I’m having a similar issue, system started discharing at max power when the battery is full and I’m running in ESS mode. So not reason, at least in my view, to start discharing at max power and returning to grid. My expectation is that solar power is keeping the battery full/floating, the current load comes from solar and everything from solar generation which is left over is returned to grid.
My other post with more details: Running ESS and systems starts discharing 4kW, why?
I’m having the same problem.
In the advanced tab i’m not able to see the fast discharge of yesterday, maybe because it’s a very short period.
See screenshot below.
The problem is only when the battery goes to 100%. (jk-bms CVL 55.2V (bulk) → 100% → CVL 53.6V (float))
In DESS mode the battery is then quickly discharged (P > 20kW) to the grid to get to 53.6V as quickly as possible. This only takes a few minutes (until 53.6V) and then DESS functions normally. (as per the settings)
If I switch off DESS it does go normally (CVL 55.2V → 100% → CVL 53.6V). The battery is then not quickly discharged to the grid to get to 53.6V as quickly as possible.
This is exacly the same problem.
Maybe I’m doing something wrong (I haven’t had my system up and running for very long) or it’s a limitation of the JK BMS, but I find it strange that in ESS operation everything just works as expected, but in DESS operation this effect occurs.
You would expect that if you lower the CVL below the battery voltage the charging will simply stop (this seems to happen without DESS = ok) and not if the CVL is lower than the battery voltage the battery will be discharged (max power) until the CVL is reached. (this apparently happens with DESS on).
Maybe I’m thinking too simple…
I have now turned off float mode in the bms. CVL should then always stay at 55.2V. I will watch it for a few days.
DESS does not read or do anything with DC-Voltages… The whole DC-Side is just a blackbox providing “Power”.
If DESS wants to discharge “15 kW” it just tells the inverter “discharge 15 kw, please” no matter what voltage or something there is.
So, IF a discharge due to overvoltage is happening, that is caused by the underlaying ESS-Logic or maybe even down deeper to ensure system stability. (Where we ofc. have to investigate why this is happening when switching from bulk to float)
@Richard28 I’ve send you a pn, if you could give me your VRM-Portal ID, I may review if the relation to DESS is jus a coincidence, or if there is something that “triggers” this behaviour.
This can’t be the solutions to this problem. If my PV system is generating 12 kW and I have the feed-in limited to 14.6 kW (same as my ESS setting for maximum inverter power), my batteries are limited to discharge only 2.6 kW, so basically limiting my system. Our grid connection is capable of 28 kW.
The questions is; why does the system handles the requested battery voltage differently without DESS en with DESS. Without DESS it will slowly lower the voltage depending on your home usage, and with DESS it will discharge the battery at full capacity of the inverters and ignores any limit set in the system. It just doesn’t make sense.
I’ve reviewed two systems, there is no relation to DESS visible.
Also our DC-Expert says this is caused by the multiplus firmware itself, so even way more technical handling than even the regular ESS-level. Lowering the CVL brings the multiplus into an overvoltage situation, where It tries to get rid of excess DC Voltage at the highest pace possible.
Seeing this pace “vary” from day to day may more be a result of the PV Power and actuall cell-voltages available, the moment the CVL is lowered.
This happens on systems where you have it configured to feed excess DC coupled PV into the grid.
When the battery suddenly lowers the charge voltage, the Multi sees it as an overvoltage. It has no way to know whether this is due to a lot of PV, or because the battery charge voltage was lowered. It just sees an excess. And it feeds it into the grid.The inverter power limit does not apply to this scenario. But if you configure a lower system feed-in power, that would indeed apply.
I would raise the float-voltage to a level where this effect becomes neglectible. Floating is not really required for LiFePo-cells, so the voltage-drop can be avoided and so would the “emergency-overvoltage-discharge” be. (Floating is a process commonly used for batteries with a high self-discharge, LiFePos not doing that)
I’m going to experiment this coming week. I’m pretty sure this effect only happens when DESS is enabled. That seems strange to me. My system worked fine for 2 weeks without DESS (CVL55.2V ->100%-> 53.6V).
I do agree that floating (lowering CVL->100%) might not be that big of an issue.
Feel free to share dates, where you have an observation. DESS also may have different strategies around the time this CVL drop happens.
It may be, that a certain strategy will increase the effect, but bottom line it’s the very basic firmware causing this.
For example, it may make a difference, if the inverters are already converting dc to ac during that time, or if they are currently also charging the battery. (Based on that, the multiplus voltage readings are suspect to a voltage drop correction and hence the emergency-overvoltage-discharge may be higher or lower)
Does DESS have a strategy when the SOC becomes 100%? This is also the exact moment when the CVL is lowered to float voltage. (BMS reports SOC 100% and lowers CVL at the exact same moment).
btw I only have AC coupled PV
On my system it happens every day. In DESS with trade mode there is no active strategy, charging or discharging. With trade mode there is only charging the batteries to 100% at least once a day. It doesn’t matter if there is PV generation or not, only if the energy prices are low enough.
So when the CVL drops to float voltage there is no charging nor discharging at that moment. Float voltage only happens when batteries are 100% for one hour so there can’t be any charge/discharging at the same time.
I did some testing
DESS ON
6s, CVL 55.2V → 55.0V
17s, discharge battery to grid to match voltage
28s, PV shuts down. (beperken systeem terugleveren staat op 16kW)
This was only with a voltage difference of 0.2V. You can imagine what happens when CVL goes from 55.2V → 53.6V…(huge discharge) I don’t dare try that anymore…
Last time I tried this (V3.55 no limit Solaredge) more than 35kW (20+kW multiplus + PV) of power was exported..(3x25A grid connection…)
What will happen if someone accidentally sets the float voltage much lower…(I assume that with DESS OFF the charging will simply stop until this voltage is reached (as expected). With DESS ON I expect a huge discharge until CVL is reached)
DESS OFF
10s, CVL 55.2V → 55.0V
Nothing crazy happens. Exactly as expected. No problem at all.
The problem only occurs with DESS enabled. I can consistently reproduce this problem.
The system has been running for 2 weeks without DESS. No problems at all. (CVL 55.2V-1h-100%-53,6V)
For now I have set float voltage and absorption/bulk voltage equal in the BMS
Thanks for testing and showing the problem. When I get home I will just disable float mode in the JK bms completely until this issue has been resolved. In the menu you can switch off fload mode.
Thx for the videos, will watch and check your system details when I have some time.
On first view, it looks like we need to be very carefull about figuring out what is the actual cause, and what is the consequence.
At the beginning for example - before PV drops out - DC PV ramps up from 11 to 15 kW. So, the multis cranking up their discharge as a consequence of “this”, until dcpv pulls the hand break - and in that moment, the multis power drain hits the battery as it out of a sudden became the only dc power source.
In your second video, DCPV is quite steady sitting at 15kW, so not causing a change in the power the inverters need to invert.
I will check some details and also ask our DC-Expert about this as well.
I don’t think he has DC PV. It is his PV on the AC1 out side, as I can see 3 phases on his PV system.
I don’t have DC PV as well and experience the same issue.
Your right. Just watched on the phone, thought bottom right is the DCPV box.
Then, it is even stranger. AC PV is usually running pretty much on it’s own.
Maybe I don’t understand you well, but I don’t have dc coupled pv, only ac coupled pv (ac-out)
It seems to me a real structural problem in DESS.
When DESS is off, everything (change in CVL) goes as expected.
It has nothing to do with DC PV. (I don’t have)
Also at the moment that the CVL → float voltage the battery (100%) is not active (idle). There is no charge or discharge demand. (especially since there is plenty AC PV)
Just sharing some findings and thinking a bit loud - hope you don’t mind 
By the time you made the first video (14:05, with DESS enabled), DESS was in Scheduled Self-Consume mode. That is, it is doing nothing (!) but regular ESS-Mode operating on the grid-setpoint.
Generally DESS has no control over the DC-Side, it does not even read any dc-value. DC for DESS is just a blackbox (like the grid), where DESS is saying “Please give me X000 Watt now”.
DESS really did nothing different here than regular ESS would do. Or to be more precicse: DESS does not do anything, it just says “Hey inverter, do your ESS thingy” 
Looked a bit around your values, there is not really much to see. The moment is “too short”, and you seem to use a vrm reporting interval of 5 min? (even 1 min wouldn’t show much more details I think).
Only thing I noted is that your unit two is showing quite some different terminal-voltages than the others do. Should check connections and cable length there. Most likely not related, just noted.
Does the Solar-Edge provide some internal logs? I really wonder, why that dropped to 0 - and how. You’ve been grid connected, so can’t be frequency shifting. Power Limit (through Modbus?) Also doesn’t show something.
Maybe figuring that out could give a hint.
Unfortunately my BMS doesn’t use a reduced float voltage, so can’t reproduce if that would happen on my system as well.
The different voltages seem strange to me. Between 13:00 and 18:00 the soc is 100%. There is then no or little current in the dc cables.(battery idle) Even with a bad connection (which I think is not the case) if there is little current flowing, then there can also be no voltage difference. I have checked the installation (under full load) with a Flir camera. There is almost no heat development at the cables/connections/fuses. (all cables (2x70mm2 with 2xNH00-125A fuses for the mp8000 less then 1m length) exactly the same length to busbar (600mm2 copper)). I will also check it tomorrow with a multimeter directly on the multiplusses.
I have checked the voltages directly on de terminals of the MP. They are all the same (now 300A discharge to grid → 100A/MP). The voltage drop between MP and busbar is 45mV (+) 12mV (-). ((45mV+12mV) x 100A = 5.7Watt). I think the voltage drop (cables&fuse) is very low. (All 3 mp the same voltage drop)
Is there a problem with unit 2?
Solaredge is controlled by Venus OS (vrm logging was on 15min, I think that is the cause you see nothing in the vrm log, even when I set to 1 min log, you probably not see it)
