3-phase system: L2 and L3 do not invert when grid is connected (only L1)

[Netrange (Steffen)]

Last idea for today…

Are the current transformers CORRECTLY positioned on the line… Arrows … etc …

[ravo (Radim)]

Dear Steffen,

Nothing has been changed on the CTs — positions and arrows are exactly the same as when the system was working.

Just to be sure, yesterday I tested it by disconnecting the external CTs (so the system measured only the backup/critical-load part), and then I switched back to the external CTs again.
The readings were correct in both cases.

So the CT wiring/position shouldn’t be the reason.

I will update later when I got Mk3 and If I don’t see any “wrong” value in the CT readings or VE.Bus configuration, then my plan is to reconfigure the whole 3-phase system from scratch and set everything up again step by step.

Sometimes a full clean configuration is the only way to rule out hidden inconsistencies in the Assistants or VE.Bus settings.

Thanks again for your time and suggestions.

Radim

[ravo (Radim)]

Dear @Netrange

We have exhausted all options we could think of. Yesterday we completely removed the existing VE.Bus configuration and rebuilt the whole 3-phase system from scratch using Mk3 USB cable. After that, we re-applied all VEConfigure settings and manually added the ESS assistants with the original parameters.

Unfortunately, the behaviour is still exactly the same — nothing has changed. L2 and L3 do not export/invert under ESS control, even though the 3-phase setup and AC measurements are correct.

While checking through the forum, I found two or three other recent threads describing very similar symptoms when the inverter goes into pass-through.

ESS problem with firmware 3.66

Has anyone else noticed ESS problem with firmware 3.66. I have Multiplus 2 GX and it work fine with firmware 3.62. I updated firmware to 3.66 and then ESS not work anymore. DESS works still like before. All settings was fine, but inverter never discharged in ESS mode with this firmware. I installed firmware 3.62 again and everything work well again. Am I only one who have issue like this? What could be reason about that?

Multiplus II Low Battery and missing Dynamic Cut-Off page in assistant

Hello, I have a 3 phase ess system with one multiplus II 48/5000 GX and two multiplus II 48/5000, cerbo gx, Pylontech battery. Lately I got a low battery (Multiplus switches to pass-thru) warning on the multiplus but not on the cerbo, SoC is the same as on the cerbo (e.g. 85%). When I cut mains multiplus switches to discharging. When I restart the multiplus everything is working fine again. After some time the issue occurs. Yesterday it did not work, no power, I think it thinks it is completel…

This might indicate that the issue could be related to the latest 3.66 firmware, but I cannot confirm it. At this point, all hardware, wiring and measurements seem correct, and only the ESS logic is refusing to activate discharge on L2/L3. If needed, I can provide full logs (VRM diagnostics, gx-device logs, VE.Bus status, Modbus register snapshots, etc.).

Thank you in advance for any additional suggestions on what else we could try.

[Netrange (Steffen)]

… slipped way down

It’s a shame that the rebuild wasn’t successful. Strangely enough, many people in the forum, including in the German section, have a 3-phase system and it is running.
The only difference could really be the use of CT sensors. Many use an external meter… if it really is due to the firmware in connection with the CT sensors, the only solution would be to downgrade the firmware… You can find the old FW on the Victron Professional website.

[ravo (Radim)]

Dear Steffen,

Yes, I also noticed that most 3-phase installations in the forum run without issues, which is why this case is so unusual. There are, however, a few similar reports — maybe users have not updated to the latest firmware yet, or they simply did not report it. After we rebuilt the entire 3-phase system from scratch, re-applied all VEConfigure settings and manually added the ESS assistants, the behaviour is still exactly the same.

Regarding the CT sensors:
we also repeated the test with internal metering only, but there was no change — L2 and L3 still do not export/invert under ESS. I am also considering testing with an external meter as suggested in the documentation, but it is strange that identical configurations elsewhere work perfectly with CT clamps. This increasingly points to some behavioural change in the recent firmware.

One more thing I would like to add, because it might be relevant:

According to the official Victron documentation:
https://www.victronenergy.com/media/pg/Energy_Storage_System/en/multiphase-regulation—further-information.html

If I understood it correctly, in section 7.2 “Individual phase” it states:

“With ‘Individual phase’ option selected, the (single-phase) ESS uses the battery to balance only L1 to 0 W. L2 and L3 are visible on the CCGX, but not used by the ESS in any way.”

We discussed this internally, but in reality most 3-phase ESS systems on this forum do compensate L2 and L3 down to 0 W even with “Individual phase” enabled — including our other installations.

So the behaviour we are seeing here is inconsistent with both:

• the documentation, and

• other real-world 3-phase systems using CT sensors.

Of course, I also tested switching to 7.3 Three-phase ESS, but the result was the same — still only L1 is inverting. Unfortunately, we cannot use Three-phase ESS permanently because our grid operator bills per phase rather than by total export.

This makes the issue even more confusing, because under the same configuration other sites still export/invert correctly on all three phases — only this installation refuses to discharge on L2/L3.

Tomorrow we plan another full restart and completely wipe the VE.Bus configuration, and I am thinking to also try physically swapping the L1 and L2 inverters to see whether the behaviour follows the phase or the device.

Regards
Radim

[ravo (Radim)]

@Netrange

The only difference could really be the use of CT sensors. Many use an external meter… if it really is due to the firmware in connection with the CT sensors, the only solution would be to downgrade the firmware… You can find the old FW on the Victron Professional website.

I also tried to find older firmware, for example version 502. However, on Victron Professional I can only find the latest 558 release and the older 459. (files 2629558.vff, 2629459.vff). I tested 558 today as well, but the situation still persists.

Small update — I managed to find the 502 firmware in the public archive here:
https://updates.victronenergy.com/feeds/VEBus/firmware-archive/
File 2629502.vff is available.

When I get back home, I will try the downgrade and report the results here.

[Netrange (Steffen)]

Great… thanks for the info… it’s always nice to get feedback. That helps other users too… even if things don’t always work out… that’s part of it.
We all learn from the experiences of others… positive or negative.

I was also referring to the GX when I mentioned the downgrade…

[ravo (Radim)]

Thanks Steffen,

just to summarise our progress so far:

we already fully reset the entire 3-phase VE.Bus system configuration and rebuilt it from scratch, and today we downgraded all inverters to firmware 502. Some time ago we also tried running the Cerbo GX on version 3.51, but the behaviour was exactly the same. At this point I feel we have exhausted every technical possibility of what could be wrong on the configuration side.

I also considered completely wiping the Cerbo GX and installing a fresh Venus OS image, but that would most likely mean losing all history and dashboard data, so I will leave this as a last resort.

As the next step we plan to physically swap the L1 and L2 inverters. If the unit currently on L2 starts working normally when placed in the L1 (master) position, then we can rule out any hardware issue.

[ravo (Radim)]

Small update regarding our troubleshooting:

We performed a complete physical swap of the L1 and L2 inverters and tested again. The behaviour is still exactly the same, so I can now fully exclude any hardware-related issue with the units themselves**.**

1. Cerbo GX reset to factory defaults
   We also tried restarting the Cerbo GX and performing a full reset according to the official procedure:
   https://www.victronenergy.com/media/pg/Cerbo_GX/en/reset-to-factory-defaults-and-venus-os-reinstall.html
   This included wiping the configuration and reinitialising the device, but it did not change the behaviour either.

2. Swapping L1 and L2
   After swapping the units, the system continues to invert only on L1. When the former L2 inverter was placed in the L1 position (master), it behaved exactly like the previous master. Therefore, the inverters are OK and hardware is now fully ruled out.

3. Testing older assistants from another identical installation
   We even loaded the exact same (older) ESS assistant configuration that we use on another site with identical hardware, where L2 and L3 invert/export normally. Unfortunately, this made no difference — the issue remains unchanged.

4. Modbus register comparison
   I repeatedly compared all relevant Modbus registers between both installations, and there is no mismatch or invalid value that could explain the behaviour. Everything appears consistent across both systems.

Additional finding – DESS settings cannot be changed

During this process I also discovered another serious issue:
After resetting the Cerbo GX, I was unable to re-enable the DESS system. One record showed a battery capacity of 0 kWh, so I attempted to correct it to 21.6 kWh — but DESS parameters cannot be changed at all.

The system shows the green confirmation message (“Setting accepted”), yet none of the changes are applied.

It appears I am not the only one. Multiple users report that DESS settings simply cannot be modified and always revert back, even though the GX device confirms the request:

• https://community.victronenergy.com/t/dess-settings-cannot-be-adjusted/46378/4

• https://community.victronenergy.com/t/dynamic-ess-setting-changes-fails/24419

• https://community.victronenergy.com/t/unable-to-change-dynamic-ess-settings/46522/9

At this point we have exhausted almost every technical possibility on the configuration and hardware side.
Since the system behaves exactly the same after all resets, firmware downgrades, inverter swaps and complete reconfiguration, it increasingly looks like there may be some hidden firmware-related issue involved.

[ravo (Radim)]

Final update

We tested both the firmware path and the GX hardware path to fully rule out a Cerbo-related issue.

1. Downgrade test (3.67 → 3.62)
   The downgrade did not complete successfully.
   After the firmware was written and the unit rebooted, it entered a power-issue fault state.
   This matches the known hardware advisory for older Cerbo GX units series HQ2201xxxx–HQ2220xxxx where, on 48 V systems, inrush current can cause the GX to fail to boot or become unstable.
   Official reference:
   Cerbo GX power supply issue in 48 V systems — Technical Note August 2022
   PDF link: https://www.bimblesolar.com/docs/2022-08%20Cerbo%20GX%20power%20supply%20issue%20in%2048V%20systems%20-%20v5.pdf

Therefore, we could not proceed with any 3.62 diagnostics on the original Cerbo GX – the unit failed during the downgrade process and is no longer operational. The device will now be sent in for warranty inspection and replacement processing.

Therefore, we could not proceed with 3.62 diagnostics on original Cerbo GX - it is dead now.

2. Full hardware swap – Cerbo GX → Cerbo GX Mk2
   We replaced the entire GX device with a new Cerbo GX Mk2.

• FW 3.54 → same behaviour

• Upgraded to 3.67 → still identical
  The issue is not caused by the Cerbo GX hardware nor firmware.

3. Phase-swap test (L1 ↔ L2)
   After physically swapping L1 and L2 between the inverters, the problem moves with the phase role, not with the hardware unit.

• The new L1 becomes the non-inverting phase

• The former L1 (now on L2) behaves normally when becoming Master

This strongly indicates the root cause is inside the VE.Bus cluster state / role configuration, rather than the Cerbo or individual inverters.

At this point we have exhausted all reasonable tests on both GX hardware and firmware, including full device replacement and phase-role swapping. We are still at the same result, and we currently have no further direction to move forward. Any additional guidance from Victron engineers or the community would be highly appreciated.

Regards
Radim