Which makes it extra strange that there is a difference with DESS on or off…
With DESS off I can lower CVL without any problems (charging stops until battery < CVL; just as expected). With DESS on always (I can reproduce consistently) a big discharge to grid when CVL is lowered (until battery voltage=CVL). (and I am certainly not the only one who has this same issue)
You’re definitely not the only one. Me and many others in a Telegram community with Victron users experience the same weird behavior.
This is not OK and can blow main grid fuses and can cause dangerous situations where you select cable and fuse sizes and it just ignores the limits you’ve set to comply with those cables and fuse sizes!
who dares to try float 48V… 
I think the results will be the same. The inverters will invert at maximum power. It is doing it now till 53.6V and will be doing it till 48V. It will discharge just longer at maximum capabilities. Nevertheless it is not nice for the LFP battery cells!
In my situation with 3 batteries it is drawing almost 400A from the batteries, so 0.42C. Imagine if I only had one battery. It would definitely trigger the overcurrent alarm on the bms and shutdown the battery.
I reported the same issue in January in a posting that was referenced earlier in this thread (“DESS and float mode is not a good match”). Since then, I reenabled float mode, but with float voltage only slightly lower than the charge voltage to limit the discharging at the end of a charge cycle (RCV 3.450V, RFV 3.444V). Disabling float mode altogether did sometimes cause another undesirable effect: when the target SOC of DESS is close to 100%, say 99% and the BMS resets to 100% before the end of the hour, DESS would aggressively try to reduce the SOC back to 99%. With float mode enabled this doesn’t happen because the JK inverter BMS doesn’t reset to 100% until after the end of the absorption period.
I have experienced that (100% soc algorithm not working;too soon 100%) with float mode disabled. Now i have set RFV 3.45V (same as RCV). I have to see if this works well.
btw when you disable float mode only on the master bms, the slave bms soc 100% still working. In Victron I see only soc 100% when all bms are at 100%. CVL parameter is only reporting from the master bms.
Thanks for pointing that out. I’ll try too to change the RFV to 3.444V instead of switching float mode completely off.
I do agree with @darryl12123, that this strange behavior can be potentially dangerous.
Is it possible that victron technical staff can look into it more thoroughly. It seems to me that it is really a technical problem in DESS. It is no problem for me to do some more testing if I can help with this.
Sure we are looking into this.
So, you could run “the light test” again. Set your VRM reporting interval to 1 minute before, eventually more distinct values may help.
I mentioned that I can see your DESS reported to be in “regular self-consume mode”, I’m not 100% sure if this report is suspect to a 15 min imprecission with a lower reporting time as well.
So, to be sure about this, the following dbus-paths would be quite interesting, while this issue happens:
/system/0/DynamicEss/ChargeRate
/system/0/DynamicEss/ReactiveStrategy
/system/0/DynamicEss/Restrictions
(just to make sure by that time DESS is really in the mode expected and supposed to “do nothing”)
I’m also looking for internal test-systems we could potentially use to reproduce this.
Well, it is an undesired behaviour I agree on that. But dangerous? When you setup a system with a potential output of - for example - 15000 kVA, you should size everything in a way to be able to withstand that power.
Just as an example, the german-grid-code forsees that any ESS has to ramp up to 100% nominal output, when grid frequency drops bellow 49.8Hz. That includes ignoring any (soft-)limitations.
(kind of last resort to try to stabilize the public grid)
Purely based on that, relying on existing software-limits set to a lower nominal output should never be considered a safety-measure.
I have set interval reporting to 1 min.
I will try to do the same tests this afternoon. I am not 100% sure if I will get to it, but if not I will do it tomorrow. I also have to wait until soc=100%…
I assume that if the “Doel laadtoestand” shows 0%, DESS is not doing anything and I can start testing.
I agree that it is only dangerous if the cables are undersized and the fuses are oversized. But during this strange behavior the Multiplus does give more than the nominal power. (3xMP8000=3x8000x0.8=19kW nominal, I have set inverter limit to 15kW). Also the limits are ignored. For me personally no problem. All wiring is oversized.
1 Like
Watched both videos in Full HD again.
The drop out of the AC coupled PV seems to be tightly related. The Multi starts to discharge while PV goes from 11 to 15, then drops out. So the huge discharge visible may not be caused by the reduced CVL as a direct consequence, but by the sudden absence of ACPV. (The multi has to provide 3000 Watt loads + 8000 Feedin out of sudden. It also quickly normalizes down to only match the loads)
To see, if that really has an impact it would be also nice to see the reaction, when the ACInverter is temporary turned off. (But then your BMS doesn’t switch to float… )
But you could probably try to limit the Solar-Edge to a minimum power (only covering loads + a minor feedin, 500W or so), so the BMS can switch to float, because the multi doesnt need to discharge. Just for seeing the reaction “then” ofc.
So, the happening is probably: “lowered CVL cuts off ACPV, making the multi to have to take over, causing discharge spikes”.
About a month ago I have finished installing the system. I only had a 628Ah battery then. I tried DESS trade mode for a few days. Then the effect was also with soc → 100% → discharge battery to match float CVL (and overcurrent alarm bms) (I didn’t know then that CVL->float was the problem). SolarEdge was not yet controlled by Venus OS (V3.55) 35kW (DESS+Solaredge) was then fed back into the grid..(17kW aansluiting)..The days after that I manually switched off SolarEdge just before soc->100% (there is a counter in the bms (standard 3600s (1h absorption), but for testing I can set 360s), so I can see exactly when soc goes to 100%).
After that I switched off DESS and had no more problems. (CVL 55.2V → 1h → 100% → 53.6V)
Now (1256Ah battery) I have DESS in green mode because this is more predictable and I can experiment more easily. Fortunately, the SolarEdge is now (v3.60) controlled by Venus and a very large feed-in to the grid is prevented. (this is what you see in the video)
I think the AC-PV has nothing to do with the discharge when CVL is lowered.
For the next test (probably tomorrow) I will turn off AC-PV feedin,
Ofc, it is inevitable that a discharge will happen, wenn the CVL is lowered.
See, usually Lead-Acid-Batteries have a immediate drop in Voltage, when charging stops. That drop is what a floating-voltage should account for - so the lead-acid battery is transitioning from a “beeing-charged” to “idle” state smoothly.
LiFePos don’t have this voltage drop, when the charging-voltage is removed. Once charged to 52.8V they will sit at 52.8V for a long time. Thus, by using a lower float-voltage you are actively pulling the battery voltage down. What else should the battery do then, but discharge?
I’ve tested this on my system using DVCC. With or without DESS doesn’t make a difference:
Usually my CVL is 52.4V (15S), lowering that to 52V has no impact at all, probably within voltage compensation range, just when I set that down to 51V I can see the multis start to slightly discharge the battery:
https://static.equinox-solutions.de/Victron/cvl_with_dess.mp4
Using DVCC to mimic this ofc is not 100% the same, but it just shows that the behaviour of this is quite deterministic: ChargeVoltage < BatteryVoltage = Discharge.
I wil try lowering the max charge voltage in DVCC and see what happens.
But it remains strange that without DESS lowering the CVL does not give any problems and has exactly the effect you expect. (stop charging until battery voltage is lower than CVL again).
The problem really only occurs when DESS is on.
To be continued.
I think what happening (coarse) is the following:
- MP detects an overvoltage due to lowered CVL.
- MP starts to discharge and sends out a throttle advice to the acpv
- ACPV stops completly
- MP now has to account for the power of the missing acpv temporary, leading to a high discharge spike.
Now, we would need to break that into individual chunks and figure out why things happen:
- why is the MP trying to throttle at all? Would acpv plus mp discharge violate some limit settings?
- why is the acpv not just throttling, but shutting down completly? (Is this overshooting throttle request, or missinterpretation by the solar edge? Some inverters also can minimum run at 50%)
Point 1) brings me to an idea: What Maximum Grid-Export values do you have set in DESS? That is a limitation applied by DESS, unknown, when running in regular ESS mode.
Just read that you use the new dynamic power limiting:
Could also give it a test to disable that. Eventually that has a not yet known bug, causing the ACPV to throttle to 0, leading to the observed chain of events. 
Seems to work very well.
For the test tomorrow I will turn this (dynamic power limit) off.
The DESS/CVL → float problem also occurred before I had dynamic power limit enabled (V3.55)
1 Like
I experience the issue since the system was installed in February, so way before v3.60 Venus OS. Also our ACPV doesn’t effect this issue as when DESS plans to charge the batteries at night or claudy weather so there is no PV generation, the problem still occurs.
Can you show the battery parameters (CVL, CCL, DCL) when the max charge voltage is set to 52V and also after 51V, including the moment when there is a discharge.