Wolf's PowIRwall

[OffGridInTheCity]

So far so good, The replacement of Frankenstein has proven to be a good idea as my cell voltage difference has notably settled down,
At some point I will check all logmon voltages against my fluke and see how much deviation there is,
For now I am happy with the 20mv to 30mv jitter. After all a 0.01V difference on a longmon is the difference of 10mv so what can I say. |
Wolf

Yes, there's a practical limit. As I read this line from the the https://wiki.batrium.com/en/product/cell-monitor/longmon -> Longmon... DataSheet.pdf

longmon voltages can be off (in theory) as much as 0.1v = 100mv. Originally Batrium allowed user calibration but I think that software/tool is no longer available.

My sense is, my longmons have not been off 'that much' because when I build the packs I discharge the new packs to a common storage voltage of say 3.8v via the iCharger X8 (and verified by random voltmeter) and then bring the battery battery bank to the same voltage. So there is some small current flow but the longmons typically show as 3.78v - 3.82v rather than 3.75v or 3.85v as an example. But this comment is anecdotal - have not researched longmon accuracy. With a 3.5v - 4.0v operational range it doesn't matter, but it also means in the back of my mind, I would not go above 4.15v as my max charge.

 

[Wolf]

Ahhh total bliss.
The power company thinks I'm burning a 40 watt light bulb 24/7. Little do they know AC is running and all the rest of the stuff.
This fall I am going to install a 3 zone 2 ton Ductless Mini Split AC With Heat Pump and the window AC's well go out the window so to speak.
This should also improve my HVAC efficiency by a long shot.
The 4 day graph is showing my 4 GTIL operating just about right. For some reason L2-2 likes to play the boss so I throttled him down to 600W max to keep the Mosfets from popping. You can see L2-1 playing catchup when the boss runs out of steam. Other than that all is well.
Wolf

 

[Wolf]

Wow I just upgraded to Grafana V10.0.2 and they finally got transparent right.
The feature set is wow. I am impressed. It makes stuff pop out and really look great.
Here is my Solar and Victron Dashboard now. I actually had a lot of fun redoing it with limited frustration.
This is all while sitting in Austria and logging into my house Linux server. Incredible technology theses earth people.

Wolf

 

[Wolf]

Update:
Well the 7 day trace with the newly replaced battery is showing that perfect C/D/C sine wave. we are all looking for.
The small glitches you see are just that, network connectivity etc. They are replacing the underground power cables in the neighborhood and without fail the ditch witch will snap an internet cable line and of course it is mine. I almost yelled at them "Hey guys you just pulled my cable modem through the floor". Well its fixed again and the purple differential trace is looking good and stable.
So far so good.
Wolf

 

[Wolf]

I now have a dilemma with my power wall.
Basically it concerns battery #3 The are all LGFBM261865 with all cells used have a IR of min 35.09mΩ and a max of 38.97mΩ

Pack #11 and not quite as bad pack#6 have an interesting issue.
When the pack Voltage goes down to ≈3.48ish (my inverter cuttoff≈48V Battery 3 pack #11 has a voltage difference of over 100mv, however when the charging begins again the difference goes down to ≈ 50mv. At this point in the season I cannot charge to full ≈4.05V to see if the difference goes away. What should I do? I do have enough LGFBM261865 cells to build another pack for replacement.
I also am not opposed to get another 1500 or so 2600mAh cells of the a same manufacturer to replace the whole battery
Here are the charts to see what I mean.
The 4 batteries and their v difference in purple trace. as you can see battery #3 has quite a difference.
I have also confirmed this with my Fluke.

The pack showing low V

The Pack showing the change

There 2 excel sheets that show all the cells that have been tested and which ones have been used.
It also includes which cells and at what position in each pack it was placed .
I am digging in the sheets to find out why this is happening but have not found anything yet.
Maybe someone can find out why this anomaly is happening all the info is here https://1drv.ms/f/s!AmNMFw8cEOSHhvoyNSyMf1Du5pXGiw?e=otNu0T
Wolf

 

[Wolf]

Small update,

Down to a 47mv difference.

Wolf

 

[Oleksii]

Pack #11 may have self-discharging cells and lost SOC comparing to others. And in case of being at lowed voltage, its voltage decreases faster comparing to other.
When you start changing - diff decreases, as all packs are more like at beginning of linearly voltage growing stage, where diff is less visible with different SOC.

Haven't you meet self-dischargers in your packs?
I did already, disregarding that I had same? level strict testing/checking rules when build my packs.
I continue to watch on the self self-discharger and one day I'll replace it.
A way how I was searching for it I showed here

View: https://youtu.be/TwbIJ23i8Ws
(auto-subtitles with auto-translation work quite good) I believe your nickel fuses might help you in a similar way.
Thermal camera just confirmed my measurements by DMM.

 

[OffGridInTheCity]

My operational range is 3.54v to 4.06v. My absolute minimum max difference over the whole range is 30-40mv no matter top, middle, or bottom balance.

However, I have pack #31.... In the lower range of 3.54v to 3.8v, pack #31 does OK but when I get a burst of sun after several months of bumping along the lower voltage range and charge up to 4.00v - this pack goes from 60mv max difference over the range to 100mv max difference (sagging) in the upper part.

So... I pulled out a hobby charger and added 18ah of charge to just this pack of 260ah and now it's back to 60mv max difference again over the entire range.

Last balanced to 40mv max difference range 8 months ago. 18ah of charge today / 8 months of no balance = ~75mah/day of self discharge of 260,000mah pack. So small a self-discharge it doesn't show in 3.5v to 3.8v range until pressured up to 4.0v?!? I don't know!

I've done this - e.g. added a touch of charge - a few times over the years. It's not radical enough to actually replace that single pack since the direct charge approach is an easy fix and there's no sign of degradation as this pattern has been steady for 4 years. Can 75mah/day self-discharge stay steady for multiple years?

Following along for ideas

 

[Oleksii]

Can 75mah/day self-discharge stay steady for multiple years?

In my case, as I know my "fuses" resistance, I calculated the single 18650 (on photos of prev post) self-discharge current as 5mA. Per day it would be 24*5=120 mAh, so pretty much close to your case too.
I give my pack to work almost another year with the self-discharger cell and I do not see rate change. So maybe it could last long time this way.
In ~1 year period I had to fill ~10Ah to the 160P pack to compensate 40mV diff on idle.

 

[Wolf]

Haven't you meet self-dischargers in your packs?

I have not as far as I know. This may be my first experience but i am not sure.
I can see why you may have an issue as you have a bunch of Sony cells.
I do know that Sony cell do not work nicely with other cells in a pack.

So... I pulled out a hobby charger and added 18ah of charge to just this pack of 260ah

I have done the same however this pack within a week or so went back to its behavior.
Here is a 6 month analysis of that pack #39 wich is #11 of the battery it is the blue tracer.
Orange is a properly working control pack #32 and the purple is of course the mv difference of the high and low cells.
Skip the spike noise and in all actuality this pack stays at 3.41V when my inverters cut off and just stays there especially
during winter and snow covered panels with very little sun at my latitude. It really never discharges beyond this point.
Now in mid February we are getting some sun and I am charging again with panels producing 10kWh to 13kWh.
Wolf

 

[OffGridInTheCity]

If I'm understanding, we're possibly seeing similar behavior of a pack that seems to hang in there at low stress at the lower voltage range (e.g. winter) for several months, but then shows some sag when pressed into moderate stress, higher voltage ranges as more sun comes around?

All I can think is that self-discharge is less (less mah/cycle) under low stress and more under moderate stress / higher voltage range? But if so, I don't have a good mental picture of how this would work cell/chemistry/capacity wise. Apparently it's not fatal as it's minor and seems stable over multiple years in my case. And at this level, active + ongoing balance would hide it. One of the reasons I avoid active balance is I want to see what's happening in the worst case.

Here's a recap of min/max mv pack differences for the given month since the last auto-level in May 2023. From June 2023 to Jan 2023 you can see the gradual drift from 30/60 to 70/90 mv min/max differences.

Mid Feb reached 60/100mv min/max. The 100mv turns my dashboard RED In this case I added 18ah of charge to the sagging pack #31 which brought things back under 100mv. I'll see what happens in March. Soon, I'll need to do another Auto-Level touchup.

 

[Wolf]

If I'm understanding, we're possibly seeing similar behavior of a pack that seems to hang in there at low stress at the lower voltage range (e.g. winter) for several months, but then shows some sag when pressed into moderate stress, higher voltage ranges as more sun comes around?

Actually I think we are seeing a reverse of each others curios issue. In my case the vast difference happens at a discharged level below ≈3.58V

However looking back in time to august of last year it appears that this pack was also having an issue at full charge, but the mv difference was below the threshold and finally setting an alarm in november.
It wasn't that worrisome to me but more of a mystery. Now that charging season is getting closer I am more concerned. I have inspected the
cells in my detailed excel sheets and am yet to find an aha moment. In any case I am going to build a replacement pack with the same part number cells ( I have plenty left over) and replace it to see what happens.
Wolf



.

 

[Wolf]

I'm so highly intersted in the testing results after 3 years of operation.
What are your voltage limits? 4.04V max to 3.45V min?

Yes me too however the frankenstein battery cell level retest will have to wait some more till I replace an odd pack on Battery #3.
I did check the individual pack voltages july 9th 2023 and all where holding ≈ 3.5V. They have been in my shed since then and soon I will check them again. Yes to your question ≈ 4.05V to ≈3.45V.




In the meantime I am retesting a bunch of LGFBM261865 so I can build a comparable pack for Battery #3.
I am now running 4 Megacell Chargers to speed up testing.
Wolf

 

[OffGridInTheCity]

Love the test setup
After 5 years, one of the slots on my 3 x OPUSs has started behaving erratically - so I'm down to 11 cells at a time.

 

[Wolf]

one of the slots on my 3 x OPUSs has started behaving erratically - so I'm down to 11 cells at a time.

So you think my 64 cells at a time is a bit overkill?
You want me to send you an OPUS? I'm pretty sure I can blow the dust balls off of one of mine, throw it in a box, stamp it and its on its way.
Oh BTW they all have the fan mod also.

Wolf

 

[OffGridInTheCity]

So you think my 64 cells at a time is a bit overkill?
You want me to send you an OPUS? I'm pretty sure I can blow the dust balls off of one of mine, throw it in a box, stamp it and its on its way.
Oh BTW they all have the fan mod also.

Wolf

OMGosh, I'm working on battery #10 and was thinking it would be my last, but I have room for more batteries. 6 more would take me over 200kwh and up to 238 longmons. Agh, will it ever end

 

[Wolf]

So as you all where well aware I had a battery (#3) with pack #11 showing me the toothless wonder.
If it wasn't embarrassing enough I had to put an arrow on it so you could definitely see.
As of late the pack would see wild swings between 55mv to 120mv during charge and discharge cycles.


Anyway after finishing testing another 430 LGFBM261865 cells I came up with 80 of the best cells and built another pack
to replace the offending pack.
The end result was nothing but astounding------- so far. we will see if this solves the problem.
The biggest surprise was that today 4/1/2024 was the first that this year I hit 100% charge

I have already disassembled the pack and am going through it cell by cell.
All spot welds where intact, no broken fuses all cells where checked for voltage after the positive side fuse sheet was removed.
All 80 cells measured 3.6745V/3.6746V so no big difference there.
I am now in the process of checking IR and capacity on all 80 cells.
Since my cell recordings from initial first touch to pack placement is considerable I hope to find something that can solve this mystery.
I will update accordingly.
Wolf

 

[OffGridInTheCity]

Since my cell recordings from initial first touch to pack placement is considerable I hope to find something that can solve this mystery.
I will update accordingly.
Wolf

Looking forward to more info on this!

 

[Wolf]

Now this is what 4 batteries in sinc should look like!
Wolf

 

[Wolf]

Some interesting updates.
First off the pack replacement has been more successful than I anticipated.
My pack Voltage differences are now running between 10mv to 20mv.


I was in the process of a major home network upgrade, WiFi, repeaters, installing a new RasPi 4 with an upgraded version of the Venus OS from Victron which scrambled my Node-Red flows, so I needed to rework those, all while setting up a new Linux server with Docker on it to host the Venus Interface. MQTT server, influxdb, an updated version of Node-Red, and also updated Grafana.
So basically a from the ground up new installation.
I have been hesitant to do the firmware and software upgrade to Batrium as it meant a complete rewrite of the Nod-Red flows.
Since I was diving deep into the revamp of virtually everything, new database structure, renaming objects and devices. Although this system was working just fine, it was in the true sense of the word a "patchwork andy" with not much of a proper structure. I was /or still am running 3 linux (ubuntu) laptops all doing something different. It was time to consolidate. Oh yea I am also putting together a HomeAssistant server which at this point controls my 3 Mini Split heads with automated settings. Nothing fancy yet but at 5:30 every morning they warm up the house till 7:00Am.
So back to upgrading Batrium to 2.17.15, which went surprisingly well. It did however trash my Nod-Red flows.
There are some Node -Red flow versions out there that rely heavily on some of the legacy node topics which some of them actually work and still send data.However none of them met my immediate need.
I hastily patched (here we go again) togather a reasonable working flow from @Slowfinger which gave me time to build a new flow to my liking.
Got to have my data. I ended up using the still operational Legacy "4232 Cell full details Node " as it appears this one was kept functional, There are 3 versions of the payload data. Topics that run up to V1.0.29, then from V1.0.30 up, and V2.14 and up.
Here is a screen shot of the completed flow. I will share if anyone is interested. You do need a1.8.10 influxdb set up somewhere or you could create a very nice dashboard in Node-Red. Also if anyone wants the other nodes decoded let me know although there is much repetition in the data.

The last show and tell I have is the analysis of Pack 11, 3rd battery aka longmons pack #39.
The only thing I found was that when using the same Mega Cell Chargers as I did when initially testing these cells is that 2 cells showed up "bad" on the MCC. I re-tested these cells with one of my SkyRc MC3000 and they where fine, although the SkyRc showed an IR at a laughable 9mΩ.
I have attached the the Excel sheet If you want to dig but I could not find anything definitive.
Here is the representation of the study.
Wolf