WallK

It should!
As far as I've read it's balancing by using input current, not by discharging the highest cells (or something along the lines)
With the idea behind it being all the cells will become fully charged thus really balanced, not just equal voltage
I'm not sure about the specifics, but the disconnected battery with BMS doesn't balance, at least perceptibly
The highest cells just passively lose charge a bit faster. But this is an anecdote, I haven't left them for any substantial amount of time
I will check when I have good grid power if it goes over the float
But mine stops at 100% or close to it, it doesn't stop THAT early...

This is exactly how balancing works
It goes into OV and then charges the cells that are under this
How else do you plan to actively top balance cells?
Your link recommend not charging to 100%, but most sources say it's fine or even recommended
And I'm not talking about seller here, I'm talking about battery designers here, Current Connected
 

Just to reiterate -- this only happens on solar charging, grid charging is fine for some reason...

Sorry, what? If I put 30A into system of two parallel batteries they each will take a part of it depending on their resistance
They will take fractions of available current, so that <full current> = <bat1 current> + <bat2 current>
So in my case of almost identical batteries one will take half and the other will have remaining half
Literally <available current> / <number of batteries>
 
I don't have JK BMS and theirs charger settings will not allow the battery to ever balance
I will not change the BMS settings from designer's recommended (that we discussed with them) for the fear of losing a pretty long warranty
Also, those charging numbers are what BMS requests when connected over CAN or
 
I've tried other settings and it's getting stuck at whatever is set as float

30A max for two 100A batteries in parallel is low enough I think (15A per battery)
 
It is literally the spec

And it happens only on solar charging
 
I don't think so, voltronic said it was a bug and provided me with 60.66 firmware built to fix it

My bulk is set to 57.6V, and 55.2 is exactly what is set as float voltage
My battery never goes even close to the CV for some reason
It did before on grid and even overshoot that voltage for a bit

My batteries are SOK LiFePO, BMS goes into OV at some voltage and starts balancing
Never happens on solar for some reason

Also, going to this firmware from 60.66 to 60.96 patched reintroduced the bug when inverter is using ~2A (~400W) from grid to charge the battery even is Only Solar Charging is selected

I've flashed this firmware with the hopes to see my batteries finally balanced
But it just stops charging at float voltage still (please check attached image)
This happens on solar-only charging, with grid it goes to float (at least it did on other firmwares, can't check right now)
I have still plenty of solar available at the last point
Was I mistaken and the fix is not for this?
Thanks!

Ah, it must be 16S charged to 3.6 VPC average. I only charge mine to 55.2V. 54V is 3.375 VPC, which is slightly above the really flat part of the LFP voltage versus SoC curve, so your LFP battery should charge fairly well to a fixed 54.0V. Much better than lead acid would. 

Yes.
That should work, but if there is not enough solar, then you might run out. 54 V is a float voltage, good to maintain a battery that is already full, but it will take a very long time to fully charge a battery at that voltage.
Oh. Well in that case, yes, you will want to target about 54.0 V, otherwise you will eventually over-charge your battery. The float voltage is the highest voltage that a battery can safely handle continuously. Such a power supply is not ideal for battery charging.

Ah, OK. When you say blackout, it sounds like you're more or less guaranteed grid supply, but only for a limited number of hours per day.
If it's at float voltage, it doesn't even matter if it does.
I guess so. You may need to experiment. I'm rusty on how lead acid works; I seem to recall that at float voltages, the battery just won't take much charge current, regardless of how much current the charger can supply. That's why a fixed voltage charger is so far from ideal. But since your energy needs are so low, and you seem to have access to at least some grid energy every day, it might be OK. With an LFP battery, this would be fine. With lead acid, where you don't want to regularly  go below 50% SoC, I don't know.
By the way, I'm assuming that this Power Supply Unit has current limiting. So if you set it to 54 V and the battery is at 49 V, it will just limit the current to a safe maximum (safe for the PSU and also for the battery), assuming that pushing it to 54 V would exceed at least one of those limits. Many fixed voltage PSUs won't have this capability.

For all those that have used my patched VM III and VM IV bugs, my humblest apologies! I included a fix for the premature float bug, thinking that it was a two-word patch, but I misunderstood the logic, and as reported by several users, the patched firmware never goes to float! These are the perils of patching code that you can't test on an actual machine.
The first fixed patched firmware I'm releasing is for factory firmware version 60.95, mainly because user @roadrunner79 sent me excellent graphs detailing exactly what the problem is. This change affects all the variants, so I've numbered the fix for these as x60.96, even though they are based on factory firmware version 60.95. I hope that there is never a factory firmware version 60.96 to get confused with.
If you are using any of the x60.95 patched firmwares, please update to x60.96 at your earliest convenience.
Once I get some feedback that I've correctly fixed this bug, I hope to apply it to other VM III and VM IV patched firmwares.
These are patched firmwares version 860.96, 260.96, 560.96, 160.96, and 960.96, all based on factory firmware version 60.95 for the Axpert VM III Twin 6kW and 4kW, and the Axpert VM IV Twin 6kW and 4kW. These patches implement most of Georg's MPPT "stuck at 90V" patches (here it would be "stuck at 60 V"), and fix the premature float bug. The only parts not applied are for the forward current constant, which doesn't seem to be changed based on solar voltage in any VM firmwares. The version number should display in full when using monitoring software, but will still show as 60.95 on the front panel display. Do not use with any other models, not even older VM III and VM IV models with power ratings like 5 kW or 3 kW, or non-twin models. Use at your own risk.
In the reflash tool, you may have to navigate to the folder with the patched files the first time it is run. Choose the hex file with the minimum MPPT voltage that you want:
860.96: float bug only
260.96: 200V min MPPT
560.96: 150V min MPPT
160.96: 100V min MPPT
960.96: 90V min MPPT
Use the 200 V variant if your PV panels are nearly always well above 200 V. Otherwise, use the 150 V version if your panels are almost always above 150 V. Use the 100 V version if your panels are particularly low in voltage. Try the 90 V version only if you have just 3 panels, the very bare minimum.
You can also of course choose the original factory firmware.
Firmware upload instructions.
 
dsp x60.96 VM III Twin patched.zip

Yes. There is an AC->DC converter, followed by the usual DC-AC converter. I'm not familiar with the King II, but I believe it's the same. 
 
I have no idea what the frequency sync parameter is about. It sounds like you can optionally ignore the frequency and phase of the input AC, but if so I'd expect that to be present in the non-twin and even the King 1 models. 
I'm 99% sure that the second output merely parallels with the first/normal output. 
Yes, that's an option. I ran a single inverter like that for several years. The only thing is you should set the external charger to about 0.1V lower than the inverter(s). The inverter can top off the battery on its own, and won't get confused by the battery voltage not behaving as expected. 

Not what you want to hear, but I wonder if the King models (original and/or King II) are better with poor grids. They have double conversion. I suspect that means that they don't have to synchronise with the grid, and so may not care nearly as much about grid quality. 

It could be unhappy with the grid frequency, or the rate of change of frequency, or perhaps rate of change of voltage.
What's the grid like? The voltage seems to have a quite wide range, suggesting poor voltage regulation.

None of the VM models can be paralleled. They come with half-sized flash; no room for all the paralleling logic.

Yes indeed if i am not mistaken the max input current of the VMIII Twin is 27A and with 2P you will be 2.52A over the manufacturer spesification, but you should be ok on that margin the array won't spend much time at Imp in any case. The tracker should clip the access current should it be necessary.

Great, thank you!
 
 
If you want to use esphome -- I use this project: https://github.com/syssi/esphome-pipsolar
I've done a fix for dual output inverters (longer QPIGS) and it's live in pip8048 branch

That's weird, coz I have 44-55 range in 25.06 already

 @Coulomb will you try to apply your mppt fixes to the new firmware or is it finally fine as it is?
I'm ok with even older firmware capabilities, but I really want the "premature float" bug fixed 🙂

This happened to me when I tried to update my VM III Twin with monitoring app still open
I, actually, did it twice
One time I recovered using a USB stick with firmware
The second time I just made sure nothing is blocking COM port and flashed from PC
This required starting flashing as soon as inverter starts up, if I let it sit for a bit -- the flashing will error out for some reason

Unsurprisingly, it's the same as other 60.xx; it has the current control parameter (which the early implementations lack), but not what I'm now considering the "full fix".
[ Edit: Oops! That was an error in my disassembly. Sigh. It doesn't even have the current control parameter. So the good performance is likely due to the ideal PV voltage, I'm now thinking. ]
But you don't seem to be losing much of any significance on those particular cloudy days either. So maybe the current control parameter is the main improvement. 
Ah. I should have realised when it was so very flat. Thanks. So that's even less time where you might have been missing out on solar production.
Or maybe you just have "sweet spot" panel voltage or something. Not too many, but enough.

You haven't got it too bad on those graphs. Yes, on the tenth of June you had it for about 10 minutes, when the PV voltage was ruler flat, but then the current was less than an amp, so you didn't miss much.
It's mainly cloudy days that owners notice it; it can stay at 90 V (or 60 V on VM III/IV) for hours. Then suddenly the voltage triples and the PV power triples with it.
I believe that I can recognise the code that works pretty well now. And sadly none of 60.89, 60.95 and 70.00 seems to have it.
Do you happen to have some records from a cloudy day?

I have 60.66 installed

Two previous days were cloudy and very cloudy
Here are voltage and current graphs for those


I don't have enough power storage to survive on solar only two days like this in a row, so I imported some power from grid

 
  
I think that part was a data gathering issue (if we are talking between 7:00 and 8:00), it has only two datapoints for 10 minutes, so this is why it's so flat there 

I don't get any overvoltage alarms
My battery is connected straight to the inverter with RS485 cable
I charge using grid power
Here's voltage and current graph from today when power was available and I didn't drain battery up until 8am (volts blue on the left, amps orange on the right)
You can see how often it charges battery without any power going out of it

Also a graph of battery SOC (blue, left) vs max charging current requested by BMS (orange, right)
As you can see, it goes to 20A on 99% and to full 80A on 89% SOC