Closed Loop Accuracy.

[42OhmsPA]

Watch out, I've been thinking....

With multiple packs in parallel is using closed loop really worth it?

How many users are running closed loop with SOC that is not accurate across multiple BMS?

I've ran open loop since day 1.
I attached what my packs looked like this morning, big differences in SOC...

In an open loop setup SOC doesn't matter, voltage rules; open loop is opposite.

So someone using closed loop with 6+ packs in parallel will more than likely have extreme SOC variations due to the poor accuracy of BMS at low current.
For example, under 0.5A or 26W per BMS will not be accurate; with 6 packs in parallel anything under 156W will be inaccurate.

Another issue is cells out of balance. SOC hits 100% or 20% because a single cell hits the trigger while the rest are not full or empty... In closed loop wouldn't charging / discharging stop vs open loop would rely on overall voltage.

I just don't see any benefit to closed loop and often it is recommended to disable it to try and resolve issues...

Make sense or should I get back to work?

 

[marionw]

And you are right. The industry does not have any standards as to when a BMS resets SOC to 100% during a charge. Some reset as soon as at least one cell reaches the Balance Voltage setpoint whereas other require several cells to reach the setpoint for a minimum time. My AIO (SGP 10KW) in closed loop uses max charge voltage and current during the charge, disregarding the BMS charge current and only slows down the charge at around 95% SOC as reported by the BMS and then stops the charge at 100% SOC thus not allowing any time for cell balancing.

Open loop does not ensure BMS cell balancing occurs as it should.

Custom software acting as the "Master" BMS that monitors all cells in all batteries and lies to the AIO is the only way to get a complete and full charge. Some are trying to do this with Solar Assistant automation or their own custom software using an Arduino, Raspberry PI or ESP32 chip.

There is one thing I need to back and check on. When I was futzing with the new EG4 CV-GC communications I noticed that the "Master BMS" was reporting to the CV-GC the same current (around 2.5 amps, idle, no charging) that was the same as my Victron Shunt. With six EG4 LifePower4 batteries this would suggest that each battery BMS was reporting around 0.416 amps to the "Master BMS". The EG4 CV-GC as a "Master" sends an undocumented command to the "Master BMS" and receives an undocumented response so it might (just might) be possible the BMS can actually measure current less than 0.5 amps but does not report anything less than 0.5, except in this undocumented stuff

 

[kommando]

SMA SI ignores SOC in closed loop and goes by voltage target given to it by the Canbus BMS. So a false 100% reported by the BMS before the target V is reached is ignored and the SI continues charging.

 

[marionw]

BTW. One other "Advantage" (if you call it that) with closed loop is that the "Master BMS" reports various violations of battery alarm and protection setpoints; i.e. Cell Over/Under Voltage Alarms, Cell Over/Under Voltage Protection, Pack Over/Under Voltage Alarms and Pack Over/Under Voltage Protection, Over/Under Temperatures etc.

Depending on the AIO and the BMS alarm condition, some AIO's will stop battery charging/discharging. If discharging, the AIO should shift to grid if it is available and stop it's load on the battery while (hopefully) initiate a battery charge.

If only there where "Documented" industry standards we could count on and that each AIO would document compliance in it's manual. The AIO is getting all the battery info it needs, just do the right/expected thing with it. Some firmware programmers are better than others..

 

[42OhmsPA]

I suppose that could be an advantage as long as charge parameters aren't overly aggressive from the manufacturer.

If you're using 'safe' charge settings in an open loop config most AIOs and regular inverters should protect the pack the same way.

 

[marionw]

"charge parameters aren't overly aggressive from the manufacturer." Hence the need for Industry Standards.

In my limited experience the AIO doesn't charge based on all the BMS closed loop provided data.

AIO and BMS manufacturers need to establish a set of "Industry Standards" that clearly define how closed loop communications work, particularly with respect to battery charging. Master BMS reported SOC should not be the only criteria that the AIO uses when charging batteries. The BMS should control bulk and float/absorption charging and the AIO should honor the BMS data as long as AIO charging power is available to meet the requirements.

It has always been assumed that the closed loop "Master BMS" controls the battery charging when in fact it does not. This is why so many are disappointed in "closed loop" and opt for "open loop"

 

[Solar Guppy]

It has always been assumed that the closed loop "Master BMS" controls the battery charging when in fact it does not. This is why so many are disappointed in "closed loop" and opt for "open loop"

My Seplos Hiten's do, and there are many parameters one can adjust to trigger the 100% SOC

 

[On_The_Road]

Why don't these AIOs all in ones have a high quality shunt after the battery connection? You know something like a Victron BMV?

 

[robby]

Running six eFlex Batteries in closed loop @54.5V Max charge Voltage. So long as they get to a 100% SOC all of them are balanced (Reading 100%).
If I have 3-4 days of cloudy weather and the Max SOC only reaches 50-75% then I see that drift between highest and lowest pack can be up to 15%. I think a lot of this has to do with the quality of the Cable as the newer ones that are on a different (inferior) brand of wire charge at a slightly lower amperage and discharge at a slightly lower amperage.

I am pretty sure if I matched the wires resistance exactly I could get rid of this problem.
Closed loop is still far better than when I tried open loop. I would get up to 30% differences on cloudy weeks with Open loop and voltage mode.

 

[On_The_Road]

Running six eFlex Batteries in closed loop @54.5V Max charge Voltage. So long as they get to a 100% SOC all of them are balanced (Reading 100%).
If I have 3-4 days of cloudy weather and the Max SOC only reaches 50-75% then I see that drift between highest and lowest pack can be up to 15%. I think a lot of this has to do with the quality of the Cable as the newer ones that are on a different (inferior) brand of wire charge at a slightly lower amperage and discharge at a slightly lower amperage.

I am pretty sure if I matched the wires resistance exactly I could get rid of this problem.
Closed loop is still far better than when I tried open loop. I would get up to 30% differences on cloudy weeks with Open loop and voltage mode.

What are you seeing for min and max cell voltages at 54.5 / 3.41?

 

[hwy17]

I'm with you, the more I develop my system the more it seems like closed loop would only add headaches.

I have one master battery that's in control of the grid charger input and I will baby that battery and make sure that it's reading it's own SOC correctly, but the rest of them can take what they get or drop out if they have a problem but I don't really care what they think their SOC is.

I'm actually working on an expansion plan where subsequent parallel packs from now on won't even have any current tracking. They'll just have balancing and high/low cell cutout but will otherwise just blindly follow.

The high/low cell cutout is really only for emergencies and malfunctions too, and my open loop voltage scheme is designed that no battery should ever be getting close to a cutout.

 

[robby]

What are you seeing for min and max cell voltages at 54.5 / 3.41?

I don't remember the exact number off hand but they are all balanced because the BMS uses an active balance that balances during charging. I will have to wait until tomorrow and see what the cells are reading.

 

[robby]

I'm with you, the more I develop my system the more it seems like closed loop would only add headaches.

Closed loop was what got rid of all of my headaches. That is why every company pushes it heavily.
If your running a smallish battery bank vs your charging ability then none of this is ever a problem.
When I had only 15Kwh of batteries they would hit 100% SOC by noon and everything was always perfect.
At 35KWh the story changes during the Winter months or the cloudy days. Just watch as winter sets in your going to see more and more threads about batteries being out of balance.

 

[Crowz]

I have 4 batteries in parallel and NONE of them have the same soc charge showing at any given time. Not a single match.

Now voltage looks good across all the cells but soc is useless from them. I use the victron smart shunt for all of my soc calculations.

As for how the inverter handles the batteries its strictly by voltage and works flawless. You couldn't pay me to run closed loop even if I could after all of the headaches I have seen people have on this forum.

 

[hwy17]

Running six eFlex Batteries in closed loop @54.5V Max charge Voltage

That's interesting to me that eflex runs such gentle charge profiles. I'm running 54.8v bulk/absorb right now but they've out-gentled me.

This implies that they believe they can get good top balancing at 3.40 volts. I don't think it's out of the question and I might explore that.

 

[hwy17]

Just watch as winter sets in your going to see more and more threads about batteries being out of balance.

We need to bring back the practice of monthly equalizing charges like they used to do on delco lights back in the 1920's. That's how I do mine.

 

[robby]

I have 4 batteries in parallel and NONE of them have the same soc charge showing at any given time. Not a single match.

They all must be at 100% SOC when they are fully charged unless you are having cell balancing issues.

Now voltage looks good across all the cells but soc is useless from them. I use the victron smart shunt for all of my soc calculations.

To be clear you are talking about individual cells in the packs because adding is SOC to the statement makes it a bit confusing. If you are talking about SOC then your talking the whole pack and all packs in parallel are always going to have the same voltage.

As for how the inverter handles the batteries its strictly by voltage and works flawless. You couldn't pay me to run closed loop even if I could after all of the headaches I have seen people have on this forum.

You could not pay me to go Open loop so I guess we have completely different experiences.

 

[robby]

That's interesting to me that eflex runs such gentle charge profiles. I'm running 54.8v bulk/absorb right now but they've out-gentled me.

Fortress actually told me to charge at 54V but I find 54.5V to be better for my setup. The voltage really needs to be measured with a good meter at the battery terminals due to loss in the line and some bits of inaccuracy with the Inverters readings.

This implies that they believe they can get good top balancing at 3.40 volts. I don't think it's out of the question and I might explore that.

It works for me and it certainly is better to go easier on them if you can.

 

[On_The_Road]

So, how important is an accurate SOC reading to an all in one inverter if you want to do peak shaving and load shifting?

 

[hwy17]

So, how important is an accurate SOC reading to an all in one inverter if you want to do peak shaving and load shifting?

Quite important. It's important any time you want to make a decision based on SOC and those functions require decisions.

But that doesn't answer the question of whether it's better to decide based on 1 battery's SOC or 6 batteries' SOC. If 1/6 batteries is spazzing out for some reason, should the spaz be listened to at all? I would say no in my system. Only the golden battery gets decision making authority.

 

[On_The_Road]

Quite important. It's important any time you want to make a decision based on SOC and those functions require decisions.

But that doesn't answer the question of whether it's better to decide based on 1 battery's SOC or 6 batteries' SOC. If 1/6 batteries is spazzing out for some reason, should the spaz be listened to at all? I would say no in my system. Only the golden battery gets decision making authority.

So how does EG4 and Solark handle this?

It's a total system soc? Average of all batteries? Soc of the lowest?

 

[hwy17]

So how does EG4 and Solark handle this?

It's a total system soc? Average of all batteries? Soc of the lowest?

I don't know exactly but it's yeah it's some kind of combined SOC.

 

[Brett V]

My 4 "identical" batteries all show slightly different SOC readings as well as charge/discharge imbalances according to the internal shunt on the Overkill BMS. It varies as much as 10% but if I average the SOC or current in/out numbers, they're generally very close to what the Victron shunt measures. My natural OCD struggles with this and I've found the best solution is to not look at it too often.

 

[marionw]

The EG4 LifePower4 "Master" battery averages the SOC from all attached batteries as long as the battery is not reporting a protection alert and as long as the Discharge MOSFETS for the battery are On. The "Master" battery also reports total capacity for all batteries where the Discharge MOSFETS are On (If the Discharge MOSFETS are Off then there is no usable capacity for that battery).

The same appears to be true for SunGoldPower's 48v100ah batteries.

The Master battery reporting to the Inverter will of course be based on the model/flavor of BMS and it's firmware. There is no Industry standard so what I observe with my two types of batteries may not be the case for others.

 

[ChrisG]

I've been successfully using one of the ESP32 CAN solutions with my non-inverter JKs which works well overall, but individually the SOC on the multiple JK's are off and needs to reset to 100% at least a few times a week to be accurate. They do not track low loads accurately as someone mentioned above which my Victron shunt does capture. System works, allows the inverter to do 'automation' based on SOC% which is easier to navigate vs guessing voltage. Smart load dump is one example were SOC is easier to use vs voltage.

Edit: See a lot of people struggling with EG4, SOK, and other batteries around SOC full charge and full discharge with battery balancing. These ESP32 projects definitely allow greater control while using active balancing BMS's like the older JKs with custom batteries.