Thank you in advance for any responses that might help me.
Here are the device versions:
Cerbo GX V3.71
Multiplus v558
Jk-bms V19.04
My DIY battery pack with MB31 cells balances perfectly at 55.2 V. After an hour or half an hour, I’d like it to switch to “float” mode at 53.6 V. That’s what I’ve configured, so the BMS reduces the MPPT to 10 W, which I understand (I’m not sending energy to the grid). But once it reaches 53.6 V, the battery pack continues to discharge due to household loads, and for some reason, the MPPTs no longer increase in watts, even when the sun is shining. As a result, it does not maintain the charge at 53.6 V and does not switch to Float mode; it remains in Absorption mode.
So, for your information, I’m using an external control on the Multiplus and the MPPTs because I configured the ESS via VE Config.
I’ve correctly configured the MPPTs to the appropriate voltage: 55.2 V in Absorption mode and 53.6 V in Float mode.
I plan to update the JKBMS to version 19.13.
I’m waiting for my RS485 adapter.
I’ve seen posts about a similar issue with the DESS. I’ve already tried disabling the DESS, but I still have the same problem.
For now, I’m disabling Float mode in the JKBMS app.
Here is the JKBMS configuration
Don’t do that. floating LFP cells is just EXACTLY resulting in what you observe. LFP Cells do not drop in voltage if the charging is stopped. By lowering the chargevoltage, you are actively pulling the desired voltage down, which results in a discharging.
Floating is required for batteries that needs to be charged ABOVE their maximum voltage to reach 100% and therefore need to be floated to prevent damage when reaching 100%. LFP should be charged AT their maximum voltage, so no need to reduce that voltage, when they reached 100%.
What voltage does your PV Array have? MPPTs need battery_voltage +5V (iirc) to start, then can run down to +1V.
So, if your PV Panels are low voltage, what may happen is: Floating enforces overvoltage to be seen, resulting in a mppt shutdown. Once the battery reached the 53.5V, discharging continious because your solar panels do not provide the +5V required for MPPT startup at 53.5V battery voltage?
That is probably missing some context. You only have to look at what the commercial battery makers recommend, and in most cases there is either no difference or a marginal difference to float. Unless you want your battery to discharge.
That is the now the tricky question. First, from a cell-chemistry-voltage-curve-pov, floating lfp is not required. Charge at 3.5V till full and done. No current flow, no need to reduce chargevoltage.
Now, when you have a pack of cells, imbalances may occur. For a passive balancing, there is no other option than stop charging, if the highest cell hits 3.5V. This however is to be taken care of by the bms by reducing the chargevoltage now and is not floating, more an emergency break due to abnormal cell conditions.
When proceeding to look at active balancing BMS, it becomes trickier. Active balancing means, as the name implies, that bms can actively move energy from one cell to another to balance them out. I cannot see a reason, why such an active BMS would need a voltage “over chargevoltage” to do that, but if that guy is educated on that BMS, there might be a reason for that.
So, the tricky part is now to figure the right strategy: don’t float, cause lfp doesn’t need that generally, or float, because the JK BMS requires that for reasons?
But bottom line at least: If you see a discharging kicking in AFTER entering float, the voltage difference is to high, actively pulling the packs voltage down again for a while.
Here is an example from a bms (non-JK) that does not float but does adjust its CVL.
You can see the cell imbalances increasing at top of charge, still under the 5mv resolution the BMS targets (the left hand scale is mV). At the end sunlight is reducing, so does the battery voltage.
When it has not been able to sit at top of charge for a while the imbalances are much greater, up to 20mv but this resolves itself fairly quickly.
(you see deltas developing at the bottom end of discharge, influenced by the loading).
Works exactly as explained clearly by Andy Offgrid Garage YT.
Am also having an off grid setup with Multiplus 2, SmartSolar MPPT, Smartshunt Cerbo Gx with DIY battery with JK BMS V15.41 EVE 280Ah with similar settings.
Am not using ESS as completely off grid. MPPT charges the battery to full and once absorption is complete at 55.2V for 30 min, shifts to Float mode (53.6V).
During the transition from 55.2V to 53.6V, MPPT reduces charging current. It again increases charging current from 53.5V and maintains PV generation equivalent to AC houseload while keeping the battery at idle
No, the JK BMS doesn’t require it. Andi recommends maintaining a cell voltage of 3.45 V for a period of time (e.g., 1 hour) so that the JK BMS’s active 2A balancer can balance the cells, and then to reduce it to 3.35 V, as this is the “natural open-circuit voltage” to which a fully charged LiFePo4 cell would return on its own without an external voltage being applied.
This is an optional feature of the JK BMS that you can enable if you want to. It changes the CVL, sent to Victron.
All these values must be set correctly in the JK Inverter BMS. 3.45V is the typical value for “SOC-100%”, “RCV-Time” - typically 1 hour - is the time after reaching 55.2V (= 3.45V x 16) for which the voltage is maintained to ensure adequate balancing. It’s important to know that the voltage of individual cells can exceed the “SOC-100%” value, and may even become too high. In such cases, the BMS will temporarily stop charging.
So, this is actually where the OP has the Issue / started the topic. Let’s not drift into a general floating/Bms discussion, even tho I somehow started it and it’s quite interesting
What you describe would be the normal way, how the system should behave. OP reports that MPPTs do not resume production, when falling down to floating voltage.
And as far as I understood this is the main concern, not the short discharging when entering float.
Thus, @Stefou you may have overseen my edit and this question?
- Once the BMS enters “Float Mode,” the battery does not stop discharging after 53.60 V, and during this time the MPPTs remain at 16 W indefinitely in absorption mode with the Multiplus.
It’s possible that I configured the Absorption and Float modes incorrectly in the “Charger” tab
And at the same time, I’ll install the latest version of JK-BMS 19.27
because the release notes mention things about Victron communication and the CAN protocol
V19.24 Upgrade logs
1.Optimize the heating control logic.
2.Optimize the bug causing failure in importing parameters of V19.21.
3.Bluetooth communication optimization.
4.Optimization of Victron’s Inverter Communication Protocol.
5.Fix some known bugs
-@2025-12-18 by JKBMS
V19.27 Upgrade logs
1.Added the “MOS anomaly” option to DRY trigger list.
2.Optimization of CAN Communication Protocol
3.Fix some known bugs
-@2026-01-23 by JKBMS
I ran into a very similar issue with a Victron + JK BMS setup (JK firmware 19.18), and after quite some testing the root cause became clear.
This is not really an ESS or inverter issue, but how DVCC handles the charge voltage coming from the BMS.
When the JK BMS switches to float (for example from 55.2V → 53.6V), it updates the charge voltage via DVCC. Victron does not interpret this as “let the battery naturally settle to that voltage”, but as an active target that must be reached as fast as possible.
In my case this resulted in a very clear (and unwanted) behaviour:
as soon as the BMS switched to float, the system started actively discharging at full power (~15kW) to bring the battery voltage down from ~55V to 53.6V as quickly as possible.
So instead of a smooth transition:
the battery was forcefully discharged
energy was effectively wasted
and the system behaviour became unstable
This is exactly the same mechanism that can also explain:
why the system does not properly stabilize at float voltage
why transitions between absorption and float behave inconsistently
Key takeaway
Victron should be the only system controlling the charge phases (Bulk / Absorption / Float).
The BMS should only provide limits (max voltage, current, protections).
If both try to control the charge voltage, DVCC will continuously react to changes and can cause aggressive corrections like forced discharge.
What worked for me
The most stable solution was:
Disable float behaviour in the JK BMS (or prevent it from lowering the charge voltage)
Keep Victron in control of:
Absorption (e.g. 55.2V)
Float (e.g. 53.6V)
Important nuance:
On JK firmware 19.18, the float timer can still be used for SOC reset, even if the voltage itself does not drop.
An alternative is to use a “fake float” in the BMS:
Float voltage = same as absorption (55.2V)
This allows SOC reset without triggering a voltage drop via DVCC
Result
No forced discharge (no more 15kW spikes)
Natural voltage drop instead of aggressive correction
Clean transition handled by Victron
Stable ESS behaviour
In short: avoid having both the BMS and Victron control the charge voltage — that’s what causes the conflict.
Hmm. I run my system for little over a year now, but never seen that odd discharge behaviour. Not with my two v19 Jk-BMS’es and yes float is still enabled. For the exact same reason Andy mentioned it in his video.
Here’s an update:
After checking, my Multiplus settings are correct.
My JK BMS has been updated to version 19.27.
I enabled “Float” mode, but the battery continues to discharge after 53.6 V. I checked my MPPT settings in the “Cerbo GX” menu > Device > MPPT 150/70, and there it is! I see that the charging voltage is set to 52.5 V.
When I got close to that voltage, the MPPT started working again to recharge the battery / maintain it.
So this value is somewhere on the JK BMS, but I can’t see where… (my JKBMS settings are in my first post)
I’m thinking of the value 3.28 V (52.5/16 = 3.28)
I’m going to try another test, by changing the RFV or something else…
(my JKBMS settings are in my first post)