Sol Ark 12K SOC estimation

[629658]

If we can get some of our resident computer experts to help us get this working it would be a win win.

 

[mvonw]

I wonder if that's the problem with my EG4 LiFePOwer4s too. I've noticed watching them that unless two amps or more is flowing into or out of the battery it doesn't report as charging or discharging. That would also explain why the inverter estimate is better. Whenever my wife is gone my average use is less than the 500+ watts I draw before I get to two amps.

Yeah, that’s what I thought after I read your post and talked to Trophy. If you just use one or two batteries in parallel, it probably wouldn’t happen, because each battery would be over the 2 amp threshold, but in a larger bank, the draw is more dispersed. I’ve had the Trophy’s for a year now, and it takes that long to really assess a product. These “smart” batteries really aren’t all that smart, it turns out.

 

[mvonw]

I have Solar Assistant reading my Trophy BMS. Seems very accurate. I have a Victron shunt also. It correlates within 1% the BMS data and then I have my Sol Ark data. During December and into January I see my Sol Ark off as much as 10% almost always too high. When we have consistent sunny days not usually an issue. During multiple cloudy days it becomes an issue. I spoke with Sol Ark yesterday in fact and my thoughts are if the Victron shunt can track it accurately then the Sol Ark should be able to also. My next thought is how can i interface my Victron shunt to my Sol Ark for SOC. That would pretty much solve any issues no matter which battery brand is attached to the Sol Ark. This would solve a multitude of battery and BMS issues and the Sol Ark.

Manually enter in your correct charge parameters and have the Victron supply the SOC data to the Sol Ark. Problems solved.

This issue really on seems to manifest if you have prolonged, low draw (under a certain threshold - 2A in my case). All the Trophy’s use different BMS models, I believe so they many not all behave as mine. It’s a know issue to Trophy. Hopefully they can fix it with firmware or settings. I have several other acknowledged issues with these batteries, they are not perfect in real world use. Hopefully they can get addressed. Dan is very responsive, but he clearly has severe limits on support.

I’m definitely a beta tester, which I hate, but I knew that buying a new product (had them over a year now).

 

[Ampster]

If we can get some of our resident computer experts to help us get this working it would be a win win.

Initially my solution was to use Solar Assistant and the emulate BMS option. Now I have closed loop communication with an Orion BMS JR2. I have been using Orion BMSs for eight years and have confidence in their Coulomb counter. I love a lot of things about the SolArk but someone else can try to correct that one weakness.

 

[629658]

Yeah, that’s what I thought after I read your post and talked to Trophy. If you just use one or two batteries in parallel, it probably wouldn’t happen, because each battery would be over the 2 amp threshold, but in a larger bank, the draw is more dispersed. I’ve had the Trophy’s for a year now, and it takes that long to really assess a product. These “smart” batteries really aren’t all that smart, it turns out.

100% agree. They make too much fuss about open/closed loop communication. It’s not that hard. IMO in the future BMS will be connected on a LAN where the BMS data can be retrieved by whatever device asks for it making setup as easy as entering in an IP address for the BMS data.

 

[629658]

I too have been very happy with my Trophy batteries. They have functioned perfectly. Would I like closed comms yes but it is manageable. The batteries themselves are well constructed, and perform well with a reasonable cost. When I installed my system I added a Victron shunt and it just gives me another source of confirmation

 

[mvonw]

100% agree. They make too much fuss about open/closed loop communication. It’s not that hard. IMO in the future BMS will be connected on a LAN where the BMS data can be retrieved by whatever device asks for it making setup as easy as entering in an IP address for the BMS data.

Oh yeah, now your’re talking. Or better yet just cell modules you can plug in with a centralized BMS/bus backbone. Why pay for a BMS each time you add a new battery….

 

[mvonw]

I have Solar Assistant reading my Trophy BMS. Seems very accurate. I have a Victron shunt also. It correlates within 1% the BMS data and then I have my Sol Ark data. During December and into January I see my Sol Ark off as much as 10% almost always too high. When we have consistent sunny days not usually an issue. During multiple cloudy days it becomes an issue. I spoke with Sol Ark yesterday in fact and my thoughts are if the Victron shunt can track it accurately then the Sol Ark should be able to also. My next thought is how can i interface my Victron shunt to my Sol Ark for SOC. That would pretty much solve any issues no matter which battery brand is attached to the Sol Ark. This would solve a multitude of battery and BMS issues and the Sol Ark.

Manually enter in your correct charge parameters and have the Victron supply the SOC data to the Sol Ark. Problems solved.

So, it's your experience that the Victron shunt is more accurate than the built-in sol ark shunt?

 

[629658]

So, it's your experience that the Victron shunt is more accurate than the built-in sol ark shunt?

100% I was just relaying this to Sol Ark yesterday. Stupid that a $130 shunt is more accurate but once you get the shunt settings right it’s very accurate. I’m now trying to figure out a way to get the shunt attached to the Sol Ark for SOC data. That would be almost perfect or even better get the Solar Assistant to provide info to the Sol Ark.

 

[Ampster]

100% I was just relaying this to Sol Ark yesterday. Stupid that a $130 shunt is more accurate but once you get the shunt settings right it’s very accurate

It might be the algorithm in the SolArk which is the source of the difference. As i mentioned above I found the Solar Assistant algorithm more accurate.. Presumably Solar Assistant uses the same SolArk shunt readings.

 

[mvonw]

100% I was just relaying this to Sol Ark yesterday. Stupid that a $130 shunt is more accurate but once you get the shunt settings right it’s very accurate. I’m now trying to figure out a way to get the shunt attached to the Sol Ark for SOC data. That would be almost perfect or even better get the Solar Assistant to provide info to the Sol Ark.

Nice. I'm going to buy one. It's surprising the sol ark is so different, especially given the price..

 

[629658]

Nice. I'm going to buy one. It's surprising the sol ark is so different, especially given the price..

Amazon
Victron Energy SmartShunt 500 amp Battery Monitor (Bluetooth) https://a.co/d/798gQ6N

I put mine inside a blue sea enclosure.

 

[629658]

Sol Ark recommended that I go into battery settings and change my battery pack size by one up or down to force it to relearn over a 3 day period. I think it’s a short coming but manageable once you know. The nice thing about the shunt is if you have home made or batteries from different manufacturers you can still get an accurate SOC.

 

[mvonw]

Amazon
Victron Energy SmartShunt 500 amp Battery Monitor (Bluetooth) https://a.co/d/798gQ6N

I put mine inside a blue sea enclosure.

I'll copy you. I would need to get their cerbo gx so I can remotely monitor it..

 

[629658]

I'll copy you. I would need to get their cerbo gx so I can remotely monitor it..

I haven’t started to work on it yet but apparently you can install Venus OS on a Pi and plug the shunt into the Pi with this cord

Victron Energy VE.Direct to USB Interface https://a.co/d/3VVakSG

 

[robby]

I don't have confidence in the SOC estimates from my SolArk..

I agree! I was told that it is very complicated to make an accurate algorithm that can hold true for multiple battery packs that are not charging or discharging equally. Lots of variable involved including the temperature when the packs are being charged and discharged. It does do a good job in the summer when my packs all get fully charged and then it is dreadfully inaccurate in the Winter when my packs are only getting to 50-80% SOC daily.
In closed loop mode it's pretty good all the time because each pack is reporting it's own SOC and the Sol-Ark is using the average value.

 

[Ampster]

I was fortunate to get communications working with the SolArk and my Orion JR2 BMS which I reported on in another thread. Now I can use the Battery % values from the BMS for my TOU settings. i am forcing a grid charge to 100% every day so that helps the accuracy.

 

[mvonw]

I agree! I was told that it is very complicated to make an accurate algorithm that can hold true for multiple battery packs that are not charging or discharging equally. Lots of variable involved including the temperature when the packs are being charged and discharged. It does do a good job in the summer when my packs all get fully charged and then it is dreadfully inaccurate in the Winter when my packs are only getting to 50-80% SOC daily.
In closed loop mode it's pretty good all the time because each pack is reporting it's own SOC and the Sol-Ark is using the average value.

I’m probably grossly under-appreciating the complexity of this, but from the perspective of the inverter, doesn’t a bank of of batteries, no matter how heterogeneous, just present to the inverter as source of energy at a certain voltage and amperage for given lengths of time with a bank of a known amp-hour capacity. I get there is inefficiencies and loss in the system and it will never be 100%, but the inverter should know exactly what goes in and out of the ‘battery’.

I would love to see the logic of these algorithms published, if only to have a better sense of what is factored in and how complex it is. Sometimes engineers use ‘complex’ as a euphemism for ‘we don’t really understand’ how to do it right now’ ..

 

[629658]

I would conditionally agree that the AIO inverters should be better at tracking SOC and if you have closed loop comms they do. Shunt technologies are not new so it can be done. In the meantime it’s an area that certainly needs improvement but is a very manageable problem once you understand what is going on.

Just remember you cannot trust the SOC information provided by your inverter. You need another accurate method of SOC measurement for verification and troubleshooting either BMS or a shunt.

 

[Ampster]

the inverter should know exactly what goes in and out of the ‘battery’.

I agree any Frankenpack will present as a load or source of energy to the inverter. I wonder if the complexity the Engineer was referring to was the algorithm to estimate the State of Charge.of such a pack.