Overvoltage spike and then OFF problem with an MP2-GX-3K

DIY
1

Hello,

A system consisting of:

  • MultiPlus-II 3000 GX
  • V-TAC battery 5kW

The system is used at the moment as test system / ups system.
No assistants installed.
Is configured to disconnect the grid after absorption finished and reconnect the grid when the SOC is 20%.
For that I am using the virtual switch function, as “dedicated ignore AC input”.
Charging bulk and float configured as 56.4V, 3.525V/cell.
DVCC disabled, no BMS control on DVCC.
Plain and simple as possible.

As seen in the pictures below, it connects the grid, it bulk charges and then enters absorption phase (which is 1 hour - minimum possible).
During absorption, at some point, before 1 hour absorption end - at about 45 minutes - the BMS says that it’s done and sends CCL = 0A.
In that moment the MP2 voltage spikes to 63V and switches OFF !!!

You will say - hey, the power doesn’t have where to go and therefore the spike…
Just take a look in the picture at the currents (MP2 and battery) in that moment… 0 (zero) Amps !!!
So the spike is NOT justified.
More than that, the OFF state is not justified !
No errors logged.

You will say: why not using DVCC?
Because all parameters can be set from the charger - current limit (18A), voltages (56.4V) and there is no need.
And still, I’ve also tried with DVCC enabled. The same !!

What happens???

It happened multiple times, so it’s repeatable…

If I try to turn it on from Remote Console, it spikes again and cycles: bulk → absorption → float → sustain (!!!) → off quickly.
Strange, huh ?!..

1
11612×911 150 KB

2
21608×913 167 KB

3
3510×308 20.3 KB

4
4636×503 68.9 KB

5
5636×503 68.8 KB

3

Like I’ve said in the post, at 0 (zero) amps consumption, the situation is almost balanced so a jump so big is at least strange…
And again, only the charge fet is disconnecting. The discharge fet remains on, so the battery is not disconnected!

And for sure doesn’t warrant the switch to OFF of the inverter.
In other words, possible critical loads will be turned off.
It is not normal!

4

Hmmm

So it is or isn’t controling the system?

Am i understanding correctly here?
It is definitely happening at full, after absorption. And then it is supposed to disconnect from grid and invert, but is failing the switch over.
In rush problems…?
(Or the battery doesn’t like the sudden change in state its going from being idle to suddenly having to discharge)

6

@lxonline
Yes, the DVCC is off. In a way like a system without Cerbo. You can have that, right?
Only a MP2 and a battery. And all the settings are in the MP2.
The same in this case. I am using the GX card inside just for the MP2 to be aware of the SOC, because I didn’t activate the MP2’s battery monitor. It’s more accurate this way.
So, again, for a system as simple as that, without solar and co, there is no need for DVCC.
And it’s not after absorption, is during absorption, when the CCL becomes 0.

@Ludo
No, setting voltage 0.2V lower will not solve the problem. It will only happen later or sooner, depending on the final voltage.
For example V-TAC is saying in the manual that the float voltage is 56.8V. Setting that will move the time when this happens only after 15 minutes in that one hour absorption. So following the manual is the same and sooner.
So, by saying that “its not the multiplus”, you are suggesting that it’s normal for the MP2 to turn OFF in a situation like this? :wink:

In fact this is what bother me the most: the fact that the MP2 is turning OFF, which I believe it’s not normal at all !
Nowhere in Victron documentation is mentioned that the MP2 could turn off when the battery sends CCL=0 !!!
And this is a dangerous situation, because someone could configure a system where loosing AC could become life threatening and after that take action against Victron for such thing…

7

Checking that it is not plugged in at all here as well. Removing bms control sometimes with some batteries somehow there seems to be some interference anyway.

I agree it is not normal, but since it takes two in partnership to invert - it is difficult to pinpoint who isn’t doing their job or maybe it is both.

Soc is not necessary if the inverter is using absorption voltage target and time to determine when to disconnect from grid (which VS uses).

So not when it has determined to disconnect from Grid?

That is because it doesn’t cause a shut down. I have some ups type systems that sit with CCL at 0. (DCL at 0A does cause shut down.)

9

Do you know how charging fet and discharging fet are working?

10

If you read carefully you can see that I’ve said that.
1 hour absorption and happens after 45 minutes since absorption started.
So during absorption, as seen from the graphs too… :slight_smile:

11

Wondering if the inverter can tell the time well enough…

Curious to know if it will do it with zero comms. Maybe put the battery onto a different GX if you want to monitor it and test again?

The voltage spike could be caps discharge in the power supply.

12

Yes, it can. It’s not that, because if I increase, like I’ve said above the final target voltage to 56.8V, like the battery docs say, it happens after 15 minutes.
I don’t even want to think how will behave if the target voltage will be 57.6V (3.6V/cell) how it’s CVL communicated by the battery… :slight_smile:

A capacitor can’t discharge with a higher voltage that the one it’s charged to… :slight_smile:
This is a voltage regulation inside MP2.

And if you carefully look a the graph, even the charging current its a little bit strange…
It’s set at 18A and the battery is charging with currents between 15A and 19A !! …
So a pretty strange regulation…

13

Then it does make sense what ludo is saying, that the battery is somehow disconnecting as its protections are being triggered.

Is it sending it to the right place? Maybe it is sending CCL to DCL address? :wink:

14

Yes, the places are correct. And remember, DVCC is disabled so those values are just a tell-tale.

15

This could be interesting… Eventually use inverter’s SOC monitor…
But I will loose the battery as device on the VRM and Remote console… crappy… :roll_eyes:

17

Yeah. I have a second GX as one of my tools for this kind of diagnostics. Still need to monitor the battery but run the system with only the electrical connections.

Is maybe the dcl and ccl dropping at the same time? I don’t see DCL on any of the graphs.

18

@Ludo
I will. Thanks!

@lxonline
DCL remains at 100A all the time.

1 Like
19

If you get one cell going over volt, and the current is not curtailed by DVCC, then the BMS charge fet will switch off, and you get the volt spike with no amps.Best to always use DVCC.

20

Thanks for the input.

The battery is quite new and the BMS is active type.
Never saw a difference more than 10mV between cells, on the entire range of voltages, from 51.2 to 56.4.
When this is happening, the current is less then 50mA, because the BMS finished final balancing and considers that there is no need for more and sends CCL=0. This is why it’s so strange, a spike so high, on a so small current…

And again, the reason I am so pissed off is the fact that the MP2 is turning OFF, without warning and without error !..

21

If the battery suddenly blocks charging that causes a voltage spike and that makes the MultiPlus crash.
But “normally” that only should happen if the MultiPlus is charging with max. current at that moment.
But in you case the current is already 0A, that’s strange.

22

Thanks for input.

Exactly my thoughts!
I would not mind if that would happen on full charge current or even half.
But for that to happen when the consumption is less than 50mA it’s strange!

1
1397×354 33.3 KB

23

Found the problem, as I’ve had this morning more time to investigate…

There are, for sure, cases when the battery reports CCL=0. Many BMSes do that.
In my case, the V-TAC is using a BMS that it’s used in LG battery types.

Victron, instead of not allowing in the first place for the voltage spike to occur, applying, for example, a simple single-pole recursive low-pass filter for the voltage regulation, they choose to shutdown the system !! :zany_face: :roll_eyes:

They even made a specific script for that, that memorize at a second time interval the MP2 voltage and store it in a 40 seconds buffer.
When any of the values are below or above a certain level (choosen 10%), they shutdown the vebus device, in my case the MP2.

Here is the relevant code in their scripts, where is clearly seen on the last line that the /Mode is set to 4, which is OFF.

Wonder why is that?
Is it not a dangerous situation to shutdown the system like this and all to remain in blackout?
Why only on LG type batteries?

Imagine, the code is executed only if the current is below 0.01A, 10mA !!! Seen in the code below !
Why don’t they regulate the voltage better ??? Especially at so low currents !!!..
Because min and max vebus voltages, meaning MP2 voltages, stored on that buffer, are compared against the battery voltage, which seems to be rock solid if it’s taken as reference for comparison…

Please, Victron developers, could you shed some light on this?
Especially how such a big jump in voltage can happen on such LOW currents (<10mA) for it to warrant such script?
Thank you in advance!

if battery_current is None or abs(battery_current) > 0.01:
	logging.debug('LG voltage buffer reset')
	self._lg_voltage_buffer = []
	return
vebus_voltage = self._dbusmonitor.get_value(vebus_path, '/Dc/0/Voltage')
self._lg_voltage_buffer.append(float(vebus_voltage))
min_voltage = min(self._lg_voltage_buffer)
max_voltage = max(self._lg_voltage_buffer)
battery_voltage = self._dbusmonitor.get_value(self._lg_battery, '/Dc/0/Voltage')
if min_voltage < 0.9 * battery_voltage or max_voltage > 1.1 * battery_voltage:
	logging.error('LG shutdown detected V=%s I=%s %s' % (battery_voltage, battery_current, self._lg_voltage_buffer))
	self._dbusservice['/Dc/Battery/Alarms/CircuitBreakerTripped'] = 2
	self._dbusmonitor.set_value(vebus_path, '/Mode', 4)
24

Maybe because they burn your house down. In my country a big recall is in place.
And the Victron code some kind of safety response?
Just guessing, a path you could follow up as you choose.