Just had a look, none of the battery packs show internal resistance.
Thanks for the suggestion!
In my MultiPlus-II the minimum configurable value is 37.2 V, and that is exactly what I currently have set. The inverter does not allow setting anything lower than 37.2 V, so unfortunately I cannot reduce that parameter further.
As for dbus-serialbattery. As far as I know, dbus-serialbattery is not available on stock Cerbo GX firmware. It requires a custom Venus OS build (“Venus OS Large” or community builds), and stock Cerbo GX units do not include it. At the moment I’m running the official firmware, so I can’t switch the MultiPlus to use the BMS voltage/current values through dbus-serialbattery.
This is exactly why the mismatch is confusing:
Even though the Seplos BMS reports a stable ~53 V during the event, the MultiPlus internal measurement briefly drops to ~23.46 V and triggers the shutdown and since dbus-serialbattery isn’t available on the standard Cerbo image, I can’t override that behavior.
If needed, I can switch to a custom Venus OS build later for testing, but for now I’m using the official one so the inverter continues to rely on its own DC sensing.
you always have to install it yourself - but no need for any custom build or OS-Large
I switched frpm a raspberry to a Cerbo GX, dbus-serialbattery was installed in minutes
If you install it, you could select in the VENUS GUI what is the conrolling thing, switch easily back to MP2 or a smartshunt
I’m also considering adding a Victron SmartShunt 500A to my system.
Do you think this could help in my case, or improve stability of the battery voltage readings for the MultiPlus-II?
Would installing a SmartShunt make any practical difference for preventing these unexpected “Low battery / DC low” shutdowns, or will the inverter still rely only on its internal DC measurement?
Any advice from people who have used SmartShunt together with Seplos batteries would be appreciated.
Update with new findings (confirmed with logs and tests):
After extensive testing and diagnostics, I can confirm the following:
-
Battery and BMS are NOT the cause
-
Battery: LiFePO₄ 16S 280Ah with Seplos BMS
-
Integrated via dbus-serialbattery (SerialBattery) on Cerbo GX
-
BMS reports stable voltage (~53–55 V), valid SOC, and:
-
Charge FET = ON
-
Discharge FET = ON
-
No undervoltage, no protection events
-
-
AllowToDischarge = trueduring the event
This fully excludes the battery and BMS logic.
-
-
CAN-BMS conflict excluded
-
VE.Can profile disabled
-
Only one battery service present on D-Bus:
com.victronenergy.battery.ttyUSB0 -
No duplicate SOC/voltage sources
-
-
DC wiring excluded
-
Direct connection to MultiPlus-II
-
2 × 70 mm² per pole
-
No DC fuses, breakers, contactors, or remote voltage sense
-
Very robust DC path
-
-
Key reproducible behaviour
-
Local AC breaker off → inverter transitions correctly
-
Real grid outage (slow / unstable AC collapse) → shutdown
-
VE.Bus System [276] – Low battery
-
VE.Bus System reports DC voltage briefly dropping to ~35 V
-
-
At the same moment:
-
BMS voltage remains stable (~53–55 V)
-
VE.Bus current ≈ 0 A
-
No real battery sag is present
-
-
-
Important log evidence
VRM logs show:VE.Bus System Voltage: 55.1 → ~35.0 → 47.4 → 55.1This happens only during unstable grid loss, not during a clean breaker-off test.
-
Conclusion
-
VE.Bus [276] is triggered by the internal DC voltage measurement of the MultiPlus-II
-
This measurement becomes invalid during unstable AC input collapse, before the inverter fully transitions to inverter mode
-
This is independent of BMS, DVCC, or dbus-serialbattery
-
There is no configurable delay or debounce for VE.Bus [276] in VEConfigure
-
-
Interpretation
This strongly points to a firmware / internal state-transition edge case in MultiPlus-II when AC input decays slowly or irregularly:-
ADC DC voltage briefly reads an invalid low value
-
Low battery protection triggers immediately
-
Inverter shuts down even though the battery is healthy
-
-
Next steps
-
Case has been escalated to Victron Support with VRM logs attached
-
Request is for engineering review and guidance on firmware-level mitigation
(e.g. DC voltage filtering or decision timing during AC collapse)
-
If anyone else experiences shutdowns only during real grid outages but not during breaker-off tests, this may be the same issue.
The thin cables were of course also connected correctly, the photo was taken to show that the lower nuts were securely tightened, the cable did not move, a torque wrench was used
Hello @YuriyK . Did Victron react anyhow to your request?
I am obverving exactly the same behaviour with my MP-II 6k5 GX. When AC goes slowly down, then MP-II shows Low DC Voltage and switched off the AC out for 10 seconds. The only thing is that DC voltage drop in logs does not happen every time. I don’t know if there is a way to setup granilarity of logs from MP to VRM.
Hello @yuras.
Yes, I have already contacted Victron Support. The request has been submitted and acknowledged via the official dealer from whom the MultiPlus-II was purchased. At this stage there is no technical response yet. I assume the delay may be related to the holiday period, so I am currently waiting for feedback from Victron. As soon as I receive any updates, recommendations, or firmware-related information, I will post them in this topic.
Your observations match mine closely, the issue only appears during slow/unstable AC voltage collapse, and the DC voltage drop is not always visible in VRM logs, or this is not available after export from VRM…
Like it or not, you have a battery issue. This is not uncommon behaviour on a grid brown-out.
The low-frequency inverter has to “lean” more on a battery when this happens. This is why the required battery for an off-grid system is much higher than an on-grid one, but people forget that this also applies to poor quality grids.
That Seplos battery, which is unsupported, and there have been numerous topics on Seplos issues over the years, at a max discharge of 200A is way too small for 10KVA inverter, and too small for a smaller system attached to a bad grid.
There have been various related topics on other batteries where marginal sizing and bad grids are involved. This is also why many supported batteries provide clear guidance for sizing.
I would not expect to get much from support beyond a suggestion to add more battery.
Very much appears to be an installation/design issue.
Good luck.
Thanks for your input, I understand this point of view and I have considered it carefully.
However, based on testing and logs, this does not appear to be a battery sizing or discharge capability issue:
-
During the event, battery voltage reported by the Seplos BMS remains stable (~53–55 V).
-
No discharge FET is opened, no undervoltage or protection is triggered on the BMS side.
-
The shutdown happens with near-zero DC current, not under high load.
-
The issue is not reproducible when AC input is cleanly disconnected via breaker — only during slow/unstable grid collapse.
-
VE.Bus logs show a very brief internal DC voltage reading (~35 V) that is not reflected anywhere in BMS or external measurements.
A real battery limitation or undersizing would normally manifest as:
-
sustained voltage sag under load,
-
increasing DC current,
-
or BMS intervention — none of which are present here.
I fully agree that poor grids require more robust systems, but in this case the behaviour points to an internal DC voltage detection / state-transition issue in the inverter during unstable AC loss, rather than an actual battery inability to supply power.
That is why the case was escalated to Victron for review, and I will share their feedback once available.
I question how well the BMS is logging data. I’ve watched circuits with my clamp-on ammeter and sometimes it catches the surge and sometimes it doesn’t.
In this case, I would trust Victron logging over the BMS logging. In my first RV, two batteries, each with their own BMS, is what I went with to ensure that the surge of the air conditioner didn’t overwhelm a single BMS.
Hello Jim. That was one of my initial concerns as well, but the behaviour does not match a battery or BMS limitation. If the issue were related to battery surge capability or BMS current limiting, it should also appear when I disconnect the AC input locally using the breaker. In that scenario, the inverter takes over cleanly every time, even under load. The shutdown only happens when the grid collapses centrally (slow/unstable AC decay). That difference strongly suggests the problem is related to how the MultiPlus handles the AC → inverter transition during an unstable grid condition, rather than a battery or BMS current or voltage limitation. Additionally, during the event the BMS still reports stable voltage and discharge allowed, which further supports that this is not a battery-side issue.
Can you test to enable shutdown on SOC instead of DC voltage? Maybe the V gets ignored and the MP stays on.
In addition of mentioning MP-II 6k5 GX, here the spec of my battery NKON ESS Pro - 51.2V 16.1kWh:
Built using Seplos BMS and JK 2A Active Balancer.
Battery configuration 1P16S
Rated voltage 51.2V
Operating voltage range 43.2V ~ 58.4V
Continuous charge/discharge current 157A / 157A @ 25±2℃
Maximum charge/discharge current 200A / 200A @ 25±2℃
I have tried to measure the battery voltage in the moment when brown-out of AC IN happens, but it is hard to measure it reliably because it does not happen exactly on the schedule and multimeter might not be fast enough to capture it too. I see regular voltages.
Is it possible to setup special device that logs voltage on the battery but sample more often than every second?
Will SmartShunt be able to log it with millisecond granularity?.. Nevermind, it records with 1 Hz frequency.
EDIT1:
To compare, adding how the last brown-out looked like:
And the file with the VRM excerpt:
883078_0_Home_log_20251230-0600_to_20251230-1137.xlsx.zip (17.4 KB)
The fact that is is accompanied by DC ripple means there is insufficient amps from the DC side.
I have seen on a few batteries where the bms is caught “sleeping”.
And we have used a pretty large spectrum of batteroes form different manufacturers over the years.
In absorption state not much is moving in and out of the bank, the brownout happens and it seem like the battery is caught offgaurd. (The biggest offender is the pace bms in my experience)
As an fyi some battey bms measure the pack voltage before rhe disconnect so the reported voltage sometimes isn’t useful for diagnostics. And their interval logging is not sufficient to catch the disconnect. We caught two in diagnostics without our own logging system outside of both Victron and the bms.
I have decided to record with multimeter how brown-out looks like on the battery poles. I caught a drop to 42V for one second. Since it is averaged value, it means that the drop could be even lower than that.
Still from the video:
As you can see V on the battery pack has not moved one bit…
I wonder, if Voltage Regulator [installed between the Grid input and inverter] could cause this somehow? I doubt it because inverter takes current from the VR and I don’t have feed back to grid enabled in the settings of MP2.
I see that VR shutdowns first because it cannot maintain the proper voltage and then inverter goes into alert mode with low voltage on the battery poles.
Not sure what to do. I am sure that producer of the battery will start looking problem elsewhere, but to me it looks like a problem with battery.
Hello. That was also considered, and I agree that some BMS implementations can momentarily “miss” fast transients or appear to lag in their reporting.
However, what makes me question a battery/BMS-side limitation in this case is the following:
why does the behaviour differ so clearly between a manual breaker-off test and a real grid outage?
From the battery’s perspective, there is no difference:
-
same battery
-
same SOC
-
same load
-
same DC cabling
Yet:
-
manual AC breaker-off → inverter takeover is clean and repeatable
-
central grid collapse (slow/unstable AC decay) → Low DC Voltage shutdown
If the root cause were insufficient battery current capability, DC ripple, or a “sleeping” BMS during absorption, I would expect the issue to appear in both scenarios, not only during unstable grid loss.
This strongly suggests that the trigger is related to the AC-side transition timing and internal state changes in the inverter, rather than a battery disconnect or current limitation event.
I fully agree that short BMS disconnects can be difficult to catch in logs, but the repeatable dependency on how AC disappears — rather than on load or battery state — is what makes this case different.
Remove the Voltage regulator as a test
Thats the thing all the conditions have to met. A breaker switching off is a totally different event to the brownout condition of a grid drop.
Where in the sine wave as it is following the sine incoming.
But that does affect how the core of the transformer is saturated (you hear the hum change) and DC insufficiency affects how that happens. There are two things switching power here as there are also mosfets in a battery bms.
But you saw it on the meter.
By the way there is a definite dip to around 40v there… Remember the bms is also switching power… If it is slow to make the surge it is that.
Adding another battery will help.
After working with my distributor and trying all kinds of settings in the inverter, I am thinking that no configuration will solve this. I have investigated forums for similar cases and came to the following conclusion.
In the case of brownout MP2 really pulls the peak power from the battery. There is no way around it. My suspicion is that its capacitors bank is being emptied and then it needs to cold restart it when it is too late. It draws peak power from battery, but no luck there. Why no luck? I have decided to open manual finally and read some specs 
It says that inverter can demand up to 11 kW of peak power. My battery is rated 200A for the discharge peak power. I hardly doubt it that peak power is correct for the battery. It is from China after all 
. I mean, it has a fuse of 250A, but real power it could provide with some error margin: 170-180A.
180A*52V=9540W → I am screwed.
For example, VIctron recommends to install 4 US2000C Pylontech modules for 5000VA model. Those 4 modules could provide up to 19200 W of peak power.
I guess, I will be ordering another bank of batteries soon.
I understand the reasoning and I also considered peak power and capacitor recharge as a possible explanation. However, the same fundamental question remains:
Why does this not happen when the AC input is disconnected manually via the breaker, but only when the grid collapses centrally (brownout)?
From the battery’s point of view:
-
the required peak inverter power is the same,
-
the DC voltage, SOC, and wiring are unchanged,
-
the inverter still has to take over the full load instantly.
If the issue were purely insufficient peak current capability of the battery (or emptying of the inverter’s capacitor bank), I would expect the shutdown to occur in both cases, including clean breaker-off tests. But in practice, manual breaker-off transitions are stable and repeatable.
The fact that the problem is strictly correlated with how the AC disappears (slow / unstable decay vs clean loss) strongly suggests that timing and internal state transitions during brownout play a major role, not only absolute peak power availability.
Also, in the problematic events:
-
VE.Bus current is near zero at the moment the DC voltage suddenly reports ~35 V,
-
BMS still allows discharge,
-
battery voltage remains stable in parallel measurements.
This makes it difficult to attribute the behaviour solely to a battery peak-current limitation.
I agree that under-sized battery banks can absolutely cause issues, but in this case the selective dependency on brownout conditions is what keeps pointing me away from a purely battery-capacity explanation and toward inverter AC-transition handling.