Sol-Ark SOC tracking issues

[Cmiller]

Sol-Ark's SOC tracking algorithm runs off and jumps around! And so far I have gotten no satisfactory response from them on any solution!

The scenario is that we have quite a few Sol-Ark's out there with LifePo4 batteries that have no closed loop comms, but with shunt based SOC screen built-into the battery. (Ark battery brand) The Sol-Ark inverters are terrible on tracking the SOC!!! We see them often being off by 50% or more, and the SOC will jump as much as 50-70% at the drop of a hat!

I have worked with battery monitors from Schneider, Magnum, Midnite, Outback, Tri-metric, Victron, etc., etc......... And Sol-Ark is the worst, by far!

It is extremely disappointing that they are not better, when their spec sheet even says they have a shunt "for Accurate % SOC".

Where we really run into issues is on the (quite a few actually) jobs that we are using Time-Of-Use settings to do Limited to Home with no grid sell. When you go SOC based, there are issues where the Sol-Ark says SOC is 50% and the batteries are actually at 15%.... or 5%.... or even completely dead.... I'm talking 49-50V on 16s LFP batteries. Then when the voltage hits the battery empty volts setting, (49V is where we generally have it set) the SOC snaps to 0%.

If we are using the gen start relay this obviously creates issues if we want to start the gen at say, 15% when grid is down, but the Sol-Ark thinks SOC is still at 50% until the batteries are flat dead! There is a huge risk of the system shutting down or even the BMS's shutting down in this scenario. We have one customer who goes to Florida over winter, and he wants to be able to not worry about the system while he is gone, and of course with this issue you can't really actually depend on the backup generator to for sure start in time.

Switching to voltage based is not a viable option, as the Sol-Ark uses the Time-Of-Use setpoint to limit gen charging as well......

So for reference: Using voltage based settings, TOU set to 52V, gen start set to 50.7V (roughly 15%). Now when grid happens to be down for a few days and the battery volts hits 50.7V and we begin to charge at 175A (using Kohler 14KW standby gen), how long until our 300-500AH battery bank hits 52V????? 30 minutes..??? How many times would the gen start in 24hrs? Many times!

Sol-Ark says "use a battery with closed loop comm". Why? Why do they say they have SOC monitoring, when in fact, it doesn't work? Why do we need to come up with work-arounds to overcome the shortcomings of an inverter not having what it is advertised to have?

I emailed them about this issue as well as some others and they at first said they need a few days to look over the issues. Then I got no response and finally after 4 weeks I emailed them and told them I need answers! Now they emailed me back and here is the response:



Here is my original email:



The spec sheet for the 15K says they have a shunt "for Accurate % SOC". I would not say that showing 50% when batteries are at 25% is anywhere near to "accurate"! Neither is it an acceptable margin of error. That is what we saw yesterday on our one job. (Actually the customer sent me these pics wondering if this is right or how it should be.... How do I respond to that...??)




This battery's screen is actually shunt based, and these screens have always been very accurate! Obviously any SOC monitoring device can, and will, sometimes run off! But in the grand scope of things, these have actually done better than any battery monitor from any inverter manufacturer that I have ever worked with. So as you can see, the Sol-Ark isn't even remotely close.....

As far as the jumping around of the SOC on the Sol-Ark goes, here are some screenshots showing that:











Some of these screenshots are from PowerView, others are from a Solar Assistant that is on our one job, so all of that data would have been directly collected in realtime from the 3x Sol-Arks that are on that job.

I really wish this post would somehow get the attention of Sol-Ark, and they would work on this. But at this point my faith in that is pretty small...

The Sol-Ark inverter can't be beat in terms of simplicity and also they just have an absolutely great all-in-one platform! That is why we use a ton of these units! And compared to other brands they rule in areas like stringing solar panels (26A pr mppt, why doesn't anyone else do that????), 200A passthrough is great and not many others have that, the menu is better than pretty much any other all-in-one (although it is way behind when compared to non all-in-one setups like Schneider, Victron, etc., etc., etc.), built-in battery breaker, we really like their battery terminals as well, and even their solar wire connectors are hard to beat!

But if I constantly have customers calling with issues that have to do with a problem with the programming of the inverter, I very quickly start feeling like it would be worth it to revert back to a non all-in-one system. A system that we can install and know pretty much with 100% certainty that we can walk away from it, and it'll just work! A system where we can tell the customer "your generator will start before the batteries are dead". A system where we can say, just look at this screen to see what % your battery charge is at.

Sadly, this issue is literally only a software issue! All it would take is some programming, and it could be fixed! But the response is that there won't be a change.....

P.S. For a backdrop, I have been in the solar industry for over 10 years, I have personally installed probably 400+ battery setups, and we as a company have installed 100+ Sol-Arks. So I'm not at all new to this stuff. I have seen things that work, and I have seen things that don't work. I prefer what works! ??️??

 

[SeaGal]

A system where we can tell the customer "your generator will start before the batteries are dead".

Maybe I'm over-simplifying things, or have missed something... but if you don't want to connect comms from battery to inverter, couldn't you make a little controller (I'm thinking of ESP32 or RPi here) that reads the SOC from the battery's RS485/CANBus port and then uses the battery's SOC measurement to activate a relay to start the generator?

 

[Cmiller]

Maybe I'm over-simplifying things, or have missed something... but if you don't want to connect comms from battery to inverter, couldn't you make a little controller (I'm thinking of ESP32 or RPi here) that reads the SOC from the battery's RS485/CANBus port and then uses the battery's SOC measurement to activate a relay to start the generator?

We actually usually basically do this, only we use the Victron BMV-712 battery monitor, and set it's aux relay to our desired start/stop parameters. Where we have issues is when we are doing partial discharge using Time-of-use settings. As long as the Sol-Ark's SOC stays (within reason) on track it all works fine. In fact we can actually use the Sol-Ark's gen relay then. (The stop is then determined by the SOC setting in TOU, which works.)

But when SOC runs off, we are forced to revert to voltage based settings. A Time-of-use setting of say 52 volts will also then cause the inverter to stop charging from gen at that voltage. So to start with an external device creates the scenario where the gen can auto start and be running for hours on end but simply not charging, because the voltage came up to 52V. (Or wherever the Time-of-use setting is at.) In that scenario the gen could potentially also not stop because you have a "deadlock" situation where gen is running, Sol-Ark isn't charging because of battery voltage having reached TOU setpoint, but battery won't hit stop trigger because Sol-Ark is no longer charging.

 

[Cmiller]

Maybe I'm over-simplifying things, or have missed something... but if you don't want to connect comms from battery to inverter, couldn't you make a little controller (I'm thinking of ESP32 or RPi here) that reads the SOC from the battery's RS485/CANBus port and then uses the battery's SOC measurement to activate a relay to start the generator?

One more note on what you suggested. This issue is mostly related to having batteries that have no closed loop comms option, or a scenario where there is a software issue with the batteries, therefore not being able to use the closed loop comm.

If closed loop comm is available, and working correctly, then the simple answer would be to just use closed loop comm to the inverter.

 

[robby]

There is no scenario in which Voltage or Batt-% monitoring will give you accurate SOC readings with LFP batteries no matter which Inverter you use. It's always going to be drifting and typically it is reading a lot higher SOC than what is actually available. I have had a Sol-Ark 12K for 2.5 Years, from back in the Days when closed loop was not even well known.

At about the 1 year mark I tried open loop with my 12K for several months because I wanted to give up the RS485 port that was monitor my batteries and instead use the Port for Solar Assistant.
Yep same sudden drops in SOC as you are showing. On top of that I had an EG4LL with three bad cells that would keep drifting out of balance. That added even more issues and more sudden drops.

The system worked fine in the Summer when at least once every three days I was guaranteed that the batteries would reach 100% SOC and keep charging until the Inverter was re-calibrated.
Once winter started to set in and full charges became rarer the SOC would drift by 20-30% upward. At that point only grid charging them back to 100% would re-calibrate the Inverters SOC reading for a few days until it would slowly drift again.

BTW I have investigated this for months and even the BMS SOC readings drift if the battery is not re-calibrated at intervals. The problem is compounded with an Inverter that is open loop mode because it cannot read the Individual packs which may vary wildly in SOC after several days of minimal charging. Also more packs in the system means greater variations.

The overall problem also increases as more batteries are added with the same amount of PV available.
At 16Kwh of Batteries with 10KWp of Solar it was not even an issue but now at 32Kw/h of batteries connected to the same 10KWp of Solar it requires a more strategic management system even in Closed loop mode.

Anyway the best solution is Closed loop and even then in the Winter some amount of Grid or Gen charging is needed to keep the SOC accurate.
Since you cannot use closed loop you need to keep the SOC margins higher in the TOU settings and then turn on the Generator at a higher SOC number so that the batteries keeps getting fully charged daily. That way you can confidently run them down to 20% at night.
There is always a reluctance to do this but Power is Power and the only downside is a bit of lost power in the conversion process. If the customer needs 60KWh of power per day and the PV is only providing 20KWh in the Winter then the other 40KWh is going to come from the Generator or Grid one way or the other.
Tapping up the batteries completely may mean a little extra Gen time but it sure as hell beats the inconvenience caused by not doing it.
The TOU slots are your friend and by using them and the Gen charge system properly you can get around this issue.

 

[Cmiller]

There is no scenario in which Voltage or Batt-% monitoring will give you accurate SOC readings with LFP batteries no matter which Inverter you use. It's always going to be drifting and typically it is reading a lot higher SOC than what is actually available. I have had a Sol-Ark 12K for 2.5 Years, from back in the Days when closed loop was not even well known.

At about the 1 year mark I tried open loop with my 12K for several months because I wanted to give up the RS485 port that was monitor my batteries and instead use the Port for Solar Assistant.
Yep same sudden drops in SOC as you are showing. On top of that I had an EG4LL with three bad cells that would keep drifting out of balance. That added even more issues and more sudden drops.

The system worked fine in the Summer when at least once every three days I was guaranteed that the batteries would reach 100% SOC and keep charging until the Inverter was re-calibrated.
Once winter started to set in and full charges became rarer the SOC would drift by 20-30% upward. At that point only grid charging them back to 100% would re-calibrate the Inverters SOC reading for a few days until it would slowly drift again.

BTW I have investigated this for months and even the BMS SOC readings drift if the battery is not re-calibrated at intervals. The problem is compounded with an Inverter that is open loop mode because it cannot read the Individual packs which may vary wildly in SOC after several days of minimal charging. Also more packs in the system means greater variations.

The overall problem also increases as more batteries are added with the same amount of PV available.
At 16Kwh of Batteries with 10KWp of Solar it was not even an issue but now at 32Kw/h of batteries connected to the same 10KWp of Solar it requires a more strategic management system even in Closed loop mode.

Anyway the best solution is Closed loop and even then in the Winter some amount of Grid or Gen charging is needed to keep the SOC accurate.
Since you cannot use closed loop you need to keep the SOC margins higher in the TOU settings and then turn on the Generator at a higher SOC number so that the batteries keeps getting fully charged daily. That way you can confidently run them down to 20% at night.
There is always a reluctance to do this but Power is Power and the only downside is a bit of lost power in the conversion process. If the customer needs 60KWh of power per day and the PV is only providing 20KWh in the Winter then the other 40KWh is going to come from the Generator or Grid one way or the other.
Tapping up the batteries completely may mean a little extra Gen time but it sure as hell beats the inconvenience caused by not doing it.
The TOU slots are your friend and by using them and the Gen charge system properly you can get around this issue.

I agree that after a few weeks of not getting fully charged you will definitely see drifting. Any time the batteries get completely fully charged whatever monitoring you are using will re-calibrate at the top.

But how do you think the batteries are tracking SOC internally?? They are doing the exact same thing that an external shunt based battery monitor does! Track power into vs power out of the batteries!

Differences between batteries DOES NOT make a difference in overall SOC! With your 32Kw/h of battery, if you take out 16Kw/h you are at 50%! Doesn't matter if half the batteries are at 55% and the other half are at 45%..... you are still at 50% SOC!! I have seen batteries with closed loop comms give the lowest SOC while discharging and the highest SOC while charging... that is a disaster! Creates all kinds of charge/discharge swings if you are using TOU to do partial discharge! The best way with closed loop comms is for the master battery to give the AVERAGE SOC to the Sol-Ark!

I personally have 30.7Kw/h of battery bank at home and my Victron battery monitor stays within 5-10% unless the batteries don't get full for more than probably 2 weeks. After that it starts running off further. A good battery monitor is designed to figure in all kinds of stuff like Peukert Coefficient, charge efficiency, temperature compensation, etc. If the settings on the backend are configured correctly, and if the battery monitor measures small enough fractional current you get a pretty good end result! The Sol-Ark has a built-in shunt that is supposed to do all of this.

If my Victron battery monitor can do it with a shunt...... Then why can't a Sol-Ark do it with a shunt??

Everyone keeps acting like it's impossible to track SOC on LifePo4 batteries with anything but using closed loop comms and listening to what the battery says..... This is not true!! We use other battery monitors ALL THE TIME with good success! And furthermore, how do these beloved BMS's with closed loop comms track the SOC? How are the BMS's any better than any other battery monitor?

And to be honest I have seen some pretty terrible SOC's on battery BMS's!! I have more confidence in a good quality external shunt based battery monitor than I do in a lot of battery BMS's!

There is a false hype on the whole closed loop comms being more accurate than other SOC monitoring devices!

Any power monitoring device is only as good as it's backend program is!! That includes any AIO inverter, battery monitor or BMS!

Also, I have side-by-side comparisons between Victron and Sol-Ark and the Sol-Ark runs way off compared to the Victron!! And please note that ultimately the only way to know if SOC is off is looking at measured SOC vs battery volts at the bottom or top end! On a good quality LifePo4 (16s) battery you will be around 15% SOC at 50.7 ish volts and at the top, once you get up to 56ish volts you are probably 99-100% full.

The issue with the Sol-Arks is that they run way too far off for the calculated SOC to be usable!! 5-10% is usable. Anything more than that and you have a disaster once you throw in gen start and if you try to do anything other than a flatline TOU number.



This was within a few days or maybe max 1 week of being completely fully charged! You simply cannot operate using TOU with this wide of a spread! This homeowner wants to have TOU set to 35%. With this wide of a spread, we would be at 10% when the Sol-Ark "thinks" we are at 35%.
We have a different issue on this unit that we were trying to figure out and yesterday afternoon, and Sol-Ark tech support changed a setting that caused the inverter to discharge (and sell battery power to grid) all the way down to our TOU setpoint of 35%. The battery voltage went down to 50.7V and triggered my Victron battery monitor to start the generator.... (that is voltage based, because voltage based is always, always, always a dependable start mechanism, even when SOC monitoring runs off!)

But the other thing is, the SOC should NEVER!!!! jump around by 50% or more unless the volts is at empty volts (we have it set at 48.9V) or full, which should be at like 56V.

I challenge anyone to tell me a good reason for SOC to jump around anytime other then bottom end/ battery empty volts or top end/ battery full volts!! Someone tell me one good reason!! If an algorithm does that, you are 100% guaranteed to have issues tracking the SOC!!

Here you can see side by side comparison of Victron vs Sol-Ark SOC tracking on the same system, this is a triple stack 15K Sol-Ark on 208/120V 3PH, with 45kWp of solar and 76.8 kW/h of battery.

Sol-Ark:



Victron:



The charge/discharge that you see is due to this being off grid with generator starting automatically overnight for running coolers etc.

This system is also where most of the screenshots come from in my original post. If you look at the Solar Assistant screenshots you will see just how drastic those jumps really are! And the Victron NEVER!!!!! I repeat NEVER!!!!!! EVER!!!! jumped anything remotely close to that!!!

 

[Cmiller]

At about the 1 year mark I tried open loop with my 12K for several months because I wanted to give up the RS485 port that was monitor my batteries and instead use the Port for Solar Assistant.

@robby Also, just note. You can use Solar Assistant at the same time as closed loop comms by using an ethernet/RJ45 splitter! I was doing that on the job with 3 15K's. I had Rubix batteries with closed loop comms, and a Solar Assistant for realtime monitoring. The reason I don't have closed loop comms running is because the BMS's gave lowest/highest SOC data when discharging/charging respectively. This simply doesn't work, as at times the switch from discharging to dropping loads to gen and charging would literally overload the 45kw genset, as there wasn't enough time for gen2 to start up and synchronize and connect in to share the loads.

Hmm.... I just now realized that another factor in that was the fact that the Sol-Ark CT circuit can't handle more than 40kw and anything over that going through the CT's results in inaccurate data..... therefore the Sol-arks didn't even know to throttle back charging in relation to the 36kw peak shave setpoint......

On this specific job, we are literally fighting 3 or 4 different inverter side glitches at the same time.... CT reading limits, gen start parameter limitations, inaccurate SOC monitoring. It gets a bit frustrating after a while, especially when the company (Sol-Ark) really doesn't try to resolve the issues! Or tries to say there aren't issues when the issues are right there in front of their noses!

Realtime monitoring is another thing that I wish Sol-Ark would improve on, as you simply cannot properly diagnose issues remotely without better than every 5 minutes another data point! But the Solar Assistant has done wonders for me on that job!

When you have a job that is 3 hours from the shop, you can't run out every time to diagnose when there are issues. And when that job happens to be a butcher shop...... well you can imagine what happens when there is hanging meat with no power to keep it cool. ???‍?

Again on this same job...... we have a Victron Cerbo GX and Smartshunt (to have remote access to a gen start relay that has an actual gen start menu with a whole host of parameter options.), a Solar Assistant (for realtime monitoring and logging of the inverters with remote access as well) and then we have the actual inverters themselves. All of the items accomplished by additional devices are supposedly built-in on the Sol-Arks..... but we achieve much better results with other brands of devices creating work-arounds.

And still we have un-dealt-with issues......

 

[robby]

Your post is way too long to reply to it fully but no one said that you cannot monitor a battery SOC without closed loop. We have always been saying that you CANNOT monitor it by Voltage or some percentage that the Inverter creates using the voltage. Yes a Smart Shunt works very well at doing it but the problem is that it's the overall value and not a full picture of each battery pack.

I have not been following Sol-Ark manuals for at least a year now but I do know that when I was playing around with Open Loop Comms they told me straight up that it's not going to be accurate and that it will drift and keep drifting until it is calibrated by a full charge.

BTW you do know that every time you go into the Menu and change the battery setting from batt-% to batt-V you are erasing all of the SOC data on the Battery in the Inverter.
I suspect this is why you are having this big of a problem. I learned that one the hard way and then Sol-Ark confirmed it.
If your using batt-% and move out of that mode to batt-V and save it then the Sol-Ark resets the internal Shunt Data in the Sol-Ark and it will drive you crazy. It takes the Sol-Ark a full charge cycle to 100% to get back the calibration data that was lost if you then go back into batt-% mode again.

I have seen batteries with closed loop comms give the lowest SOC while discharging and the highest SOC while charging...

Sol-Arks in closed loop mode use the Average SOC of all the Packs in the Comm loop.
It never uses the highest or lowest SOC.

I challenge anyone to tell me a good reason for SOC to jump around anytime other then bottom end/ battery empty volts or top end/ battery full volts!! Someone tell me one good reason!! If an algorithm does that, you are 100% guaranteed to have issues tracking the SOC!

Bad cells in one/some of the Packs. I have seen it dozens of times!
Pack will be reading 51V and then the Bad cells suddenly drop their voltage under a heavy load.
The pack voltage then dives down into the 49V region and adjacent packs start to dump current to charge up the pack that is low. The voltage must be balanced since they are all in parallel, so as the Inverter is drawing current the defective pack is still dropping and the good packs are compensating by powering the load while charging the bad pack. You can see the current flow going between packs using the BMS software or a clamp meter. Within minutes the whole bank can drop in SOC very quickly. YMMV depending on how many good packs you have connected to a bad one.

Also, just note. You can use Solar Assistant at the same time as closed loop comms by using an ethernet/RJ45 splitter! I was doing that on the job with 3 15K's. I had Rubix batteries with closed loop comms, and a Solar Assistant for realtime monitoring.

Thanks, but I think I was the first one on here to get that working over 18 months ago.
It was not possible until most of the battery companies did firmware updates and started to use Canbus communications. Now a splitter is an easy fix.

 

[Cmiller]

Sol-Arks in closed loop mode use the Average SOC of all the Packs in the Comm loop.
It never uses the highest or lowest SOC.

Well this is what I saw with my own two eyes....

The Sol-Ark only shows what the battery tells it! 9 batteries with closed loop comm, SOC% spread of up to 13-15% between highest and lowest, and every time the Sol-Ark went from charging to discharging or vice versa, the SOC on the Sol-Ark showed the highest or lowest. It was 100% battery related, because we got new firmware made for it and it fixed it. But along with that fix came a new problem..... the new firmware didn't give the Sol-Ark a combined charge and discharge amps, but only gave it the rate for a single battery. So as soon as gen power dropped out and the inverters were supplying around 9kw to the loads they overloaded due to the max discharge amps being at like 100A or something....

Sorry kinda got derailed again. Back to the point. The Sol-Ark's will show whatever the battery tells them. If the master battery tells them average SOC of the units in parralel, that is what it will show. But if the master battery shows low/high SOC depending on discharge/charge then that is what the Sol-Ark will show.

The Sol-Ark can only show what it is told, after all.

 

[Cmiller]

Bad cells in one/some of the Packs. I have seen it dozens of times!
Pack will be reading 51V and then the Bad cells suddenly drop their voltage under a heavy load.
The pack voltage then dives down into the 49V region and adjacent packs start to dump current to charge up the pack that is low. The voltage must be balanced since they are all in parallel, so as the Inverter is drawing current the defective pack is still dropping and the good packs are compensating by powering the load while charging the bad pack. You can see the current flow going between packs using the BMS software or a clamp meter. Within minutes the whole bank can drop in SOC very quickly. YMMV depending on how many good packs you have connected to a bad one.

So, you are suggesting that it is ok for SOC to jump from, say, 70% to 30% SOC in a split second. Then an hour or so later when the battery voltage gets up to the 56 volt ballpark to then jump up to 100%??

And what about the fact that the Victron battery monitor on the same system isn't doing that at all...??

Doesn't matter how you look at it, side by side comparison between the Sol-Ark and the Victron plus looking at where the battery voltage is at, shows us which device is working. Or should I say, which device is not working...

I don't think that your analogy of a cell diving and causing that battery to charge off the others is a scenario that can actually happen quite like you say! As soon as that voltage begins to drop the current will shift to the units with the higher voltage, therefore causing the faulty unit to have less current draw and therefore less voltage sag. Lithium batteries in parallel tend to float their voltages very evenly. If you see anything different you either have cabling issues, or you may have a BMS that is limiting current or something goofy like that! Voltage drop on individual cells, resulting in battery unit voltage drop will ALWAYS translate to current draw shifting to higher voltage units! That also means that as long as there is proper cabling and busbar sizing (both outside the batteries AND INSIDE them!!) you will only have a certain amount of deviation in SOC between batteries before the voltage difference will stop it from getting worse! I am not suggesting there can't be variations! I have many times seen 5-10% or even up to 15% difference. But more than that and I am questioning the accuracy of what the battery is saying!

Even if your scenario would be accurate, that is still no reason for SOC to drop! If the faulty unit is charging off the others, then that unit is accumulating charge to give off as well. Net energy in the battery bank is not changing, unless you are creating heat! When we talk about dropping SOC by 50% on let's say a 25kw bank of batteries, that would be a massive amount of heat being generated if what your are trying to say were true! And then furthermore.... where does the charge come from when the SOC bounces back up again??

You can explain it however you want to, but at the end of the day the Sol-Ark still has a problem and is unable to accurately track SOC!

NOTE* THIS WOULDN'T BE A PROBLEM IF SOL-ARK WOULDNT ADVERTISE THAT THEY CAN INDEED TRACK SOC!! In fact in their response email to me, they claimed that they indeed do track SOC accurately when the unit is in SOC mode. My experience screams differently!

To your noting that whenever you switch between voltage and SOC modes the data is reset. For starters, if that is how it works then that should probably be changed as well, so that's an additional problem. But then also, it seems that the longer that I have these systems in SOC mode the worse they get, and the more they jump around!

So just out of curiosity, what is so hard about saying "wow, Sol-Ark should work on these issue!", vs trying to explain away the issues? I am a bit confused about how people seem to think these Sol-Ark inverters are perfect with no flaws! Every inverter has it's flaws. Sol-Ark just happens to seemingly think they are exempt from that.... And by the way, I actually really like Sol-Arks! I was a big push in getting our company switched over to using a lot of Sol-Arks! I have said it before, but we have installed 100+ Sol-Arks. It's not that I don't like them. But I sure hate trying to come up with ways to explain to my customers why their inverter doesn't function as advertised!

 

[Cmiller]

Thanks, but I think I was the first one on here to get that working over 18 months ago.
It was not possible until most of the battery companies did firmware updates and started to use Canbus communications. Now a splitter is an easy fix.

That's great! If you posted on here about it, then I probably learned it from you! Thanks for that! I know when I was looking into that I read a bit on this forum, and most of what I learned came from here somewhere!

I am always open to learning new things. But I will also always refute what I know to be incorrect!

 

[robby]

Well this is what I saw with my own two eyes....

The Sol-Ark only shows what the battery tells it! 9 batteries with closed loop comm, SOC% spread of up to 13-15% between highest and lowest, and every time the Sol-Ark went from charging to discharging or vice versa, the SOC on the Sol-Ark showed the highest or lowest. It was 100% battery related, because we got new firmware made for it and it fixed it. But along with that fix came a new problem..... the new firmware didn't give the Sol-Ark a combined charge and discharge amps, but only gave it the rate for a single battery. So as soon as gen power dropped out and the inverters were supplying around 9kw to the loads they overloaded due to the max discharge amps being at like 100A or something....

Sorry kinda got derailed again. Back to the point. The Sol-Ark's will show whatever the battery tells them. If the master battery tells them average SOC of the units in parralel, that is what it will show. But if the master battery shows low/high SOC depending on discharge/charge then that is what the Sol-Ark will show.

The Sol-Ark can only show what it is told, after all.

It's not supposed to work that way.
This is how it normally works and it is the way I have seen other people's systems on here work.

Today is the fifth consecutive day of massive cloud cover so PV production is only getting the batteries from 20-60%.



Here are five of the Individual packs.
Note how the Internal BMS are already drifting apart.
Pack #5 was just installed last week and needs a firmware update, but I am waiting for a sunny day to do it.




Take the Average of (54+65+66+61+48) / 5) = 58.8%
This is what the BMS App is reading and this is What the Sol-Ark is Reading



I get Zero surprises with Sudden SOC drops while in closed loop, so long as I check once in awhile that the one EG4LL pack has not gone out of balance. The eFlex cells are always perfectly balanced because it uses an active balancer.

 

[robby]

So, you are suggesting that it is ok for SOC to jump from, say, 70% to 30% SOC in a split second. Then an hour or so later when the battery voltage gets up to the 56 volt ballpark to then jump up to 100%??

I am not telling you it's Okay. You said your running in Voltage Mode and I am saying to you that this mode is Useless when it comes to SOC accuracy with LFP batteries. The only way to keep track of SOC is by using a coulomb meter.

You can have an LFP battery Bank that is reading 52.40Vdc the inverter in V-Mode it will assume you have 60% SOC.
When you put a couple of hundred Amps of load on it the Voltage might drop from 52.40V down to 51.60V and the Inverter will now show 30% SOC. If you drop the load a few seconds later the battery could recover back to 52.30V and the Inverter would show 55% SOC.
None of these readings are accurate. The Inverter needs to know exactly how many Amps have gone into the battery and how many have been taken out. If this is not tracked then the SOC will always be messed up and jumping all over the place..

And what about the fact that the Victron battery monitor on the same system isn't doing that at all...??

Because it's a coulomb Meter.

Doesn't matter how you look at it, side by side comparison between the Sol-Ark and the Victron plus looking at where the battery voltage is at, shows us which device is working. Or should I say, which device is not working...

The Victron is a dedicated Shunt and is better at doing what it does than a Shunt built into an Inverter.
PLEASE NOTE: The Shunt in the Sol-Ark is not even in use if your in batt-V mode. So don't blame it for your situation.

Manufactures like the Sol-Ark tell you straight up to use Closed Loop and so do most other Inverter manufactures.
If you got plenty of PV or use Grid charging and Tap the batteries up to Full every 3-4 days then yes the Shunt in the Inverter will give a fairly accurate reading.

I don't think that your analogy of a cell diving and causing that battery to charge off the others is a scenario that can actually happen quite like you say! As soon as that voltage begins to drop the current will shift to the units with the higher voltage, therefore causing the faulty unit to have less current draw and therefore less voltage sag. Lithium batteries in parallel tend to float their voltages very evenly. If you see anything different you either have cabling issues, or you may have a BMS that is limiting current or something goofy like that! Voltage drop on individual cells, resulting in battery unit voltage drop will ALWAYS translate to current draw shifting to higher voltage units! That also means that as long as there is proper cabling and busbar sizing (both outside the batteries AND INSIDE them!!) you will only have a certain amount of deviation in SOC between batteries before the voltage difference will stop it from getting worse! I am not suggesting there can't be variations! I have many times seen 5-10% or even up to 15% difference. But more than that and I am questioning the accuracy of what the battery is saying!

Okay well since i have seen this multiple times I won't get into it.
I am talking about a pack with bad cells. So yes voltages drop off suddenly and current is of course being drawn from packs with a higher voltage and flowing into the pack with the bad cells and a lower voltage.

Even if your scenario would be accurate, that is still no reason for SOC to drop! If the faulty unit is charging off the others, then that unit is accumulating charge to give off as well. Net energy in the battery bank is not changing, unless you are creating heat! When we talk about dropping SOC by 50% on let's say a 25kw bank of batteries, that would be a massive amount of heat being generated if what your are trying to say were true! And then furthermore.... where does the charge come from when the SOC bounces back up again??

If any source with a higher voltage is presented to a battery it is going to start charging if it can hold more charge. The battery does not know the difference between Voltages coming from an SCC vs Another battery.

You can explain it however you want to, but at the end of the day the Sol-Ark still has a problem and is unable to accurately track SOC!

NOTE* THIS WOULDN'T BE A PROBLEM IF SOL-ARK WOULDNT ADVERTISE THAT THEY CAN INDEED TRACK SOC!! In fact in their response email to me, they claimed that they indeed do track SOC accurately when the unit is in SOC mode. My experience screams differently!

Accurately tracking SOC over what time period? Even you admit that drift is always happening. Sure Sol-Ark can Track SOC in batt-% mode accurately so long as every 3 days it gets a Full charge.
If it's partially charges then drift is going to occur. Good luck finding an Inverter that will handle it in Open Loop mode without this kind of drift.
Still not sure if your getting the point that once your in Batt-V mode you have turned off the Sol-Arks Shunt.

I did several posts on this subject and I would bet that you fell into the same Trap that myself and others fell into. You were probably Switching back and Forth between batt-V mode and batt-% mode and did not realize the SOC data in % mode had been erased. The Inverter just grabs a translation of what the Voltage is indicating when you go from Voltage to Percentage mode. So yes it looks like it's right info but the Coulomb counter has actually been reset. It's not your fault, it is because Sol-Ark does not document this properly and most of the new CS Reps don't even know that it happens.

To your noting that whenever you switch between voltage and SOC modes the data is reset. For starters, if that is how it works then that should probably be changed as well, so that's an additional problem. But then also, it seems that the longer that I have these systems in SOC mode the worse they get, and the more they jump around!

It might be nice if it alerted the user to the fact that the data will be lost but I can understand why they don't want it stored indefinitely.
Suppose two days later you decide to go back to batt-% mode and it just picks up the old SOC value.

So just out of curiosity, what is so hard about saying "wow, Sol-Ark should work on these issue!", vs trying to explain away the issues? I am a bit confused about how people seem to think these Sol-Ark inverters are perfect with no flaws! Every inverter has it's flaws. Sol-Ark just happens to seemingly think they are exempt from that.... And by the way, I actually really like Sol-Arks! I was a big push in getting our company switched over to using a lot of Sol-Arks! I have said it before, but we have installed 100+ Sol-Arks. It's not that I don't like them. But I sure hate trying to come up with ways to explain to my customers why their inverter doesn't function as advertised!

I have had zero issues with my Sol-Ark. There was a time when the 12K first came out that it had some small issues but those have been squashed a long time ago.
The problems that you think are bugs are not bugs. You still seem to think that running a Sol-Ark with voltage control and LFP batteries can work properly.
I am trying to explain to you that it cannot work properly with LFP batteries.
You will have the same issues you are experiencing no matter which Inverter you use.

 

[Cmiller]

I am not telling you it's Okay. You said your running in Voltage Mode and I am saying to you that this mode is Useless when it comes to SOC accuracy with LFP batteries. The only way to keep track of SOC is by using a coulomb meter.

You can have an LFP battery Bank that is reading 52.40Vdc the inverter in V-Mode it will assume you have 60% SOC.
When you put a couple of hundred Amps of load on it the Voltage might drop from 52.40V down to 51.60V and the Inverter will now show 30% SOC. If you drop the load a few seconds later the battery could recover back to 52.30V and the Inverter would show 55% SOC.
None of these readings are accurate. The Inverter needs to know exactly how many Amps have gone into the battery and how many have been taken out. If this is not tracked then the SOC will always be messed up and jumping all over the place..

Because it's a coulomb Meter.

The Victron is a dedicated Shunt and is better at doing what it does than a Shunt built into an Inverter.
PLEASE NOTE: The Shunt in the Sol-Ark is not even in use if your in batt-V mode. So don't blame it for your situation.

Manufactures like the Sol-Ark tell you straight up to use Closed Loop and so do most other Inverter manufactures.
If you got plenty of PV or use Grid charging and Tap the batteries up to Full every 3-4 days then yes the Shunt in the Inverter will give a fairly accurate reading.

Okay well since i have seen this multiple times I won't get into it.
I am talking about a pack with bad cells. So yes voltages drop off suddenly and current is of course being drawn from packs with a higher voltage and flowing into the pack with the bad cells and a lower voltage.

If any source with a higher voltage is presented to a battery it is going to start charging if it can hold more charge. The battery does not know the difference between Voltages coming from an SCC vs Another battery.

Accurately tracking SOC over what time period? Even you admit that drift is always happening. Sure Sol-Ark can Track SOC in batt-% mode accurately so long as every 3 days it gets a Full charge.
If it's partially charges then drift is going to occur. Good luck finding an Inverter that will handle it in Open Loop mode without this kind of drift.
Still not sure if your getting the point that once your in Batt-V mode you have turned off the Sol-Arks Shunt.

I did several posts on this subject and I would bet that you fell into the same Trap that myself and others fell into. You were probably Switching back and Forth between batt-V mode and batt-% mode and did not realize the SOC data in % mode had been erased. The Inverter just grabs a translation of what the Voltage is indicating when you go from Voltage to Percentage mode. So yes it looks like it's right info but the Coulomb counter has actually been reset. It's not your fault, it is because Sol-Ark does not document this properly and most of the new CS Reps don't even know that it happens.

It might be nice if it alerted the user to the fact that the data will be lost but I can understand why they don't want it stored indefinitely.
Suppose two days later you decide to go back to batt-% mode and it just picks up the old SOC value.

I have had zero issues with my Sol-Ark. There was a time when the 12K first came out that it had some small issues but those have been squashed a long time ago.
The problems that you think are bugs are not bugs. You still seem to think that running a Sol-Ark with voltage control and LFP batteries can work properly.
I am trying to explain to you that it cannot work properly with LFP batteries.
You will have the same issues you are experiencing no matter which Inverter you use.

No, these systems are NOT IN VOLTAGE MODE! I understand fully that in voltage mode it doesn't track accurately! If I were the one programming the backend, I would design my code in a way that the Sol-Ark keeps on tracking SOC via the shunt.... or columb counter as you say it (probably the proper way, really, I will admit), because after all, that's the only point if the shunt. Right? Unless there is a different reason for having a shunt..... but I'm actually completely sure there isn't!

And they could, like Victron does, have a setting that allows you to manually set the current SOC, so you can adjust when you first start up, or add batteries, or swap out to new batteries, etc.

But really though all I'm asking for is that the Sol-Ark at least tracks accurately. (When in SOC mode, that is.... even if it doesn't in voltage mode, I'm fine with just a working SOC mode!)

So.... these issues we are seeing. They are on Sol-Arks running in SOC mode! And the mode IS NOT being switched around every so often..... and a side by side comparison between Sol-Ark SOC and Victron SOC shows a terrible inconsistency in the Sol-Ark! Also, the Victron (as well as other brands such as Schneider and Midnite) stays relative well on track out to about 2 weeks or so from not getting fully charged, then begins to run off. So the Sol-Ark is very inferior.... and yes I'm talking about when it is in SOC mode!

And another note. You have a 12K and most of our issues are actually on the 15K. I don't know why the algorithm would be different in the 15K, but it sure seems like it is! We don't have many issues on the 12K's, but then again we don't have many 12K's where we actually using TOU, so maybe that is a factor as well.

On our one job we had a 12K working quite well, actually, in SOC mode and using TOU. Then we upgraded the system to a 15K and everything went south with the exact same settings.... do you have an explanation for that?

Another note...... again...... these units are running in SOC mode..... and the longer they are left in SOC mode, the worse they seem to get.......

Again..... these units are in SOC mode.....

I think you will get the point now...

Sorry if I didn't make this clear to begin with, I thought I had.....

 

[Cmiller]

I am not telling you it's Okay. You said your running in Voltage Mode and I am saying to you that this mode is Useless when it comes to SOC accuracy with LFP batteries. The only way to keep track of SOC is by using a coulomb meter.

You can have an LFP battery Bank that is reading 52.40Vdc the inverter in V-Mode it will assume you have 60% SOC.
When you put a couple of hundred Amps of load on it the Voltage might drop from 52.40V down to 51.60V and the Inverter will now show 30% SOC. If you drop the load a few seconds later the battery could recover back to 52.30V and the Inverter would show 55% SOC.
None of these readings are accurate. The Inverter needs to know exactly how many Amps have gone into the battery and how many have been taken out. If this is not tracked then the SOC will always be messed up and jumping all over the place..

Because it's a coulomb Meter.

The Victron is a dedicated Shunt and is better at doing what it does than a Shunt built into an Inverter.
PLEASE NOTE: The Shunt in the Sol-Ark is not even in use if your in batt-V mode. So don't blame it for your situation.

Manufactures like the Sol-Ark tell you straight up to use Closed Loop and so do most other Inverter manufactures.
If you got plenty of PV or use Grid charging and Tap the batteries up to Full every 3-4 days then yes the Shunt in the Inverter will give a fairly accurate reading.

Okay well since i have seen this multiple times I won't get into it.
I am talking about a pack with bad cells. So yes voltages drop off suddenly and current is of course being drawn from packs with a higher voltage and flowing into the pack with the bad cells and a lower voltage.

If any source with a higher voltage is presented to a battery it is going to start charging if it can hold more charge. The battery does not know the difference between Voltages coming from an SCC vs Another battery.

Accurately tracking SOC over what time period? Even you admit that drift is always happening. Sure Sol-Ark can Track SOC in batt-% mode accurately so long as every 3 days it gets a Full charge.
If it's partially charges then drift is going to occur. Good luck finding an Inverter that will handle it in Open Loop mode without this kind of drift.
Still not sure if your getting the point that once your in Batt-V mode you have turned off the Sol-Arks Shunt.

I did several posts on this subject and I would bet that you fell into the same Trap that myself and others fell into. You were probably Switching back and Forth between batt-V mode and batt-% mode and did not realize the SOC data in % mode had been erased. The Inverter just grabs a translation of what the Voltage is indicating when you go from Voltage to Percentage mode. So yes it looks like it's right info but the Coulomb counter has actually been reset. It's not your fault, it is because Sol-Ark does not document this properly and most of the new CS Reps don't even know that it happens.

It might be nice if it alerted the user to the fact that the data will be lost but I can understand why they don't want it stored indefinitely.
Suppose two days later you decide to go back to batt-% mode and it just picks up the old SOC value.

I have had zero issues with my Sol-Ark. There was a time when the 12K first came out that it had some small issues but those have been squashed a long time ago.
The problems that you think are bugs are not bugs. You still seem to think that running a Sol-Ark with voltage control and LFP batteries can work properly.
I am trying to explain to you that it cannot work properly with LFP batteries.
You will have the same issues you are experiencing no matter which Inverter you use.

We switched our one job to voltage mode because SOC mode ran so far off that gen start literally didn't work. The battery would run completely flat dead with the TOU set to 50%. Voltage mode was our last ditch effort to at least keep the batteries from going completely flat dead!

2 days ago we swapped out the customer's batteries for different ones with closed comms, because we got no help from Sol-Ark on any kind of fix for SOC tracking in SOC mode!

We ran that system in SOC mode for probably 2 months after swapping to the 15K and it seemingly just got worse and worse. And again, this job had a functioning 12K before, and we set the setting identical with the new 15K!

 

[Cmiller]

It's not supposed to work that way.
This is how it normally works and it is the way I have seen other people's systems on here work.

Today is the fifth consecutive day of massive cloud cover so PV production is only getting the batteries from 20-60%.

View attachment 182743

Here are five of the Individual packs.
Note how the Internal BMS are already drifting apart.
Pack #5 was just installed last week and needs a firmware update, but I am waiting for a sunny day to do it.
View attachment 182745
View attachment 182746


Take the Average of (54+65+66+61+48) / 5) = 58.8%
This is what the BMS App is reading and this is What the Sol-Ark is Reading

View attachment 182747

I get Zero surprises with Sudden SOC drops while in closed loop, so long as I check once in awhile that the one EG4LL pack has not gone out of balance. The eFlex cells are always perfectly balanced because it uses an active balancer.

From what I learned, I believe the reason your Sol-Ark is showing that 58%, (which like you showed is the average SOC%) is because the "master" that is plugged into the Sol-Ark is getting the net average and "telling" that to the Sol-Ark.

I would be curious though to know if that makes sense in your scenario? Or perhaps you have a different communication setup..?

The way our batteries are set for communication is they get daisy chained battery to battery via the RS485 ports (standard RJ45/ethernet plugs, 2 ports on each battery) and then from battery 01 you plug from canbus port to Sol-Ark.

Is that the way yours are set up as well? I am mostly curious because of not having worked with very many different brands of batteries that have closed loop comms options.

PS. Active balancers are an absolute must-have in my opinion! I have seen so many issues related to imbalance, because unless your cells are of exceptional quality you will always end up with imbalances long term!

 

[robby]

From what I learned, I believe the reason your Sol-Ark is showing that 58%, (which like you showed is the average SOC%) is because the "master" that is plugged into the Sol-Ark is getting the net average and "telling" that to the Sol-Ark.

I don't know for sure but that is probably right. I know that people with EG4 Stuff and SOK feed the Inverter the Average value also.

I would be curious though to know if that makes sense in your scenario? Or perhaps you have a different communication setup..?

I use the Same Daisy Chain of Batteries and then into the Sol-Ark via Canbus.

The way our batteries are set for communication is they get daisy chained battery to battery via the RS485 ports (standard RJ45/ethernet plugs, 2 ports on each battery) and then from battery 01 you plug from canbus port to Sol-Ark.

Same here.

Is that the way yours are set up as well? I am mostly curious because of not having worked with very many different brands of batteries that have closed loop comms options.

PS. Active balancers are an absolute must-have in my opinion! I have seen so many issues related to imbalance, because unless your cells are of exceptional quality you will always end up with imbalances long term!

I agree that they are a must. The one EG4LL that I have has a passive balancing system and cell issues arise from time to time where as the eFlex's cells are always balanced in each pack.
Right now after my 6th straight day of cloudy and rainy weather I suspect that the one EG4 battery I have is going to need to be manually charged up. The problem with that is that I have to charge all 32KWh of batteries to 100% just to fix the one EG4LL. Today the PV has produced a measly 13KWh of PV ?

 

[Cmiller]

I agree that they are a must. The one EG4LL that I have has a passive balancing system and cell issues arise from time to time where as the eFlex's cells are always balanced in each pack.
Right now after my 6th straight day of cloudy and rainy weather I suspect that the one EG4 battery I have is going to need to be manually charged up. The problem with that is that I have to charge all 32KWh of batteries to 100% just to fix the one EG4LL. Today the PV has produced a measly 13KWh of PV ?

Yeah, that's a bummer! Unfortunately the batteries with active balancers can only "make up for" one without for so long.

You could add a balancer into that battery, although I'm sure that would void any warranty that you might have on it.

 

[Cmiller]

Sol-Ark tech support says: batteries must be charged to full every 3 days, or SOC algorithm will run off. They recommend charging to full every 2 days.....

So apparently these units are not practical for off-grid running in SOC mode without a battery that has closed loop comms.....

I mean here in Ohio there are times we don't get full for 2 weeks during the winter. This would mean in order to have accurate SOC tracking we would need to run a gen every 3 days and complete a full charge.

I understand and am fine with some drift over extended periods of time, but 3 days??

Sounds pretty absurd to me, especially with seeing side-by-side what other brands can do!

 

[robby]

Sol-Ark tech support says: batteries must be charged to full every 3 days, or SOC algorithm will run off. They recommend charging to full every 2 days.....

So apparently these units are not practical for off-grid running in SOC mode without a battery that has closed loop comms.....

I mean here in Ohio there are times we don't get full for 2 weeks during the winter. This would mean in order to have accurate SOC tracking we would need to run a gen every 3 days and complete a full charge.

I understand and am fine with some drift over extended periods of time, but 3 days??

Sounds pretty absurd to me, especially with seeing side-by-side what other brands can do!

Any company that says it's inverter will stay calibrated more than three days in open loop mode is probably just trying to create sales or is using very conservative DOD amounts in their calculations or is basing it on an ideal installation. There is just no way for the Inverter can know the state of the Individual packs!

Unless you got a perfect Install and perfectly matched packs so that all the packs are charging and discharging evenly the packs will start to drift apart quickly under partial charging.
Look at any of Will's videos where he measure the Amps going to the individual Packs and you will see numbers like 18A into one pack and a low of like 16A into another pack. By the end of a partial charge the pack that got 18A may be at 70% SOC while the one that got 16A will be at around 65% SOC.
The next night you drain the packs down to 20% according to the Inverter and the 18A one will actually be above 20% and the 16A one is actually nearer to 15%.
Repeat that same amount of charge the next day and the first one will be closer to 72% and but the second pack is now closer to 60%.
You can see where this is going and the difference between packs is going to show up in the coming days as a sudden drop in SOC as the lowest batteries start to drop the combined voltage.
People who run with Wide Margins won't have this issue as quickly but the more you try to squeeze power out of the packs is the quicker it will happen.

In closed loop this does not happen because it uses the Average of the True SOC of each pack.