Problem with discharge cut off value

[Ern5t]

A few Marstek software updates ago (around V146) they added a threshold value for the discharge cut off. I.E. if set at 12% the discharge won't start if the charge level is 14%. In self consumption mode, this threshold is overseen. If the battery with 14% is selected with a for example 100W power shortage, it will rise the discharge power up until 2500W before it adds the next battery even though there is no power coming from the first battery. If the second battery adds the 100W, the P-line stays at 2600W.
IMO it would be simple to check whether the selected battery is delivering and if not, abort the discharge, only I don't have the skills.....

[gitcodebob]

Hi Ern5t,

Interesting. Checking if the battery actually delivers is possible, but it indicates a fault state.
Most strategies won't work correctly if they hear afterwards that the control output was not executed. It causes instability if we would operate it that way.

I think the best way forward is to set the Cut Off and Power limits to match the ones in the firmware.

Apparently Marstek made changes to it.

[Ern5t]

Hello Bob,

Setting the limits at a higher value won’t work as the threshold will rise as well. It was better described on Tweakers by phkba.

==
Tweakers message=84618606 phkba wrote on Wednesday February 11th 2026 @ 09:54 AM:
[...]
From the release notes of v147:

“A 5% load hysteresis has been added. After the discharge activates the SOC ≤ DOD, the charge must reach the SOC ≥ (DOD + 5) before the discharge to the grid can resume.”

So the battery must first be charged to at least 17% before it discharges again.
—

I can confirm that it operates in this way. As long as the discharging is in progress it can reach the set minimum value of 12%, but it won’t start discharging below 17%.
I hope you can find a solution without instability

[gitcodebob]

Oh boy, well I better start updating my firmware to investigate this. Thanks for the headsup.

[gitcodebob]

Ah I found a facebook post about this as well: https://www.facebook.com/groups/9076528425802065/posts/25779982661696712/

In Dutch:

V3 update V147

Geachte heer/mevrouw,

1. De LED-indicator voor de schakelaar is nu beschikbaar.
2. De configuratie voor servers van derden is geoptimaliseerd.
3. De rapportage van gegevens buiten de piekuren aan servers is verbeterd.
4. De AI-gerelateerde protocollen zijn verfijnd en problemen met abnormale AI-werking zijn opgelost.
5. Compatibiliteit met Duitse Stromleser-elektriciteitsmeters is toegevoegd.
6. Er is een laadhysteresis van 5% toegevoegd. Nadat de ontlading de SOC ≤ DOD activeert, moet het opladen de SOC ≥ (DOD + 5) bereiken voordat de ontlading naar het net kan worden hervat.
7. De off-grid-poort omzeilt de DOD-beperkingen tijdens het testen, waardoor ontlading tot minimaal 8% mogelijk is.
8. Een bug is verholpen waarbij meerdere Venus-units in de handmatige modus elkaar zouden opladen/ontladen tijdens zelfopwekking/zelfverbruik.
   Met vriendelijke groet, Keay

[gitcodebob]

I have not had time to solve this yet. A complication is the growing multitude of invisible variations. Marstek type v1/v2/v3 + BMS versions + integration of battery to HA. This makes for an unclear issue resolution.