How to fix SOC% in Victron PV + quattro being so inaccurate as to be unable to manage Li battery

Cerbo, Quattro, RS450/100 - 6kW PV, cerbo, 4x 100Ah Smart Li.
OK, I have discovered (although I am pretty sure it’s not in the manuals or the webinars I have seen) that the Quattro/Multi charging current is a best guess of the AC I/O. I have no idea if the RS450/100 I have is any better as I have never seen a spec given for that either.
It’s implied that the cerbo receives current and voltage data from the two devices. It could of course also receive integrated charging current/energy (eg ave per sec/minute) to reduce bus bandwidth. Sadly, the information from the Quattro is apparently too crude to give a viably accurate SOC.
Both PV and Quattro change SOC.
DVCC enabled everything else in DVCC is disabled (as manual).
Installing 2 smartshunts would be physically very challenging.
The really obvious way to control charging and display SOC% (given the existing method is unusable) would be off the battery voltage. Li have a particularly reliable voltage vs SOC correlation that would be more than good enough for control, but rather surprisingly this is not implemented for this purpose, pity.
So what do I want to do?
Winter setting would be to charge to 80% SOC (this figure may change if I have an accurate SOC available) from 01:00 for 5 hours; Sundays to 100%. No grid usage between 14:30 for 6 h. Yes, I am on Octopus Agile!. Summertime I am off grid so hold charge to >40% (Sunday 100% as before).
How to do it? Before the system was installed I read all the manuals and watched many webinars but nowhere did I come across any statant of this SOC problem, yet it appeared to be quite well known.

Invest in a smartshunt

1 Like

Lithium batteries generally are meant to be CAN connected and the BMS provides an accurate SOC.
If you are running them in “dumb” mode, then you need a shunt as the inverter cannot see all the charge sources and lacks the accuracy of a shunt.
Best practice in this config has always been to use a battery shunt, if the battery is not managed.

1 Like

Victron smart Li (It’s what victron supply) do not have a CAN bus. Obviously, for (ha-har its beginning to seem) for maximum compatibility I used all victron products installed as per what victron recommend. Nowhere can I remember reading or seeing (in a webinar) that smart LiNEED TO HAVE A SMART SHUNT to work properly in a victron system such as I have. The installer and his supplier never mentioned it either.

Two smartshunts, which there is no space for.
Which, until now, nobody mentioned was required and I have read all the manuals several times and if its there, its not stressed.
It should be.
Also using battery voltage is better, it allows for ageing of the battery for one thing.

Just in passing (Nick) did I read you right in another thread that the Cerbo can only use one SOC “supply” at a time (minutes not microseconds)?

There can only be one source, in the same way the system only uses one bms. You can connect multiple in the newer OS, but only one can be the reference.
In the GX system setup the specific source is chosen as a reference.

Why two ?

First I need to be sure we are using the same words.
By “source” do you mean a device that can/is charging the battery?
So my PV inverter is a source and my Quattro is a source?
If this is what you mean then there is no intrinsic problem with having two (or 20) sources running at the same time (think constant current sources). You need to handle hunting of course.
However, it’s technically easier to have one at a time.

As far as I can tell the Quattro modulates the PVinverter so it does not overcharge them, although the carbo should have software to be able to do it too.
ANYWAY its up to the system (ie the Cerbo/Quattro whatever is in charge) to switch appropriately and measure/input the charging/discharging current from that source.
That said I will just check because I feel I’ve seen both charging at once.
As I say, for Li batteries it’s really better to use the battery voltage to estimate the SOC for battery charge management because that is capacity-independent and gets round all the problems I (and from the sound of it others) have been having with SOC%. Use the user-entered capacity for estimating how long the battery will last etc.

Ludo,
well I have two strings so It’s either two or a very expensive Lynx shunt that I have no space for. Might shoehorn one in if it can go on the left-hand side of the battery lynx distributor, which cannot be moved any sense due big thick cables and short run.

Could probably adjust the Lynx shunt to run the other direction by reversing the shunt sense leads.

TBH I am not sure the cerbo can handle two come to think of it.

You asked about SOC, in other words battery monitors.
These can be shunts, BMS"s or the inverter.
The inverter is poor at doing this and for multiple DC chargers, a shunt is best practice - which is why you will see one in most diagrams, and it is referenced over many posts on the forum over the years.

You can have lots of shunts doing lots of different things (they can do more than calculate SOC), but only one can be configured as a battery monitor.

Likewise for BMS’s that provide SOC, you can have multiple, but only one can be used.

Voltage as a basis for LiFe SOC just does not work well, the curve is too flat vs legacy lead batteries.

A single BMV configured as per the docs will track all energy flowing in/out of the battery.
In this config the quattro’s algo will be used to control charging.
With a smart battery, the battery controls the charging.

This is well documented in the DVCC chapter of the GX guide.

With AC chargers, frequency is used to control the device, via either the ESS or PV assistants (or modbus in the case of Fronius, and more recently some others).

One quattro one DC-string.

Your 4 batteries might be connected 2s2p, still they have one commen minus that goes to one shunt. So it can be the small SmartShunt that is also affordable…

Hi Nick,

  1. When I was looking at this about two years ago (before the system went in) all that was stated as a standard system was the ve bus bms to go with quattro, battery and rs450.
  2. There was no suggestion that the quattro (you didn’t answer about the rs450 charge accuracy) was not able to handle managing the SOC and I don’t think its mentioned in more recent manuals either. There is no reason to expect a top-line manufacturer like victron to skimp on measuring the charge on a charger after all.
  3. Ok so because victron didn’t spell it out clearly I how have to get a Lynx smartshunt ($$$), Rewire the quattro and the rs4500, reverse the direction of the smartshunt and cost me $$$.
  4. Lead acid using battery voltage for SOC% is not good, I wouldn’t suggest doing this. Lithium is flat, just means slightly more accuracy is needed, ideally to ~10mV, which victron devices already manage out of the box. So that is really not a problem, and luckily the curve seems to persist for the life of the battery (which is why you can mix old and new in parallel without problem. Its a better method and I am puzzled why victron do not use it as an option.
  5. As Li age and capacity drops the SOC% as measured by shunts become increasingly inaccurate, -20% by rated life and obviously more if kept longer (its economic to go to 50% or less). This is likely to cause more discharging (as a %) than the user thinks and eventually results in unexpected system disconnects due low cell voltage. Not ideal.
  6. Remember victron smart batteries are not smart!!!

by constantly recalibrating SoC over “full” voltage and tailcurrent the system is highly reliable, the SmartShunt only costs $$, a fraction of the total system costs.

Its gloomy today but here two devices charging and two discharging on a gloomy day.

It happens, not infrequently so the system should be able to handle it.
The charging current is quoted to two decimal places (3.10A) but actually only resolved to one. This suggests an accuracy of about 0.1A or 5W which over 24 hours is only about 120W/day which is far far less than the actual observed inaccuracy (93.5% claimed, more like 82%),

what are you trying to tell us ?

Sadly its done properly so both are individually fused into a Lynx distributor.
That has four connections, the other two go to quattro and rs450.
So one smartshunt is not a goer.

Ludo
So is using battery voltage which is free.
Also allows for battery cycling degradation.
When your battery drops 10% capacity, what you think is 40% charge is only 20%.
Not so good.
I was slightly surprised that cell balancing is required weekly, but its victron’s call.

Nick,
I read the GX guide, its how DVCC is set up (see earlier post).
11.2.1 says only:
“For Gel, AGM, OPzS and other lead-acid batteries, DVCC can be used without any problem. The same is true for Victron Energy Lithium Battery Smart with the VE.Bus BMS, the Lynx Ion + Shunt BMS or the Lynx Ion BMS. DVCC is forced-enabled for the Lynx Smart BMS.”
It doesn’t suggest you should use a smart shunt at all.
Also here exactly my system but no smartshunt. There are several like this over the manuals. Nothing to suggest system malfunction without a smartshunt.


I do not have a smartbattery, I have a victron smart Li (!!).
So although you appear to be right that the victron charging systems are inadequate out of the box, I do not think there was anything in the victron documentation to alert me to this problem (or my installer or his supplier),

I see only two ways out of this

  1. sue your installer/supplier
  2. reconnect the batteries to a SmartShunt, connect the shunt to the busbar.

Your choice, your nerves, your money