Pylontechs + DIY battery.. unexpected behaviour

[Caesium]

I'll start a new thread for this so as not to derail Joeboys build thread.

So as some of you know I recently completed my build and added my 15 LF280Ks in parallel with my 6 Pylontechs.

The initial evening of discharging looked promising, the voltages held up longer than they used to so the extra capacity is clearly being used, no problem there. I also monitored the first couple of hours of charging and indeed, it charged a lot slower, to be expected as there's a lot more capacity

I left it to it and went to bed. In the morning I noticed that the Pylontechs hadn't reported 100% SOC at any point overnight. They'd peaked at 88% and actually the voltages "wobbled" a bit. Here's some Grafana charts of what the Luxpower was reporting.

So first is the charge rate. This flatlines at 3.5kW for a few hours to be expected, drops a bit at 1am, then zig-zags around a bit:


Here's the battery voltage over the same time:


This peaks at 51.5v at 1am, then wobbles between 51.1v and 51.4v for an hour. The drop at the end is the end of Go Faster so charging is turned off and it drops off to 50.1v

Now here's the clue I think. This is a graph of the maximum charge current requested by the Pylontechs:


I've zoomed this in a bit to where the problems start. Its at 150A (the max for 6 packs, I think) until 00:50. It then gradually drops to 135A, then 130A, but then around 1:20am it plummets to 15A - but only for one reporting period, then it shoots back up to 130A - then back down again! It does this a few times.

So really whats happening is the inverter is just doing as its told and reducing charge current when the Pylontech asks it to.

My theory is the new DIY batteries have a slightly different charge curve to the Pylons. I think the Pylons are basically getting to the voltage that they're full, but the DIY stack isn't (and is probably eating most of the charge current by this point, I can't measure that yet). So the Pylons reduce the requested current. However this cuts off charging to the DIY stack as well which isn't full, so its voltage drops, which pulls the Pylons back down (probably even the Pylons end up charging the DIY a bit?), so they request more charge current again. This goes on for a while and I think they never have time to really get in balance so the reported SOC never hits 100%.

So the biggest problem here that I can see, if the Pylons never get to 100% (I've seen this same behaviour for two nights now) then eventually they'll request a full charge from the inverter, which can happen at any time, probably during peak period, and it could take a day or more during which time I'm eating peak electricity and not getting any battery gain.

My first experiment to try and alleviate this is to do as nowty (I think) suggested and just go the "this is a lead acid battery, dear inverter" route and turn off the smarts. I've experimentally done this this morning and its charging now with a high-set voltage of 53.0v. I want to check this works properly before trying it in the middle of the night This should avoid the zig-zagging and the battery-to-battery balancing.

I think the way forward for me is possibly to sell the Pylontechs and maybe buy another 15 LF280s. I kind of wanted to do that anyway so this is kind of cementing the decision now. Especially as I've lost the modbus connection and the SOC reading on the inverter (it immediately dropped from 85% to 25% when I told it it was a lead acid battery.. clearly not right).

Joeboy I'd be especially interested to know if you see similar things when you get your stack built!

[Oldgreybeard]

How long have the packs been running in the new configuration?

I ran three Pylontech US3000Cs for about 6 weeks, then added another three US3000Cs, and it took around two weeks for the data to stabilise and all six battery packs to get synced up and balanced. The way that the system works seems to be that the pack with the highest state of charge dominates when it comes to telling the inverter/charger to slow down and shut off, so the packs with the lowest state of charge don't get a look in. The balancing current that the Pylontech BMS uses seems very low, based on what I saw, so that explains why it takes so long for all the packs to equalise if left to their own devices.

FWIW, the indicated SoC of the three batteries in use at the time I added the three new ones was around 70%. I deliberately waited until it was around there, as I assumed that the new batteries would be at around that level when delivered. Clearly the true difference was a lot greater than that, hence the long time it took for things to sort themselves out. Over the 18 months of more that we've been running the full 6 packs they seem to have stayed in perfect step. My suggestion would be to just wait for a week or two and see what happens.

Setting the inverter up as a dumb battery charger might well be the answer, as that would stop the Pylontech BMS from trying to control the whole lot.

[Joeboy]

I will report back. I would suggest doing a full discharge at a known load to see the capacity of the combined stack and charge from there to see how it performs. The Pylontechs tend to stick at 88% displayed SOC so i would go on voltage alone and forget about the SOC from the inverter. I am looking about for a better monitor with app ability too. Found one on Aliexpress last night for about £75, not bought yet.

https://m.alibaba.com/product/160052589 ... om=Android

Last week we discussed the charging profile of the Pylontechs and the speed which the pack with get from 88% SOC to 100% SOC. Essentially, once you are up at 88% SOC as indicated on your software you are really closer to 95% maybe more. These are my readings
88% SOC @ 51.3V
95% SOC @ 52.6V
100% SOC @ 52.9V
My stack will literally leap from 88% SOC to 100% SOC in a couple of minutes (certainly under 10 mins) if its sunny.

Capacity test would be my way forward to reassure and then monitor as it goes to see if it syncs. There may also be variance from unit to unit for BMS, I have a memory of changing around my stack to have a different master battery due to some small problem as I was adding to the stack. At the moment though i'd not mess about with it beyond cap test and keep an eye as you are doing on the voltages.

Does your inverter not have an eco function where you can set a charge window for the stack?

[Caesium]

Oldgreybeard, the complete stack as it is now has only been in place 2 days, so I agree there could be a lot more self-balancing for it to do. There's such small voltage differentials for quite high differences in SOC that things happen slowly.

The existing 6 Pylons were have been in place about 2 years so they should be balanced between themselves.

I won't make any drastic changes for now and give it time to settle down. I have changed the inverter to think the batteries are lead acid though, and my test charge went ok. That will at least stop the charge rate see-sawing when they get nearly full which I don't think is very good for anything.

Interestingly even though I set the inverter to 53v, the batteries levelled off at 51.5v with only a few hundred mA flowing so I don't think they were going to get much higher. I'm reluctant to raise the inverter charge voltage much more though as I don't have any overvoltage protection on my DIY stack. The balancer only kicks in when it sees 100mV difference between cells I think; it doesn't care about overvoltage - maybe I need something else for that.

At any rate, I think 51.5v is pretty well charged. The Pylons got to about 91% but as you say Joeboy I've also seen them jump from 90% to 100% without much extra current going in so I think the SOC% is just a bit unreliable at that point.

I'll leave it like this for a few days/weeks and monitor closely.

[Tinbum]

DON'T increase the inverter voltage setting

[Joeboy]

Caesium, this a handy wee unit and is reading within 1% of the goodwe inverter displayed soc.

https://www.amazon.co.uk/Multifunction- ... 4842&psc=1

[Stinsy]

I'd have left them as they were for a few days to allow the new batteries to cycle alongside the pylontechs a few times before taking any action.

The way the pylontechs calculate SoC and therefore how much charge to demand is (I guess) based on a reference table of voltage and charge/discharge amps.

My pylontech stack often sits at 88% for ages, soaking up loads of charge, then suddenly jumps to 93 or 97%. Overnight charging too often the charge rate drops and spikes a few times as it gets to 100%. My assumption is that the reference table indicates 100% then after the charging stops the voltage settles to a point where the BMS decides "well now I think I've got 98% SoC so I will have a little more if that's OK".

EDIT:

Here is how my pylontech stack charged last night. The resolution is only 5-mins so there could be a lot more peaks/troughs inbetween readings:

[Caesium]

Had a better result last night with the inverter on dumb lead-acid charging. The inverter was set to 53v but the batteries eventually only settled at 51.7v or so, not sure why the difference - as you can see from graphs below there's no current flowing by the end of my 5 hour charging period so cable voltage drop can't explain it.

@Tinbum agreed I wouldn't go above 53v. This is already a little too high for me, I would prefer 52v, but I'm wondering why there's a 1.3v difference between what I set and what the batteries get charged to. I could try setting 52v tonight as an experiment. I suspect they will take a lot longer to get the last bit of charge in but maybe thats no bad thing.

Much smoother ramp down at the end of charging here:


Peak voltage 51.77v:


I am tempted to leave it like this for a few days/weeks and let it settle down and fully balance. Perhaps then I'll try handing control back to the Pylons for a night and see if the zig-zagging recurs or has disappeared because everything's nicely in balance.

But either way, this is a solution I'm happy with for now particularly as it removes me depending on having Pylontechs at all. Anyone in the market for any?

[Tinbum]

I use the battery CAN with my batteries but I intercept the messages and adjust them.

The thing with the Pylontech batteries is you seem to have to get to about 52.5v for the internal pack cell balancing to start to work, even then it is very little current moves between cells. You will see the ambient temp of the battery reported by the BMS rise as it starts the balancing. I've emailed Pylontech twice to try and find the voltage but no reply either time.

Victron recommend a charging voltage of 52.5v and their equipment changes the BMS messages to that.

I set my max voltage to 52.5v and then automatically once a week to 53.2v. I also have a setting that I run every now and then that can keep the pack voltage at either 52.5v or 53v to give the packs more time to balance their cells.

As has been mentioned before the battery SOC will stay at 89% for ages and then suddenly it will rise up to 100% over a few minutes. If I recall it's only at about 95% when the BMS starts to reduce the charging current. It's not very often I've seen my pack get up to 53.2v.

On the new C batteries any voltage over 54.5v will cause permanent damage to the BMS

[nowty]

I find that even if you set a charge rate of 54V to the Pylontechs, mine independently throttle off and go into standby by 52.8V. I think Joeboy's has measured a similar 52.9V max after a full charge.

Interesting @Tinbum, what you say about the Pylontech batteries needing to get to about 52.5v for the internal pack cell balancing to start working. I note some folk have commented in the past that on receiving new Pylontech batteries they were getting low and/or high cell voltage errors until their packs had done a few full cycles. The suppliers said that was normal, so seems to add weight to what you have found.