,
same CAN cables,
same parallel cards,
same supervisory software.
Since making this change, the system has now completed two full day/night cycles without a single Error 80.
These cycles included:
sunrise transition,
battery charging,
full PV production,
heavy loads (air conditioning and an approximately 2 kW angle grinder),
afternoon shading of one PV array,
sunset transition,
overnight operation.
Previously, these transition periods were exactly when Error 80 occurred most frequently.
At this point I cannot claim that Program 30 is the definitive solution. Two successful cycles are obviously not enough to prove causality.
However, the improvement is significant enough that I believe this parameter deserves attention, especially on installations where the two PV arrays do not receive identical solar irradiation throughout the day.
Has anyone else tested Program 30 = ALL instead of ONE on a parallel installation?
I would be very interested to know whether other users have observed a similar improvement.

He good news is @TaliaB has been discharge and he is busy recharging hopefully to 100% SOC on a family farm.
His oxygen levels still need to recover and could take a while.
Thanks for the support on behalf of our friend.

I have found enough time to patch 78.07. Sorry for the long wait.
This is patched firmware version 278.07, based on factory firmware version 78.07 for the Axpert Max E 11 kW Twin '28066. Do not use with any other model, not even an Axpert Max E 11 kW Twin with '2809 DSP (it will refuse to update the firmware).
This patch implements Georg594's patches for firmware version 90.06, but applied to 78.07. It also fixes the premature float bug, and changes the version number reported by monitoring software to 278.07. The display will still show 78.07. The current threshold is changed from 0.50 A to 0.05 A, and the minimum MPPT voltage is changed to 200 V.
Use at your own risk; I do not have a Max to test this firmware on.
Firmware upload instructions for models with a removable display.
dsp_278.07 patched 200V.zip

It won't update, just as your mismatched 8 kW didn't update.
It's sad that some resellers don't know about the different DSPs.
However, the difference between '28066 and '2809 firmware is just one start-up call to a function in ROM (Read Only Memory, different from Flash memory). So I might be able to modify this '2809 firmware to become '28066 firmware, then package it up with a different reflash tool and it should work. I did that once before; I can't recall if I got any feedback on whether it worked or not. I need to do some checking.

Here is my comparison between 978.11 (still stuck at 90V) and 378.11
Before:

After:
Similiar full sunny morning, same time, same configuration. Change was only FW and looks much better

Thanks for the advice.
I upgraded from factory firmware 82.07 to 82.63 about a week ago. So far, everything seems to be working well. The inverter has been stable, and I haven't noticed any problems. I'll keep monitoring it, but at this point the upgrade appears to be compatible with my inverter.

<font style="vertical-align: inherit;" dir="auto"><font style="vertical-align: inherit;" dir="auto">Panel remoto_Reflash_MCU(03.25).rar</font></font><font style="vertical-align: inherit;" dir="auto"><font style="vertical-align: inherit;" dir="auto">Panel remoto_Reflash_MCU(03.25).rar</font></font>

Update (SOLVED): I have success with 978.11 firmware update. What i did extra, hard reset of inverter (battery off, AC off, PV off) and waiting few seconds.
After starting inverter with battery and AC, communication was established correctly.

Hi
On FTP side there is DSP firmware too. Togather with Display 57.22.I think they are stable.
Here I upload 60.14 DSP firmware.
VM TWIN_60.14.7z

The display / BMS comms is likely the firmware that needs to be changed, however.
The latest (of two) 57.xx display firmwares I have is 57.22. [ Edit: Oops, I now see that's just above in this thread. I just search once I know the firmware version. ]

Ah, you actually have 60.06. It seems that the latest 60.xx firmware is 60.10 (also available as x60.xx patched). But if you need the quiet fans, stick to 60.95, or its patched versions x60.96 (note the difference in the last digit; I erred with x60.95 and the fixed code is now called x60.96, even though it's still based on factory firmware version 60.95).

Dear all

After studying the service manual, the internal board layout, the Voltronic SOP, and several teardown pictures, we have developed a working hypothesis regarding the intermittent Error 80 observed on some parallel Axpert/Voltronic 5048MG units.
The original Voltronic SOP only addresses LCD flickering by repositioning the MPPT choke wires. However, after a closer analysis of the inverter architecture, we believe the root cause may be broader than the LCD issue.
The MPPT choke is physically mounted far away from the MPPT power board. The two wires connecting the choke to the MPPT board (P2/P3) are relatively long and appear to be part of the MPPT switching current loop. These wires carry high current and high dV/dt switching waveforms generated by the MPPT IGBTs/MOSFETs operating at approximately 50 kHz.
While the toroidal choke itself should radiate very little magnetic field, the long connecting wires form a large switching loop that is likely to generate both conducted and radiated EMI. Inspection of the inverter shows that these wires run very close to several ribbon cables and communication boards, including boards associated with the parallel system (PAR-P board, PAR communication board, communication board and main board).
Our hypothesis is that the switching loop radiates broadband EMI which couples into the nearby ribbon cables and communication circuits. The fundamental switching frequency is only around 50 kHz, but the fast switching edges generate harmonics extending into the MHz range. This may occasionally corrupt communication between the parallel control boards, resulting in Error 80.
To investigate this theory, we will first implement the official Voltronic capacitor modification (2.2 µF film capacitor across PV+ and PV−), which should improve differential-mode filtering on the PV input.
If the problem persists, we plan to shield the two MPPT choke wires. The proposed design is intended to minimize additional capacitance:
each MPPT wire will be routed inside its own PVC insulating tube;
the two tubes will be placed side by side;
a copper braided shield will surround both tubes;
an outer heat-shrink sleeve will provide mechanical protection;
the shield will be connected to chassis earth at one end only;
significant air spacing will remain between the conductors and the shield in order to keep conductor-to-shield capacitance as low as possible (estimated only a few tens of pF).
The objective is not to filter the MPPT current itself, but to contain the radiated electric field generated by the switching loop and prevent coupling into the nearby communication and parallel-control circuitry.
At this stage this remains a hypothesis, but it is consistent with:
the physical layout of the inverter;
the location of the MPPT choke wiring;
the proximity of the communication ribbon cables;
the Voltronic LCD flicker correction;
and the intermittent nature of Error 80.

Update: I started replacing all of my current sharing cables and in the process I identified the problem. One cable had a ferrule that wasn't thin enough and had poor contact there. I assume the ferrule was placed there to improve contact, but that ended up being the reason for the error 72. By crimping with a different ferrule the error goes away and hopefully forever. Thanks a lot! @Coulomb Enjoy your coffee!

Just an update just received. He has recovered in such a way that he is out of ICU and in the high care ward. He looks better and can eat without help. Still awaiting results of the latest blood tests. Great to see he is on the recovery path. 🙏

The latest is they intent fitting a filter in the largest vein in the body to catch the blood clots before they get to the lungs. This should happen tonight. This is after consulting a specialist in PTA.
Will keep you guys informed. 🙏🙏

To save readers' time, the two files in the Eissa's post before last are for Ultra main firmware version 25.10 (left/first), and Ultra display version 12.18 (right/second). Don't confuse 12.1x for Ultra with 12.2x and 12.3x (etc?) for Max and other models.

SV-IV-56.14_last_from_easun_banban.rar

SV-IV-56.13.rar

Please post in English here.
Most factory firmwares leave the PV voltage too low even when the PV current increases enough to allow a higher PV voltage, and hence higher PV power. So the PV voltage is "stuck" at the minimum MPPT voltage, which for Axpert Max models is 90V. Other models can be stuck as low as 60V, some at 120V.
The result is lower than normal PV production. Most days, this reduces production by up to 50% or more for the first and last hours of sunlight. On cloudy days when you often need all the PV power possible, it can be reduced almost the entire day. Other days it might make little difference.
To test if it's happening, isolate the panels momentarily. Usually, PV production will then return to where it should be, with the PV voltage near the Maximum Power Point.
User Georg594 figured out patches for an older 8kW firmware, then "ported" the patches to a few other models. I've been porting these patches to several other models since.
Part of the patching involves choosing a new minimum PV voltage. That means that there isn't a "one size fits all" patched firmware for any particular model. So each patched firmware file now comes out as a set of patched firmwares.
A few factory firmwares have what I've been calling the "presumed good" MPPT algorithm. This seems to work a lot better than standard factory firmwares, but for some reason these don't seem to last and are not common. The patched firmwares seem to perform slightly better than these presumed good factory firmwares anyway. But this means that every time a new firmware is released by Voltronic, users have to choose between updating and losing the MPPT performance, or waiting until I or someone else gets around to patching the new firmware. Each patched firmware set takes me about 2 hours of work.