Wolf's PowIRwall

[Wolf]

Isn't it 'nerve wracking' to hook in a new battery?

Will be interested to see how long it takes for you to balance. Will you be using top (bypass volt) or auto-level balancing?

It is neve wracking to a degree but since I made all the mistakes on the first 3 batteries the fourth one was not so bad. Just waited till the Voltages were close enough and since all 4 batteries are in parallel well you know water seeks its own level.

Since the only battery that is somewhat "unbalanced" if you want to call 100mv difference between high and low cells at 3.5V that.
I will wait till I get to 4.0V or so before I consider top balancing. Battery #1 is notorious for its bad behavior at < 3.6V with max deviation between cells of 100mv. as soon as the voltage gets to 3.65V and above the deviation reduces and settles at ≈30mv at full charge 4.05V the other Batteries run in the 5mv to 25mv range. Let see what todays charge will bring. I will report later in the day.

 

[Wolf]

Well as promised the results so far. We had a great morning but after about 10:30 the weather turned sour. Heavy clouds the white fluffy kind just made for shading and then a quick rain shower. So I was not able to do a full charge. Maybe tomorrow supposed to be mostly sunny.
Nevertheless collected 22kWh and had a max of 4.78 kW from 18 250W panels. Got to love those Victron MPPTs, they will squeeze every drop of power from those panels.

So as predicted the voltage difference did go down on battery #1 to 46.5mv at about 3.7V unfortunately that was the end of the charging.
We will see if it is any better tomorrow.
Wolf

 

[Wolf]

It is 8:00 Am and charging is commencing As you can see Battery 1 is the only one that has this greater difference of 80mv to 90mv at around 3.55V to 3.65V. Batteries 2, 3, and 4 are all running a cool 10mv to 30mv difference at all charge levels.

Once Battery number 1 gets above 3.7V it settles right down to a difference of 20mv to 30mv.
My first top balance was on the Frankenstein Battery which is Battery 1 when I installed it. May 2020. I also ran a top balance on Battery 2 when I installed it. Since then I make sure the pack are all within ≈50mv before I put them in service and haven't needed ANY form of balancing since then.
So on a rant note when Batrium says my longmons are failing because I balance too much I can tell them with confidence they are full of
@Batrium are you listening?
I must admit I only have had 3 fail but nevertheless............... Warrantee and support are certainly not Batriums claim to fame.
I do still like it and for all its worth am happy with it.


Here are seven days of Battery 1 metrics. As you can see as long as the cells stay above 3.7V their difference is negligible, and are in tune with all the others. For all I know the differences could be the quality of the longmons and their tolerances.
Wolf

 

[OffGridInTheCity]

Really, really nice data/graphics!

A question - do you monitor capacity?
One could suspend regular operations and perform periodic charge/discharge tests to measure overall battery bank capacity but... From the beginning, I've been recording/logging the AH/Volt metric during 100% discharge periods (at night) with the idea of detecting loss of capacity over time. So far (3 years) this AH/Volt is generally stable but this metric varies significantly based on the load.

Are you tracking anything to try to monitor capacity over time during live operations?

 

[Wolf]

Really, really nice data/graphics!

Thank you!
This can all be yours too if you want. Its running on just an old IBM Think Pad with a SSD drive. Wow I just checked and its got a 6th gen i7 didn't know that. It was a throwaway at office 3 years ago and I figured run Linux on it.


Just need node-red, influx and grafana The code is written for all of it just got to expand it for 112 longmons I'm halfway there.
The thing that it allows me to track is the mv difference between the packs and assuming I am correct if the pattern changes a considerable amount I may have a pack/battery in trouble. Battery 1 (Frankenstein) is really the only one of concern, but so far the deviations have been very predictable and I feel as long as there are no sudden changes the battery is OK. For me also it is the tracking history I can view back in time to be able to analyze the metrics to see if anything has changed. Naturally pulling data for six months requires a bit of time, nevertheless it is available.
Other than the occasional Batrium glitches the data is consistent. You can see when I was coming out of the winter into spring.

Wolf

 

[OffGridInTheCity]

Sincerely - thanks for the offer of code sharing but I've got a bunch of custom code coordinating/integrating Midnite Classic metrics with Batrium metrics in a SQL Server DB along with automated failover + backups. Too much to change gears at this point.

Going over 100 longmons broke my code - just got it fixed. I parse the Batrium logs every 5 mins for Batrium metrics. The Batrium logs do IDs as "@ 23" or "@ 57" but I didn't realize until seeing "@100" that when it's a 3 digit longmon ID, there's no space between the "@" and the ID number and had to account for that.

As you do, I track deviations between packs as a key metric - the live metric (updated every 5mins) looks like this example showing 40mv...

with the display turning large-font/RED if its >= 0.10.

And monthly the "Min/Max PackDiffV: 0.04/0.09" as it goes thru the daily cycle for the month. But of course, balance touch-ups will reduce the max to a degree. I find that between the full 3.5v/pack and 4.0v/pack they just don't stay exactly in sync no matter what I do.

 

[Wolf]

I've got a bunch of custom code coordinating/integrating Midnite Classic metrics with Batrium metrics in a SQL Server DB

I figured as much
Tech savvy as you are.

Are you tracking anything to try to monitor capacity over time during live operations?

Well that's a good question. The only sense of capacity I have is from the original pack tests which we all do from ≈4.2V to ≈ 3.0V
That gives us the actual full cycle capacity. I enter that data into Batrium and through some magic that I haven't figured out yet it comes up with a SOC. At 100% SOC I technically should have a full battery. Do I believe that no not really since I only charge to 4.05V and when the SOC goes to 101.5 percent it resets back down to 100%. Part of the Batrium SOC recalibration. I'm sure you are aware of this.

Since live operation is so dynamic it is very difficult to determine the "actual" capacity of the whole battery at any state other than what I would consider in the infamous words of Spock in the Voyage Home a guess.

So with this "best" guess in hand and the data we can collect over time as far as charge/discharge metrics I'm sure there must be a way to
calculate this. I know that I use maybe 75% to 80% of potential capacity as our charging voltage limit is 4.05V and discharge limit is 3.5V. So that in itself throws a wrench (question) into the capacity gear. I suppose since all my 4 batteries are on breakers I could isolate one at a time hook an iCharger to it charge to 4.2V discharge to 3.0V and check that against the values I got at the initial test. If I feel that the "whole" battery is not working at optimum capacity I guess, no pun intended I would perform such a test.
Did that answer your question, I am afraid not.

Wolf

 

[Redpacket]

+1 nice graphs there
I'd suggest that the capacity metrics will be more accurate when the cell voltages are on the more linear middle part of the curve & less so when past the low end "knee" or the top end "J" parts of the curve. Not sure what you've found with your cells re the 3.5V & 4.05V numbers as to where those are on the curves?
Wolf it sounds like with your battery #1 <~3.6V at least some of the cells are heading "below the knee", hence the cells deviating so much.

Agree as you mentioned, capacity & curves vary under higher loads.

 

[Wolf]

Wolf it sounds like with your battery #1 <~3.6V at least some of the cells are heading "below the knee", hence the cells deviating so much.

Agreed! I am in the process of verifying the Manufacturer and part number of each cell in the Frankenstein packs maybe there is a correlation there.
Fortunately I have all the data so I can analyze it.


On another note I hit the all green on charge today and as you can see pack voltages are all coming together as we get closer to 4.0V
Maybe we still have enough sun and time left for a charge to 4.05 but I doubt it.
Maybe tomorrow.

Wolf

 

[100kwh-hunter]

@Wolf i was out for working to much in the mean time, sorry.
How much is your actief storage at the moment?
And how much you got to be added up to your storage, and your final goal?
i grand you no less then 1 megawatt...you remember me?

With best regards Igor.
Ps i changed to lifepo4

 

[Wolf]

And how much you got to be added up to your storage, and your final goal?
i grand you no less then 1 megawatt...you remember me?

@100kwh-hunter Of course I remember you how could I forget.

With Battery 1@185.87Ah, 2@207.73Ah, 3@209.34Ah, and 4@206.84Ah would be a total of 809.78Ah if I calculate from 4.2V to 3V.
But running at 4.05V to 3.5V or from 56.7 to 49V I think I may have usable about 600Ah to 650Ah.
Quite enough for me and my small house. This weekend I will be adding another 750W of solar to accommodate the AC.

Wolf

 

[OffGridInTheCity]

@100kwh-hunter Of course I remember you how could I forget.

With Battery 1@185.87Ah, 2@207.73Ah, 3@209.34Ah, and 4@206.84Ah would be a total of 809.78Ah if I calculate from 4.2V to 3V

With Battery 1 @ 186ah vs the rest at 207ah'ish, does battery 1 have trouble staying balanced with the others as you go thru the daily cycles? That's a 207-186 = 21ah difference or 21/207 = ~10% difference.

 

[Wolf]

@OffGridInTheCity No not really I haven't touched any balancing feature on the Batrium in a long time, even through the winter.
The batteries are in parallel so battery voltages are going to remain the same. The pack voltages will vary a bit as the discharge happen
for me on battery 1 and that is around 100mv at its highest which occurs right around my 3.5v cutoff. Once charging occurs and the packs get to around 3.65v the differences in pack voltages changes drastically and goes down.

This can be seen in these 5 day graphs of Battery 1,2,3,and4.

Since the 4 batteries are all hooked in parallel and battery 1 is ≈10% lower in capacity it brings me to my work on how cell behave in parallel.
We know that the cells in parallel will discharge at different rates at a constant voltage dictated by IR and capacity.
So the "balance" issue with battery 1, I think, is more the combination of 4 different manufactures and 29 different part numbers. They certainly have different charging/discharging profiles, hence the issue, at least that's what I believe. Batteries 2, 3, and 4 are all the same manufacturer cells and part numbers and the voltage difference between packs shows. It is minimal running between 10mv to 30mv.
But to answer your question " does battery 1 have trouble staying balanced with the others as you go thru the daily cycles? " no it doesn't.

Wolf

 

[OffGridInTheCity]

The batteries are in parallel so battery voltages are going to remain the same. The pack voltages will vary a bit as the discharge happen
for me on battery 1 and that is around 100mv at its highest which occurs right around my 3.5v cutoff. Once charging occurs and the packs get to around 3.65v the differences in pack voltages changes drastically and goes down.

Interesting. It's my intent to run my powerwall until the cells actually start to fail - and I constantly wonder how this will go with several parallel batteries. You're point about the batteries (in parallel) maintaining the same voltage - of course! but I tend to overlook this because Batrium monitoring is pack based... so I've been focusing on the blue-bar 'trees' without thinking much about the forrest.

If I'm following your thoughts - even though a failing battery (14s) will maintain it's voltage with the other 14s batteries - this failing battery (e.g. significant loss of capacity) will add stress to the individual packs within that battery as the capacity degrades compared to the other parallel batteries. This is likely to manifest as increasingly erratic pack-level voltage swings thru the daily charge/discharge cycle.

However, pack level voltage swings alone are a bit subjective. My 112 pack max difference varies over the 3.5v low to 4.0v hi from 40mv to 80mv - *the same* every cycle and is stable for at least 270cycles - the longest I've gone without touchup balance. 40mv at 3.6v to 70mv at 3.8v to 50mv at 4.0v max difference - every cycle the same.

So it's not just pack variance but non-stable variance - let's say noticeable changes over 30 cycles - that will ultimately herald some failing packs / failing battery.

 

[Wolf]

So it's not just pack variance but non-stable variance - let's say noticeable changes over cell differenc30 cycles - that will ultimately herald some failing packs / failing battery.

That is my opinion also. We are going to have some variance in pack voltage that is a given. As is the case with Battery 1 it just so happens that the pack difference runs between 30mv to100mv at cutoff V. But it repeats consistently within that window. So I believe the early warning signs of a Pack/battery starting to fail is a larger pack voltage deviation than we are normally seeing. Also maybe this 6 month 24hr snapshot could give us an idea of a problem.
As you can see below the trendline is slightly upward granted it is only 4mv but if the trend does continue than that may be an indicator.
Just thinking out loud. Of course I have thrown a preverbal wrench in the works as I have added another Battery to the mix and discharge rates per battery will be now reduced by 1/4 (24.8%) Considering my max discharge for me is ≈60Amps that was ≈20 Amps per battery(3 of them)/ 250mA per cell (80 cell packs). Now its 188mA per cell which will no doubt change my charging and discharging profile.
Wolf

 

[Wolf]

Update:

So it looks like adding the 4th battery and an additional 750Watts of Solar panels has given me the preverbal sweet spot for now.
With the AC running I am still charging the batteries and so far it even looks like Battery 1 (Frankenstein) is starting to behave very well.
Looking at the V difference between packs of each battery I believe I am in a position of posing the least stress on the battery as a hole.
The V difference is really low 40mv or less throughout the C/D cycle. That low of a number could also just be the tolerances of the longmons voltage measuring circuit.
Wolf

 

[OffGridInTheCity]

Really nice to see the powerwall operating succesfully!

That low of a number could also just be the tolerances of the longmons voltage measuring circuit.

Interesting you should mention this. I've never been sure of the exact specs of a longmon in terms of it's accuracy. Here's a spec sheet from Batrium.com - https://s3.amazonaws.com/helpscout..../BMon-and-LMon-LeafMon-Data-Sheet-Ver-2.0.pdf And on page 4 I see this....

which is useless. A +/- specification of accuracy would give something for context and to verify.


For example, my #6 battery was off the the side temporarily and one of the packs (longmon) showed pretty wild discrepancies as in 70mv when it should have been 40mv. When I re-arranged things and brought that battery into the main battery bank/hookup - the issue disappeared. I'm not sure if it was because of the 'side hookup' instead of the permanent apposing corners for overall + and - take-off OR jiggling the longmon connections/network wires.

Without specs, even if they longmons show 0v difference it doesn't mean the packs are 100% balanced. My sense is <30mv is probably pushing the accuracy? but I don't know! And as discussed above, once the baseline is established the change over time (from the baseline) is a key metric.

 

[Wolf]

I've never been sure of the exact specs of a longmon in terms of it's accuracy

Ah @OffGridInTheCity, you must read further down the list and you will find your answer. Typical is ±0.05 Maximum is ±0.1.
In theory that could mean in a typical measurement we could be ±50mv and in a worst case scenario ±100mv off.
We might be a bit overkilling/overanalyzing this a bit. But in the long run we may be on to something.
If we have very accurate voltage measurements the possibility exists we could determine from that data a possible prefailure condition.

Wolf

 

[OffGridInTheCity]

Ah ha... (you're right! didn't read far enough) but based on this discussion I found this link - https://support.batrium.com/article/291-calibration-of-distributed-cell-monitors - which says...

What exactly does ".... valid to 2 decimal places per our specification." mean? I'm not getting a clear understanding in my brain in context of their published +/- 0.05v typical and 0.1v max variance specification. Does this mean if their reference voltage is 3.50v then the mon could be off +/- 50mv typical and as much as 0.1v max in the user environment?!?

One implication is one may be wary allowing charge above 4.1v as this might actually be 4.2v.

And it's interesting they disabled customer calibration - it makes sense as there's diminishing returns to obsess on 1mv when the mon isn't capable of measuring/reporting at that granularity at operating voltage and temp ranges.

 

[Cherry67]

Hi Wolf, whats Up? See you still going? I May have been gone lost to the LiFePO Guys....