Formula for unreliable SOC on Seplos V3 and possibly other BMS

[dusball]

After having my Seplos BMS V 3 now for a couple of months - I've determined the SOC calculation the system gives is completely unreliable. Since I feed my data into HomeAssistant I've developed a short equation to give a better indication of state of charge based upon batter voltage. There's probably a better one out there available but until find one that uses more of a logarithmic calculation for the end curves this will have to do. Since the discharge and charge curve is relatively flat a formal is easy to extrapolate for the flat areas of the curve. It may could be tweaked but this give a pretty close approximation if you look at the graphs others have provided. Sharing here in case others have the same frustration as me.

Solve for Y to get state of charge with Voltage- Solve for X to get Voltage from SOC,


for state of charge between 0 and 14%

0-14

-9.5x + 0.5y + 23.75 = 0

for state of charge between 15 and 99%

15-99

50x - 0.1y - 158.25 = 0

Example X(soc) =( (50 * Voltage) + 158.25 ) / .01

for state of charge between 99 and 100%

99​

x - 0.275y + 23.85 = 0

Not perfect and if you're using HA- you'll want to use filters "limits" on the end points to ensure it doesn't go past 0% and 100%

My Value Template for HA:

{{(((float(states.sensor.seplos_bms_0_pack_voltage.state) / 16 * 50.0 ) - 158.25 ) / 0.1) | round(2) }}

 

[sunshine_eggo]

Voltage is a demonstrably worse means of determining SoC when compared to coulomb counting. This is compounded by the influence of current and direction on voltage, which will induce a notable SoC error.

On what basis do you find this more accurate than the BMS reported SoC?

 

[dusball]

I understand and completely agree, but when my BMS is showing 78% SOC charge and the voltage is 51.2V - as it was this morning -there's a problem. its not easy to coulomb count on a live system that constantly going up and down coulombically, That is probably why the built in systems dont do a good job either. The coulomb formula is based on time and starting points of charge times current -+. At 51.2 volts I can guarantee I dont have 78% left. I need a better tool to tell me where I am- and this may be simplistic but its better than what I have and SOC is somewhat subjective anyway.

 

[Quattrohead]

At 51.2 volts I can guarantee I dont have 78% left. I need a better tool to tell me where I am- and this may be simplistic but its better than and what I have and its all subjective anyway.

Sounds good to me.

 

[marionw]

The "Best Better Tool" might be an accurate shunt

 

[ksmithaz1]

Battery voltage changes based on SOC and load. The higher the load the lower the voltage at a given SOC. Unfortunately this is not going to be easily compensated for, and would in fact require some 'learning' on the part of the BMC because the internal resistance of the batteries is extremely low thus even small variations of the chemistry create significant variations in the readings. I have no idea how the BMS's do the maths, but it should be monitoring load/time and should have a baseline based on averages.

 

[bolhuijo]

It's kinda frustrating because the BMS has a shunt and seems to be coulomb counting already with the "TotalCap" and "RemainCap" numbers and yet the SOC number gets worse and worse between full charges

 

[sunshine_eggo]

It's kinda frustrating because the BMS has a shunt and seems to be coulomb counting already with the "TotalCap" and "RemainCap" numbers and yet the SOC number gets worse and worse between full charges.

Hence the need for regular full charges to re-sync the "shunt" to 100%. Victron, Bogart, etc., states this is required to retain SoC accuracy with their battery monitors.

If someone isn't fully charging to 100% regularly, they're forcing the necessity for manual adjustments. I would still hesitate to trust results from voltage measurements in an active system.

I have the luscious situation where I can't ever sync my shunt because I don't charge to 100% but more than 1-2 times per year, and that's for the purpose of a capacity test. THe upside is my chemistry has a very strong voltage to SoC correlation. I'm regularly resetting my Batrium SoC to 75% when my NMC hits 3.92V/cell. It wanders about 5%/month - usually reading high due to it not factoring in any burn-off from balancing. My BMV-702 tends to wander even more.

 

[bolhuijo]

Hence the need for regular full charges to re-sync the "shunt" to 100%. Victron, Bogart, etc., states this is required to retain SoC accuracy with their battery monitors.

If someone isn't fully charging to 100% regularly, they're forcing the necessity for manual adjustments. I would still hesitate to trust results from voltage measurements in an active system.

I have the luscious situation where I can't ever sync my shunt because I don't charge to 100% but more than 1-2 times per year, and that's for the purpose of a capacity test. THe upside is my chemistry has a very strong voltage to SoC correlation. I'm regularly resetting my Batrium SoC to 75% when my NMC hits 3.92V/cell. It wanders about 5%/month - usually reading high due to it not factoring in any burn-off from balancing. My BMV-702 tends to wander even more.

Sure but this seems to get wildly far off in a couple days, right after doing the official 100-to-zero calibration maneuver.

 

[sunshine_eggo]

Sure but this seems to get wildly far off in a couple days, right after doing the official 100-to-zero calibration maneuver.

Hmmm.... Okay. Maybe it just sucks worst than most...

 

[Bob B]

Sure but this seems to get wildly far off in a couple days, right after doing the official 100-to-zero calibration maneuver.

Are the current readings accurate?

I have a JBD BMS that wasn't coulomb tracking SOC correctly until I calibrated the charge and discharge current readings.

 

[bolhuijo]

Are the current readings accurate?

I have a JBD BMS that wasn't coulomb tracking SOC correctly until I calibrated the charge and discharge current readings.

That's not a bad thing to look into... I see there is a calibration function for current in the bms-studio software.

 

[bolhuijo]

So, what to use as a calibrated current supply or load? I ain't got no fancy power supply capable of accurate current measurement in the 50 volt range.

 

[Bob B]

You are going to need to calibrate both charge and load ...... Mine was only a 12V so I used a cheap power supply for the charge .... just need a steady current and and accurate way to measure it .... a good clamp meter or better yet, put a meter in series with the load. Don't rely on the current reading from a cheap power supply.

 

[dusball]

Looking into the current calibration is a good idea- but since the battery its in use, supplying the house- I have the same dilemma on how to get a accurate steady current to work off of.

maybe a generator at night for the plus side?
and a late night period when the house is calm on the draw side?

 

[kommando]

For the JK PB BMS they state calibration is required, recommend Voltage is done at rest and current under a constant charge. It does improve SOC but does not change the constant current reading changes the BMS shows at very low charge and discharge currents.