Dual battery SOC discrepancy after firmware update

Firstly I want to join the chorus recognising @TheDragon (GivEnergy) for his efforts. Grafting all over a bank holiday weekend helping people with their systems is going way, way above and beyond and goes a long way to rebuilding confidence in the GE offering.

I upgraded my dual 8.2 batteries over the weekend to 3012. My batteries work OK, but have always displayed a couple of anomalies. There has always big discrepancy between both SoCs. When charging to 100% one battery would always jump from 90% to 100%. Finally, if one battery hit 100% then the other would stop charging, meaning that the total store was less than 16.4kWh.
The upgrade went seamlessly, with the SOC jumping from ₆₀% to 100% and then discharging down to zero (approx 45.5V). It then charged up to 100% and completed calibration.

The good news is that I haven't seen the jump to 100% that one battery would always display.
However, I thought that this BMS would address the discrepancy between battery SoC? This is what I am now seeing, albeit only after a couple of days.

I thought that this BMS would address the discrepancy?

Zakalwe

Similar story for me. My primary is 76%, secondary is 64% with an overall value in the middle at 71%. They do get a bit closer at times (around 4%) but are usually about 10% different. Not sure if you are the same, but I have 1 battery that's had 500+ cycles and another with 5 and my DY battery is always lower than the BG battery...wonder if it's similar to yours.

BG2121 has 315 cycles.
DY2216 has 205

Zakalwe Can't find it now but I think the non-jargon release notes for BMS FW 3012 say known issues include multi pack SOC not tracking as close as they want. So it should be being worked on for a future release I would assume from that?

chris_p Could be. I do remember @TheDragon (GivEnergy) mentioning how good the SOC tracked on multiple batteries during the beta testing though.

chris_p Zakalwe Can't find it now but I think the non-jargon release notes for BMS FW 3012 say known issues include multi pack SOC not tracking as close as they want. So it should be being worked on for a future release I would assume from that?

I didn't see that, but he did say this on another thread:

TheDragon (GivEnergy) 3012 has many new features, for example it can disconnect one battery and leave the other working, if the SOC get too far out, the lower can charge with the higher temp offline.

Crappy day for solar so far today, but it looks like they are converging:

Zakalwe I think they will adjust on the fly under 50% SoC at low power. Best way to align currently is to run them to empty and then leave idle for 30 mins, they will then kick off the OCV calibration and keep adjusting the SoC until it matches the cell conditions. You'll see the SoC moving up or down. Correct that hopefully giv will improve on this in a future release.

Good news is the split no longer prevents charge/discharge when one reaches full/empty early.

So, the new firmware seems to have gotten rid of the jump from ₉₀% to 100% which is really pleasing. Both batteries also charged to 100%, (or very close to ) which is also a first. :-)

Overall, I'm quite pleased with that result. I'll see how it performs tonight and see if it manages to do a better job of balancing the discharge.

Still quite a gap between them, though it seems to be getting a bit narrower....

Seems odd that one would charge faster than the other but I guess the charge rate won't be identical?

chris_p

No idea....one discharges and charges faster than the other one. Maybe a dodgy cell in the slower battery?

One other thing that I've noticed is that on 3007, when looking at individual cell voltages via GivTCP I used to see loads of "spikes" from some cells (mainly from Cell 15 in both batteries). I never knew if this was real or just "noise" in the data. Since the upgrade the number of spikes has dramatically improved, albeit more in one battery than the other.

Zakalwe It'd be interesting if Givenergy would either send an engineer to check the cell connections, or allow you to do it (if you feel confident) without compromising the warranty. If the spikes turned out to be due to poor connections, it'd be a very clear indicator of this type of manufacturing problem.

GivEnergy support have had a look and they think that it is nothing more than a reporting blip.

Regarding the SOC discrepancy, the gap is now as wide as before the firmware upgrade (currently 19% difference). I let the battery drop to 30% last night as I was expecting a decent solar day today. It ended up with one battery at 24% and the other at 42%. Once the Timed Charge period ended then "high" battery discharged down to 19% to support the house load with the other not discharging.

Zakalwe That's an interesting graph. I've never seen the SOC of my packs swap which is higher and lower. The lower one is always lower but the gap between them does get smaller and larger.

judgepd

Its new behaviour since the 3012 firmware. Prior to this the batteries used to track together, albeit with a gap between them.

Zakalwe The change at 8:00 or so this morning looks like a deliberate adjustment of the SOC by the firmware but the crossover at 15:00 yesterday doesn't look like it to me.

....and the SoC jump is back!

Based on that, it's worse than before. Under 3007 the secondary used to jump from ~ 90% to 100% once the other one hit 100%. This time it's jumped from 81% to 100%

@TheDragon (GivEnergy) Any ideas?

Zakalwe it's because of this.

The OCV has pushed the SoC of this battery down and looks like it has adjusted too low (by 20%), so when it hit 80% it was actually full, so it jumped back up by the 20% 'lost' earlier. What was your battery power just prior to the dip? Bet it was low and solar was very close to base load

I've seen a similar issue and have reported, as shouldn't be doing it. Only difference, I haven't hit full yet to see if it jumps back

My example

Thanks for the reply.

mpartington The OCV has pushed the SoC of this battery down and looks like it has adjusted too low (by 20%), so when it hit 80% it was actually full, so it jumped back up by the 20% 'lost' earlier.

Haven't you just described the SoC bug that this firmware was to address? Namely the system unable to accuratley asses the SoC of the batteries leading to jumps up and downwards?

Zakalwe I think it's slightly different in the software is wrongly forcing this, rather than a gradual drift over time (so should be very fixable). I kept picking this behaviour up in the beta trials, but honestly thought they had fixed it in this last release as I hadn't seen it for a couple of weeks on the final version. Previous root cause was it could do the OCV at low discharge currents, which pushed the voltage down further than if the battery was at idle. Therefore it was assigning the SoC to the lower voltage and calculating a SoC value lower than it should. Interesting on yours that it seems to kick off after a small period of higher battery discharge (which would have lowered voltage), so I wonder if that is related, but why only one battery?!

I've sent yours and my example to Paul for further investigation. Probably a line of code wrong somewhere.

'Danger zone' for this happening (for me) seems to be in the morning when my batteries are below 50% and the solar matches the house demand, so effectively batteries are idle.

Thanks again for the clarification- very useful.

I let the batteries drop to 14% last night (partially as there is a good generation day today and partially to seethe battery performance at a lower SoC)

No sudden drops at the lower end, but still a substantial discrepancy between the two.

The primary battery still seems to be doing the heavy lifting- discharging and charging more quickly than the secondary (unless the capacity is somehow different?).

Still, it's an improvement on the older firmware- both batteries are now charging to full rather than one stopping the other from charging.

I have seen similar.

This is mine last time I forced charge to 100%. Still a good 10% difference, but voltages are identical (suggesting they are at the same charge)

I have 2x 9.5kWh batteries, I jumped from 3008 (from memory) to 3012 this week, I had no issues with charge/discharge sync as we are still waiting for panels I charge them to 100% on eco7 overnight and they have been fine. Following the upgrade this week they are no longer in sync and have varied by as much as 9-10% as they charge/discharge - Overnight the primary maxed out at 99% and Secondary 97% and currently site at 74% & 79%.

Dunno where I am going with this thread, but anyhoo....

I'm seeing lots of these SoC drops since the firmware upgrade. I'm pretty sure that they are just spurious noise in the data, but I rarely if ever saw them before. Nothing has changed on my network so I think that it is safe to assume that the firmware is responsible.

The SoC still isn't tracking across both batteries, despite idle periods and discharging down to 20%.

It's also interesting to see that although the battery voltages track very closely, the SoCs vary wildly

Also interesting to note that that battery BG2121 is displaying a lower voltage than DY2216, yet the displayed SoC has been below and above DY2216. That, to my uneducated eye, tends to show that the new BMS still cannot accurately calculate the SoC......am i incorrect in this assumption?

The Remaining issues part of the BMS Announcement / PDF did note:

1. Multi Pack (Inter Pack) online balancing: Not yet implemented. Offline (Idle) will, as mentioned
above.
2. Scheduled Offline OCV: User initiated, disconnect, idle for 30 min, test, adjust, and then reconnect**

So may be something to do with item 1 although if you've left them to idle they should track which leads me more to:

Battery Serials - BG21(2021) & DY22(2022) - so they are different vintages (I can't see you mention their capacity, are they the same? BG/DY would suggest they are different but I forget how that part of the serial maps out) .
I see your cycle counts are not a million miles apart so i'm wondering if your two packs have 2 different cell manufacturers inside so they track different SOC charts/maps.
Paul mentioned a few weeks back this was something newly discovered:

The OCV-MAP tables of Anchi cells and Penghui cells are added. Other cells use the OCV-MAP tables provided by Penghui manufacturers by default.

So have you got one pack running the Penghui cells & another running Anchi?
Unless what map is in use gets pulled in to GivTCP i'm unsure we'll ever really know to be honest.

hoggy Battery Serials - BG21(2021) & DY22(2022) - so they are different vintages (I can't see you mention their capacity, are they the same? BG/DY would suggest they are different but I forget how that part of the serial maps out) .

Thanks for replying hoggy.
Both batteries are 8.2, 100% DOD.

hoggy So have you got one pack running the Penghui cells & another running Anchi?

No idea, to be honest. And I'm not opening them to find out :-)

hoggy The Remaining issues part of the BMS Announcement / PDF did note:
Scheduled Offline OCV: User initiated, disconnect, idle for 30 min, test, adjust, and then reconnect**

As an end-user I am not sure how I am supposed to initiate this. I certainly am not qualified or competent to disconnected or test the batteries. If it needs that level of physical work then the firmware is irrelevant to most end-users. or am I mis-reading this?

Zakalwe yes it disconnects itself internally from all the other packs. Doesn’t need anything physically doing to it.
I read it to be a button in the portal that you click if you really really need to force it to retry an SOC calc and it essentially turns the battery off, waits, does it’s measurements and adjustments and then switches itself back on.

And no I don’t advocate opening it up (you won’t see anyway who made them as it’s not written on them or anything and you can’t get to the QR code on them at all) - was just a suggestion that in the future GivTCP “might” be able to tell you which voltage map its basing its figures off.
How the BMS knows I have no idea, it may be set in factory or maybe the serial number dictates who made the cells -who knows.

Thanks hoggy.
Maybe it's my system, but i sure am not seeing that much difference with the new firmware. Certainly nothing like what is being claimed about batteries taking themselves off-line to allow the other ones to catch up and match SOCs.

Zakalwe have you taken all the batteries below 40% when you have idled them? I don't believe it does anything if they are above that.

I emptied mine the other day and then gave them a good 1 hour to level (highlighted in pic below). They've been much better aligned since then. At the moment letting them empty seems the best way for me to perform the 'offline levelling'. Although it should work if you pause, with all batteries below 40% as well. The full solution is still to come, I suggested a statement was added to the release notes as I knew from beta testing tracking would be picked up quickly by multi pack owners.

However, in my experience on this software, a discrepancy no longer prevents one battery from charging / discharging because the other on has already reached full/empty. There are also lots of other benefits for people seeing the large soc jumps (many only have a single battery). Therefore it's rightly been issued, whilst work continues behind the scenes on multi pack tracking improvements.

mpartington have you taken all the batteries below 40% when you have idled them? I don't believe it does anything if they are above that.

They went to <20%

However, and this is probably where I may be doing it incorrectly, how do you get them to idle?

mpartington However, in my experience on this software, a discrepancy no longer prevents one battery from charging / discharging because the other on has already reached full/empty.
Agreed, and I've seen a big improvement on that front in my two batteries.

Zakalwe I've seen if you turn eco mode off in the portal they will idle..... Just tried it and seems to work....

I normally dumped the charge into the car or something at start of cheap rate (if they're empty they must idle) and then set the charge to take place much later in the off peak period so it had enough idle time.

They need 30 minutes initially for the voltages to recover to correctly reflect their non loaded state. You'll then (hopefully) see the SoC for each battery converging. Eventually they will stop and you can turn eco on, or start your charge.

I'm really hoping they can improve on it as well.

mpartington My batteries fully cycle every day but still stay out of sync by the same amount. Here you can see pre and post firmware update... They did improve but now the individual ones don't agree with the overall SOC

naltsta they look much better, not perfect but I think 5% split (or so) is the limitation at the moment. The system SoC should be different to the battery SoC, as they present the useable capacity differently (one system level, the other is the BMS which contains the unusable buffers). Although the way it's implemented seems a bit random to me. E.g. why do I have one battery at 98% and the other at 92% when full. According to the article they should be 95% at full and 15% at empty (which gives the 20% DoD).
[https://kb.givenergy.cloud/article.php?id=10]

Top end is a known point based on the voltage profile, so I don't understand why the batteries read different here, could even hard code the BMS to read 95% when it hits that condition. I don't buy it's due to slightly different capacities due to manufacturing tolerance as:

(1) The BMS isn't aware of the other battery, so should align to its own capacity -as this is displayed as a % it should be consistent (for any battery) at top and bottom limits (agree could be variations in-between)
(2) After an off line OCV, I often see the batteries switching position and the other one reading higher, so disproves any link to minor capacity differences

@TheDragon (GivEnergy), I assume it's being looked at, but any further insights and next steps?

I deliberately ran my batteries down this morning to take them well below 50%. I turned eco off to get it to go idle and left it for 2 hours. The SOC on 1 battery moved up 2%, the other battery stayed the same and the overall SOC went up 1%...the gap between batteries therefore went from 17% to 15%.

Doesn't seem like it worked brilliantly for me considering the battery was idle from 8:30 to about 10:50

judgepd I'd feed that chart back to Paul as it doesn't seem to be doing as advertised.

hoggy The Remaining issues part of the BMS Announcement / PDF did note: bms-firmware-v3012-new-generation-ocv.pdf102kB

1. Multi Pack (Inter Pack) online balancing: Not yet implemented. Offline (Idle) will, as mentioned
   above.
2. Scheduled Offline OCV: User initiated, disconnect, idle for 30 min, test, adjust, and then reconnect**

I did say this a while back, surely makes this thread a bit redundant until this is solved by newer firmware?

chris_p Can't find it now but I think the non-jargon release notes for BMS FW 3012 say known issues include multi pack SOC not tracking as close as they want. So it should be being worked on for a future release I would assume from that?

chris_p yep definitely agree, although I think the threads moved towards how to get the best out the current firmware. Point 2 can be mimicked in this version with some user intervention and doesn't seem to be giving the expected results

We're just stabbing in the dark really but is it only when you have a 2021 pack with a 2022 pack? Had a quick scroll back up and I think that's what everyone here has?

hoggy Certainly my packs have a difference of over 500 cycles between them.

hoggy Mine (2 x5.2s) installed same time although 9 weeks manufacturing difference wk 52 2021and wk 9 2022. To be fair, I'm fairly happy with how mine are staying together (and I'm deliberately trying to break it still). 10% difference is fine for me and it does seem purely cosmetic, rather than affecting real capacity. I think it's the direct comms between packs that is harder than expected. Sure good things come to those who wait.

Zakalwe
Can you tell me how you see the SoC of a dual battery system.
I can't find it on either the Web or Android App

Thanks

Rogerbu on the app, press.the battery picture, then press "more information".

Rogerbu
Or access it via GivTCP.

Is GivTCP available somewhere or is it still a hidden members club type thing?

Look up the YouTube channel Speak to the Geek. Loads of great info on how to install HA and GivTCP.

Right, I let the batteries discharge to under 20% last night. I then put them into idle (disabling Eco mode and hitting Pause in the app). I left them like that for 2 hours. At the start of the idle period the SOCs were 18% and 14%. It looked like it did some balancing, however as soon as I made them active again and they started to charge they immediately drifted apart.

One is now 42% and the other 27%.

The individual battery voltages were within 0.1V

As far as I can see for my system the BMS does not balance the batteries. I have seen no evidence of one battery switching off to allow the other to catch up or of the batteries maintaining a similiar SOC
https://community.givenergy.cloud/d/2199-can-i-turn-off-one-battery-to-balance-them-up-charge-wise/6

Maybe I am doing something wrong?

Zakalwe I am seeing this too. I can not seem to get the ocv to run. I have tried twice, lowing my soc below 40% turning eco mode off for over an hour. Nothing happens. Not sure why, I have made TheDragon aware. Seeing a 5% jump at the top some days too. Nothing major, some small oddities.

It looked to me like the OCV did run on mine...the battery's SoC did converge. However, as soon as they charged they immediately charged at different rates. At this point, one is at 37% and the other at 62%. The OCV looks like it made the difference worse!

Zakalwe To be honest I am not too concerned about the soc's being level on my system. 1 AC3 with 2 batteries attached. Providing the ocv runs to locate capacity, but this time of year especially I have not seen it run. Looks like 2 small adjustments on the 8th. 4% in total over a 1 hour period. But having my system sub 40%, eco off should have started to the ocv. I am lucky enough to have a large battery total capacity, so no real issue to be honest. Come October November I might be emailing support :-)

THALL

I'd agree with that. However, after all the huffing and puffing it's disappointing to see it underperforming.

Overall though it has corrected the two main problems that I was seeing, namely a SOC jump from 90-100% on one battery and the BMS stopping charging on both batteries once one hit 100%.

Mine are 30% apart with a 0.2v difference

Mine have over the last week or so slowly but surely got much closer together. They have at times been around 20% apart but over yesterday and today they've got down to 2% difference and at one point today they were both exactly the same. They've also gone from the total SOC being in between the 2 battery levels to the total SOC being just slightly above both which I think is how it should be. I would say mine diverge when discharging and converge when charging but each time they have got closer together.

now 98% apart with 0v different wtf

Zakalwe it's a PIA, but maybe try it at empty. If it's going to be a nice day, discharge into the car, immersion etc then turn off eco mode.

Or if you have a cheap rate time to discharge in the first hour, then charge up a bit in your last hour.

I don't know why it would work any better, but maybe it needs to hit the lower point a few times to finish learning (calibration doesn't seem enough)

I gave mine a few full forced cycles after applying the firmware and is better than yours for tracking, although I can see it diverging at the top again 😞

Looks like I can also draw a line through "a SOC jump from 90-100% on one battery " as one of the things that 3012 has fixed.

Jesus H. Christ GivEnergy...you had one job to do!

mpartington it's a PIA, but maybe try it at empty. If it's going to be a nice day, discharge into the car, immersion etc then turn off eco mode.

Or if you have a cheap rate time to discharge in the first hour, then charge up a bit in your last hour.

Yeah, no.
Maybe GE should have another go at this instead of relying on end-users having to do voodoo to get their damn equipment to work as advertised?

Zakalwe I see yours did same as I keep reporting.
Software is flipping the batteries over after an OCV, so the opposite battery reads higher. It's not one pack being larger due to manufacturing tolerance, cell variation, temperature or any other factors that have been suggested. It's software related 100%, and surely that isn't how it should be working....

Yeah, it looks very similar, doesn't it?
It's very disappointing to see the SoC jumping again too. Hey, ho, it just looks like something that they cannot get to grips with. I can live with small jumps and hopefully it's cured in the main for those who were constantly plauged with massive jumps.

Ours appears to be tracking fairly closely now, after a few times of discharging down to below 20% & charging up to above 90%. Hopefully, it will stay that way and be even better once the Inverter software gets updated.

Just had a top end correction of about 5% (after a good sunny day of charging) both now at 100%

I did a manual idle OCV last night just for fun, from 2100-2300 hours. My packs SOC swapped just like other above.

Gained about 10% SOC on my primary (66 cycles) and lost about 2% SOC on my brand new secondary. (2 cycles)

This is what winds me up.

How can 1 pack be 27% SOC @ 52.7V and the other be 53% @ 52.6V. Yes there is 2 degrees temperature difference but it just doesn’t make sense. Doing yet another manual OCV to see what happens.

We are now on 3013 BMS and the two batteries appear to be tracking quite closely now:

Also, when the batteries filled up last evening the "Remaining Capacities" leveled out at just over 8.65kWh each.

Having just gone through all 65 posts am I right in thinking that the balancing of batteries SOC only occurs where two batteries are connected to one inverter? I have 1 x AC3 with 8.2 battery (1022 cycles) and 2nd AC3 with 9.5 battery (213 cycles) This means that the manual OCV described above won't work for me?

TrignFlo no, each battery can do an ocv. Don’t need two of them for it to happen.