LV5048 Coulomb counting ?

[AI6MK]

System: Grid tied LV5048 / 2x EG4 LiFePO4 / 16 panel GND MOUNT / SOLAR ASSISTANT monitoring.
Use: WELL PUMP + BARN.

Works great and on sunny/partially cloudy days, recharging the BATTERIES and WATER TANK well before the sun goes down from solar alone. BUT how does it work on overcast days...not so good.

My expectation was that on overcast days with minimal solar power, once the batteries dropped to a low SOC the utility would kick in to restore them back to 100%, by setting PARAM[12]=51V and PARAM[13]=FUL. 51V is the MAX value for PARAM[12].

The issue seems to be that SOLAR ASSISTANT uses "coulomb counting" and LV5048 uses VBATT to determine SOC. Today, the batteries were at <20% as reported by SOLAR ASSISTANT, but LV5048 did not switch charging to UTILITY, because VBATT was > 51V. The only way I can force the BATTERIES to get charged from the UTILITY is to change PARAM[1] from SBU to UTI and turning off the SOLAR PANELS.

So the problem is that for LiFePO4 chemistry measuring voltage, as happens in the LV5048, is a inaccurate way to determine the SOC.

Short of buying an all-in-one which uses coulomb counting to determine SOC, is there any remedy for this on the venerable LV5048 ?

 

[OffGridForGood]

My expectation was that on overcast days with minimal solar power, once the batteries dropped to a low SOC the utility would kick in to restore them back to 100%, by setting PARAM[12]=51V and PARAM[13]=FUL. 51V is the MAX value for PARAM[12].

The issue seems to be that SOLAR ASSISTANT uses "coulomb counting"

I don't believe so, SA doesn't "use" anything, it reports what it is being told - either by the inverter, or the battery pack or by a shunt.
My own experience was not great until I bought a Victron 500A shunt and programed SA to read the SOC from the shunt. Since that day, I am super happy with the accuracy of the SOC reported.

Once you have the shunt installed, you can program SA to switch over the charging at any SOC you like.