AIO 13.5 plus G/W - sudden reduction in SoC

Hi

At 07:00 this morning, the SoC on my battery was at around 39%, but now, 1 hour later it is at 15%!! I have checked the various consumption graphs on my HomeAssistant system and they show the usual 700-800w load, so there is no apparent reason why I should lose nearly a quarter of my batter in an hour

I have solar panels with a Solis 3.6kW inverter and am also running predbat (latest version). I did a recalibration about 2 weeks ago.

Where should I now be looking to diagnose the issue?

A few minutes after I posted the original message, the SoC had reached 14%. It is now around 13% so the apparent discharge rate seems to be back to normal. I will keep a close eye on it because it is predicted to be nearly flat before the solar panels really kick in

Think @Jellybaby has this kind of thing (unless I'm mistaken) - I'm not fully convinced his cells are OK on atleast one of his packs despite what Giv say.

hoggy

Thanks for the reply. I will look at the threads of @Jellybaby to see if there is common ground. Meanwhile, I will try and set my charging regime for today so that the SoC drops to 40% or so around midnight or 1am. I will then only have 3 hours before my next cosy period an hopefully the battery will last until then. This will help confirm whether this is a one-off or it is a problem

Last night I allowed the battery to drain to its 4% reserve and did not experience a dramatic drop off in SoC. Instead, ther was a small kink in the discharge line between 05:30 and 05:45 when the SoC dropped from 16% to 12%.

This level of abnormality has been seen before during recalibration and whilst it represents about 3x the actual rate of discharge, it is short lived and bearable. I will be importing for the next couple of hours whilst the solar generation ramps up.

Hopefully my issue was a one off, but I will make sure to keep an eye on things by regularly discharging to my 4%reserve, maybe once or twice per week. It's probably not a bad thing to do anyway

Weasel perhaps also charge to 100% from time to time - easy while there's so much solar.

DD I have been doing just this. In fact I only started deeper discharging recently

the BMS will perform a self check around 40-20% range.
It also self calibrates when going to 100%

TheDragon (GivEnergy) Thanks for the information, it's interesting that yesterday's drop was of the order of 25% whilst today's was only 6%

hoggy Funnily enough I was just coming to post in the BMS 12 thread
ARGH THIS GODDAMN FORUM ERROR

Makes any kind of battery automation 100% useless.

Weasel What I have been forced to do is keep track of daily in/out.

Say for example I charge from 4 to 100 % I will automatically deduct 1.5 from the original charge.
Now if I spill over into the next day and it gets to 100% before fully discharging I will once reaching 100 % again deduct 1.5 totalling 3, so on and so forth until it reaches 4 %.

For example

Edit
Not change that 15 on the 4>100 in a while so disregard that lol, only reason for charge/discharge excess is so I can deduct it in formula for same day in and out and ive been lazy on date changes as well it seems lol

Jellybaby
Thanks for the thoughts on tracking the "real" SoC, I will bear these in mind. Given that I am on the Cosy Octopus tariff with its 3 cheap periods, I only get down to <40% SoC when I want to.

I try to minimise exporting, so for the last week or two I have been deliberately skipping some charging opportunities to minimise solar exporting. I have also now tuned predbat to try to minimise exporting and this seems to be working OK, but I occasionally deliberately still want to exhaust the battery and it was on such an occasion that I expereinced thsi worrying drop in SoC.

The following day, after at 100% charge, I deliberately exhausted the battery again. This time, I did not see a huge drop in SoC but a small "knuckle" that I has previously seen during recalibration.

I will keep monitoring via HA and let predbat do its thing, along with an occasional full discharge. I will update if I discover anything new, but so far, this seems to be a one-off.

One final thought is that I had enabled low-power charging in predbat and am wondering whether this could be a contributor. I am not sure how, but it was the only thing that had changed recently in my setup

Weasel No clue on predbat thing, I would like to use it but with it tanking 10-20% between 30-40% the predbat plans are useless imo, id love to use it but I don't feel its suitable for the GE AIO system.

Jellybaby Hopefully, my experience is a one-off and does not become "normal". I will leave predbat doing its thing for now because it is not predicting that the SoC will drop below 50% and your experience is that the cliff edge is lower than this.

In a week or so, I will probably skip a charging session to force the battery lower, fully knowing that I might experience a cliff edge rather than the small bump that I see at around 20% SoC. I will update the thread accordingly

Quite by chance, I experienced another, smaller cliff-edge with a 13% apparent drop in SoC in half an hour.

I have looked at the battery voltage at this point at the behaviour seems to start when the battery voltage drops to around 305v. I am not sure exactly what this means right now because I do not have enough data points to be confident in any inferences that I make.

Jellybaby Do you see a similar thing? ie. is there a correlation between the battery voltage and the cliff-edge?

Weasel
Not sure if this is of any help, but looking back at battery data on the 12th I detected a sudden fall off of the SOC. During the morning no solar, batt at 310/311 most of time. Lunch time kettle, batt dropped to 306/7, cliff edge fall off in SOC rate for next 2 hours. Solar input mid afternoon at 2kw, batt returned to 311/12 , solar depleted next 3 hours, batt remained constant, SOC rate returned to normal.Evening kettle, batt 303/4. The batt volt remained at circa 300/306 for rest of evening but SOC just reduced at normal rate until midnight.
Since then I have fully charged and run down system to 5%. Not seen any cliff edges so it may have been a calibration thing and batt needed to find its feet.
Regards
Martin

mole Thanks for that. My concern is that I had only performed a full calibration a week or so before. Therefore it appears that my battery might be going out of calibration pretty quickly

So far, I can manage around this issue (if indeed it continues to be one) but will keep a close eye on the drop-off and the battery voltage at the time. Between occurrences of the drop-off, I have been able to discharge without the huge drop-off (just a small one); it is only when my battery drops to 305v (or thereabouts) do I see this odd behaviour

More news as I have it

Weasel No clue, will have a peek tomorrow when ive woken up.

the 3kw draw was kettle being put on

Jellybaby
Wow! your drop-off is happening from around 33% and does not (on the surface) appear to have the same cause as mine. I had another brief occurence yesterday so I will continue to monitor closely.

Weasel its random what %age it happens

This is 9th of April, full charge happened on the 5th, before the drop and after and a general system view of before and how it went down after

yay, only took 4 minutes to get the picture to paste.

Hi

A quick update:-

I raised a support ticket via email with GivEnergy and after 10 days or so received a reply which was pretty much a fobbing off. I was planning to comose a reply at the weekend giving them a piece of my mind ...

This morning from about 01:40 to 02:40 the battery SoC dropped from 45% to 13% and I happened to wake up at 04:45 and see the battery at 9% and charging whereupon I set up some force charging until 07:00. Once again, the drop-off coincided with the battery stack voltage dropping as shown below

I am a software geek, not a electricity one, so cannot figure out whether the voltage drop is cause or effect, but a pattern is emerging here.

I have composed and sent the email to GIvEnergy support with the attached pictures and asked them to initiate a warranty claim. I have also contacted my installer to inform them of the position and that if necessary, I will use the Consumer Rights Act to obtain a refund since the equipment is not fit for purpose.

I have not (yet) posted on any social media platforms about this, but it does seem that posting on Facebook into the relevant groups often gets a quick response. I suppose nobody likes bad publicity.

Just to add that I had enabled low-power export and charge in predbat and these are about the only differences to my predbat configuration. As a precaution I have disabled these.

Secondarily, last week I migrated my HomeAssistant server form HA green to a HAOS VM in proxmox, so I have lots more memory, VCPU and disk. This prompted me to install Influxdb and Grafana. I will dig around to see what dashboards I can build that might help me keep a closer eye on things, especially with respect to the stack voltage thing.

You got any weird cell voltages - there should be 24 voltages per module and there's 4 modules. It stands to reason any cells pulling low would impact SOC drop off as they would skew the entire systems voltage as opposed to the expected (for a given SOC.)
I don't know what the HV stuffs balancer is aiming for but the old LV stuff was looking to get things to 30mV difference between cells so I doubt that it would be much different for this kit.

hoggy Thanks Hoggy - I haven't (yet) got round to checking the cell voltages themselves and expect that I will do so at the weekend

This was on the 21st, was a funny one.

Here to join the conversation. I have AIO but no G/W. I have seen a 40% to 4% drop overnight on a few occasions with a load of approx 150-200W.
I also got fobbed off by email initially, but it happened last night so I tried through chat this morning. Didn't get much help tbh but have been told I have an internal ticket raised and that someone will be in touch.
As mentioned, if we can't rely on the SoC percentage then a good part of the reason I went with this brand (Automation with HA) is pretty useless and I've wasted a whole lot of money with cheaper options out there.

Mine seems to have been behaving recently, but I think that this is simply because it hasn't dropped below about 40% for a while simply due to the great solar generation. I am keeping an eye on things and have an open ticket with GE support

I was initially fobbed off a bot, but when I went back with a lot more details, justifications etc, the tome of the responses changed

Mine still the same and I see THIS STUPID DAMN ERROR IS STILL HERE SORT IT OUT FOR GOD SAKE, if you can't handle a forum error then I have no hope of this getting sorted and this battery tech should be abandoned if its so unreliable.

Edit
Whelp, that was fun trying to get that image in

I did a calibration as GE support requested, and within 4 days had a massive drop again from approx 40% at around 1% every 10 minutes with less than 200W battery load.
The official answer is that after being reviewed by the Head of Firmware this is "due to the system performing a self-check. Since the All-in-one unit had seen very little usage over several days, the SoC adjusted itself when the opportunity arose."

So after 4 days the reported SoC appears to be something over 30% out from reality?

I also notice that the reported stats in the portal for the month of May show 158kWh battery charge, and 107kWh battery discharge which if true mean there's massive losses somewhere.

I'll keep pushing support I suppose but right now I very much regret this purchase.

I must admit that my AIO has been behaving itself for the last few weeks. I have tweaked my settings in predbat to try and give myself as much headroom as possible, so rarely drop the SoC below 30%. This gives me a nice buffer in case I see a sudden SoC dropped.

It helps that my Cosy Octopus tariff has 3 cheap periods per day and I am happy with this tariff.