NMC Bank, Bus Bars and Batrium Love

[sunshine_eggo]

It has been years in the making, but I finally have the NMC battery poised for deployment... (okay, only half, but still).

Summary:
7X Ford Fusion/C-Max Energi PHEV batteries, 7.6kWh rated per pack.
Each battery: 84S 25Ah 3.7V NMC cells. Four subpacks of 21S in series
Packs have tested 19-24Ah
28 total subpacks (modules), packs of 4 pictured below:



Config:
21S modules disassembled, cells flipped to the same polarity and reassembled into a 21P module.
Cells connected with custom Aluminum bus bars (.25" x 1"), 21 holes each, 2 per module, 28 bus bars, 588 holes.
Each 21P module in series with other 21P modules for a total of 14S.

21P module with bus bars fit:


So far, I have 14 of the 28 modules ready to go - 1/2 the bank and enough to build a single 21P14S battery. That's 48V, Baby! 26.6kWh rated and about 23kWh actual based on tests.

note that these 14 modules have roughly equal numbers of cells from each pack such that each 21P module has almost identical capacity to the other modules in series.

No. I shouldn't have used copper. Raw materials alone would have been $2000 for copper bus bars with no holes. The Aluminum bus bars were $995 TOTAL with the holes drilled in them.

Anyway, the biggest challenge has been being remote. With everything at the site, but me only there ~12-18 times a year, it turned out to be nearly impossible to put in the many hours of work needed to get it done, plus my used T-1275 4S2P 48V bank was more than meeting our needs... wanted to enjoy that shit too! I brought much of it back and started working it at home.

Batrium acquired for Victron interface and cell data logging. I've been intimidated by the Batrium from the start. There's no manual per se, but a crapton of info scattered about. What I finally figured out is that everything is in videos, and if you really dig in, you can simply queue up a series of videos and get everything configured.

TRIAL RUN:
With everything pretty much ready to install, I didn't want to get up there and realize that I was missing something, or I couldn't figure something out, etc., so I did a trial run with the Batrium on an as-pulled module using 14 out of the 21S cells:



Yes, it's a mess, but I was going for function.

Success:


Keep in mind that this is NMC, NOT LFP. Voltage and SoC have a strong correlation with this chemistry, so balancing at any point in the voltage curve is worthwhile. My own charging efforts with a hobby charger capable of 7S 3.7V balancing got the cells to within 3.80-3.82. I tested the balancing function by setting it to 3.80, and it only took about an hour to pull them all down to 3.80V at a max of 0.5A.

The cool thing is it documented how much was required to balance them:

Cell	Date	Time		Session		Amp	MinV	MaxV
1​	7/1/2022​	1:06:43​	AM	336.5​	mAh	0.00A	3.80v	3.80v
2​	7/1/2022​	1:06:43​	AM	254.5​	mAh	0.00A	3.80v	3.80v
3​	7/1/2022​	1:06:43​	AM	213.1​	mAh	0.00A	3.80v	3.80v
4​	7/1/2022​	1:06:43​	AM	394.2​	mAh	0.00A	3.80v	3.80v
5​	7/1/2022​	12:39:20​	AM	309​	mAh	0.00A	3.80v	3.80v
6​	7/1/2022​	12:39:19​	AM	333.1​	mAh	0.00A	3.80v	3.80v
7​	7/1/2022​	1:06:43​	AM	240.6​	mAh	0.00A	3.80v	3.80v
8​	7/1/2022​	1:06:43​	AM	221.4​	mAh	0.00A	3.80v	3.80v
9​	7/1/2022​	1:06:43​	AM	82.1​	mAh	0.00A	3.80v	3.80v
10​	7/1/2022​	12:39:19​	AM	4.6​	mAh	0.00A	3.80v	3.80v
11​	7/1/2022​	12:39:20​	AM	177.3​	mAh	0.00A	3.80v	3.80v
12​	7/1/2022​	1:06:43​	AM	410.4​	mAh	0.00A	3.81v	3.81v
13​	7/1/2022​	1:06:43​	AM	589.4​	mAh	0.48A	3.81v	3.81v
14​	7/1/2022​	1:06:43​	AM	230.8​	mAh	0.00A	3.80v	3.80v

This was taken from a snapshot created every 15 minutes. It was towards the end after all but #13 had been balanced.

I'm running the software on a Mini PC. It seems to handle it just fine, and it only uses 18W peak power based on the power brick output rating. I still have a lot to figure out about logging, but it looks like I can get there. Remote access to the PC appears to work well too.

Lessons learned:

1. My mouth/brain readily write checks my ass can't cash. Time estimates were woefully inadequate, and my ability to get shit done is a small fraction of what I think it is.
2. Remote projects don't get done. This first half of the battery took the better part of 3 years. There were massive flurries of work spaced out with long periods of inactivity in between. The second half will likely be done faster (note how I didn't give an actual timeframe... I'm already learning).
3. LFP would have been so much easier, faster and a little less in cost. I was fully committed with 5 of these packs before I became aware of any of the Amy's, EVE or their 280Ah cells. I would have been done two years ago in a day.
4. Batriums MUST have a PC attached to log/access data.
5. Drilling 21 holes in a .25" Al bus bar isn't easy. It's very time consuming. I did 7 bus bars, and it took me about an hour each. I contracted with an acquaintance with an end mill. He was able to do them 7-8 times faster (30 minutes to do 4 stacked) and so much more precise. I paid the equivalent of about $14/hour of my 21 hours it would have taken to finish, and I got 21 hours of my life back.
6. #5 is simply a "Right tool and right person for the job" lesson. Just because I can do something doesn't mean I should.

I'm sure there are a lot more lessons in there, but over three years, they've been lost to the dead neurons.

Next week will include a post documenting success or a magnificent fire in a shipping container.

Thanks for reading/skimming/laughing quietly to yourself, etc.

 

[StuartV]

What does NMC stand for? ? Nickel Metal Cadmium?

 

[toms]

What does NMC stand for? ? Nickel Metal Cadmium?

Nickel Manganese Cobalt.

Great write up on your adventures Sunshine Eggo, i think you’ve done a great job.

 

[glamsland]

Nice!

So, nmc should ballance in the whole curve? Interresting. I have only ballanced in the upper voltage, like in float voltage . Then the cells will be ballanced while in "rest".

 

[sunshine_eggo]

Nice!

So, nmc should ballance in the whole curve? Interresting. I have only ballanced in the upper voltage, like in float voltage . Then the cells will be ballanced while in "rest".

There isn't necessarily value in it, but it actually does something. Putting NMC cells to the same voltage puts them at near identical states of charge. Of course, if the cells have different capacity, that doesn't get you much.

In my case, with tested cells and interleaved modules, I'll be using both active balancers for full range SoC balancing and the Batrium to assist at the top if needed. I will be running these cells in a very narrow range - 3.50-3.92V for absurd cycle life.

 

[Horsefly]

Great report Mr Eggo!!! Your lessons learned is much more honesty than I is easily found on this or any other forum. Wish I could buy you a cold one and have you walk through the details of this with me.

 

[sunshine_eggo]

Nickel Manganese Cobalt.

Great write up on your adventures Sunshine Eggo, i think you’ve done a great job.

Thanks!

 

[robby]

Really nice Job.
I know how satisfying it is to have a long term project come together and work just right.

 

[sunshine_eggo]

UPDATE:

Stage one complete:


Yes, it's a shitty rat's nest. Unfortunately, many of the BMS leads are too short, and I had to locate things as best I can. Active balancers installed and transferring about 1A to #3.

In case anyone is worried, there's a class T fuse and master cut-off out of frame.

Took several hours to buff the aluminum and apply the appropriate anti-corrosion grease. Yes. I actually torqued all 588 5mm nuts to 60 in-lb. No. I didn't skip any of them. Another couple hours for the interconnects and main leads, and finally...



Yes, I can't find the temp sensors. They're probably at home on my desk. It is NOT -24C here...

Yes, cells 3 and 5 are wonky. 5 was the one freshly charged and brought from home, so it was right where I expected it. Not sure why #3 is out. It's possible I didn't charge it 3 years ago... Not that I've ever been inconsistent. I was worried a dead cell was pulling the module down, but all 21 cells are pleasantly cool to the touch.

Once the evening activities have concluded (done with A/C and sous vide in the Instant Pot), I'll break the battery down enough to where I can parallel #3 and #5 as it's going to take forever to bring that in line with 441Ah of total capacity. I'll use a clamp ammeter to make sure current doesn't get out of hand. These individual cells can handle 150A, so it shouldn't be a big deal to push 50-100A total through the wires I have for the job.

Speaking of evening activities... very cloudy day with sad solar. Witnessed a 2900W drain, and the pack voltage only dropped by 0.3V. Been running A/C, instant pot and ~200W background loads for a couple hours, and voltage is only down 0.7V.

So... no fire yet, AND no BMS communication with the Victron. It's going to take me another couple hours digesting the config info to make sure I have it right. I have to properly configure the BMS, the GX and ESS on the Quattro...

As it stands, I am currently the BMS safety cut off...

 

[sunshine_eggo]

UPDATE 2:

Wired 3 and 5 in parallel for about an hour. Started at 82A and ended at 28A. 0.1V difference remains:



Managed to pull #5 below the others.

For the life of me I can't get it to talk to the GX. The instructions are clearly dated against both the Batrium and the GX. The instructions on the ESS/inverter side make zero sense, and the inverter shuts off due to low voltage... at 53V.

I can see the GX receiving data in the CAN bus, but nothing's talking. I've triple checked the cable and pinouts and even tried connecting the ground (indicated to be a last resort).

Oh well. I'll scour Google and try again tomorrow. Charging with the generator while I watch paint dry (cell voltages).

 

[JRH]

It has been years in the making, but I finally have the NMC battery poised for deployment... (okay, only half, but still).

Summary:
7X Ford Fusion/C-Max Energi PHEV batteries, 7.6kWh rated per pack.
Each battery: 84S 25Ah 3.7V NMC cells. Four subpacks of 21S in series
Packs have tested 19-24Ah
28 total subpacks (modules), packs of 4 pictured below:

View attachment 100869

Config:
21S modules disassembled, cells flipped to the same polarity and reassembled into a 21P module.
Cells connected with custom Aluminum bus bars (.25" x 1"), 21 holes each, 2 per module, 28 bus bars, 588 holes.
Each 21P module in series with other 21P modules for a total of 14S.

21P module with bus bars fit:
View attachment 100870

So far, I have 14 of the 28 modules ready to go - 1/2 the bank and enough to build a single 21P14S battery. That's 48V, Baby! 26.6kWh rated and about 23kWh actual based on tests.

note that these 14 modules have roughly equal numbers of cells from each pack such that each 21P module has almost identical capacity to the other modules in series.

No. I shouldn't have used copper. Raw materials alone would have been $2000 for copper bus bars with no holes. The Aluminum bus bars were $995 TOTAL with the holes drilled in them.

Anyway, the biggest challenge has been being remote. With everything at the site, but me only there ~12-18 times a year, it turned out to be nearly impossible to put in the many hours of work needed to get it done, plus my used T-1275 4S2P 48V bank was more than meeting our needs... wanted to enjoy that shit too! I brought much of it back and started working it at home.

Batrium acquired for Victron interface and cell data logging. I've been intimidated by the Batrium from the start. There's no manual per se, but a crapton of info scattered about. What I finally figured out is that everything is in videos, and if you really dig in, you can simply queue up a series of videos and get everything configured.

TRIAL RUN:
With everything pretty much ready to install, I didn't want to get up there and realize that I was missing something, or I couldn't figure something out, etc., so I did a trial run with the Batrium on an as-pulled module using 14 out of the 21S cells:

View attachment 100872

Yes, it's a mess, but I was going for function.

Success:
View attachment 100873

Keep in mind that this is NMC, NOT LFP. Voltage and SoC have a strong correlation with this chemistry, so balancing at any point in the voltage curve is worthwhile. My own charging efforts with a hobby charger capable of 7S 3.7V balancing got the cells to within 3.80-3.82. I tested the balancing function by setting it to 3.80, and it only took about an hour to pull them all down to 3.80V at a max of 0.5A.

The cool thing is it documented how much was required to balance them:

Cell	Date	Time		Session		Amp	MinV	MaxV
1​	7/1/2022​	1:06:43​	AM	336.5​	mAh	0.00A	3.80v	3.80v
2​	7/1/2022​	1:06:43​	AM	254.5​	mAh	0.00A	3.80v	3.80v
3​	7/1/2022​	1:06:43​	AM	213.1​	mAh	0.00A	3.80v	3.80v
4​	7/1/2022​	1:06:43​	AM	394.2​	mAh	0.00A	3.80v	3.80v
5​	7/1/2022​	12:39:20​	AM	309​	mAh	0.00A	3.80v	3.80v
6​	7/1/2022​	12:39:19​	AM	333.1​	mAh	0.00A	3.80v	3.80v
7​	7/1/2022​	1:06:43​	AM	240.6​	mAh	0.00A	3.80v	3.80v
8​	7/1/2022​	1:06:43​	AM	221.4​	mAh	0.00A	3.80v	3.80v
9​	7/1/2022​	1:06:43​	AM	82.1​	mAh	0.00A	3.80v	3.80v
10​	7/1/2022​	12:39:19​	AM	4.6​	mAh	0.00A	3.80v	3.80v
11​	7/1/2022​	12:39:20​	AM	177.3​	mAh	0.00A	3.80v	3.80v
12​	7/1/2022​	1:06:43​	AM	410.4​	mAh	0.00A	3.81v	3.81v
13​	7/1/2022​	1:06:43​	AM	589.4​	mAh	0.48A	3.81v	3.81v
14​	7/1/2022​	1:06:43​	AM	230.8​	mAh	0.00A	3.80v	3.80v

This was taken from a snapshot created every 15 minutes. It was towards the end after all but #13 had been balanced.

I'm running the software on a Mini PC. It seems to handle it just fine, and it only uses 18W peak power based on the power brick output rating. I still have a lot to figure out about logging, but it looks like I can get there. Remote access to the PC appears to work well too.

Lessons learned:

1. My mouth/brain readily write checks my ass can't cash. Time estimates were woefully inadequate, and my ability to get shit done is a small fraction of what I think it is.
2. Remote projects don't get done. This first half of the battery took the better part of 3 years. There were massive flurries of work spaced out with long periods of inactivity in between. The second half will likely be done faster (note how I didn't give an actual timeframe... I'm already learning).
3. LFP would have been so much easier, faster and a little less in cost. I was fully committed with 5 of these packs before I became aware of any of the Amy's, EVE or their 280Ah cells. I would have been done two years ago in a day.
4. Batriums MUST have a PC attached to log/access data.
5. Drilling 21 holes in a .25" Al bus bar isn't easy. It's very time consuming. I did 7 bus bars, and it took me about an hour each. I contracted with an acquaintance with an end mill. He was able to do them 7-8 times faster (30 minutes to do 4 stacked) and so much more precise. I paid the equivalent of about $14/hour of my 21 hours it would have taken to finish, and I got 21 hours of my life back.
6. #5 is simply a "Right tool and right person for the job" lesson. Just because I can do something doesn't mean I should.

I'm sure there are a lot more lessons in there, but over three years, they've been lost to the dead neurons.

Next week will include a post documenting success or a magnificent fire in a shipping container.

Thanks for reading/skimming/laughing quietly to yourself, etc.

I love it but honestly I can’t relate as I would wish too as it is so insanely over the top… looks like somthing on a video about stuff down at Boca Chica… does it fly.? Is it Gona land and be reusable.?.. it’s a beast…..!

 

[turbo6bar]

If you had to do over again, would you use the Batrium system? On secondstoragelife several Batrium users are losing love for Batrium for quirky faults and poor compatibility with older Batrium hardware.

I will be building a powerwall using LG pouch cells later this year, and I'm waffling on the BMS. The flavor of the month Chinese BMS is JK with the active balancing, but I'd be a fool to not consider REC and Orion.

 

[JRH]

If you had to do over again, would you use the Batrium system? On secondstoragelife several Batrium users are losing love for Batrium for quirky faults and poor compatibility with older Batrium hardware.

I will be building a powerwall using LG pouch cells later this year, and I'm waffling on the BMS. The flavor of the month Chinese BMS is JK with the active balancing, but I'd be a fool to not consider REC and Orion.

Hey guy , I’m probably not the right guy to ask on that topic.. I’m still trying to get all the battlebornes and Victron stuff paid off of my PayPal acct.. and also not get electrocuted running a 100 volt system…so far so good with my new build ….
im sure there are some of these really smart guys who would be better served on this topic for your question.
but thanks …
Jim.

 

[sunshine_eggo]

I love it but honestly I can’t relate as I would wish too as it is so insanely over the top… looks like somthing on a video about stuff down at Boca Chica… does it fly.? Is it Gona land and be reusable.?.. it’s a beast…..!

LOL... me too. Keep in mind that I'm only halfway done... I have a whole 'nother battery to build/deploy.

If you had to do over again, would you use the Batrium system? On secondstoragelife several Batrium users are losing love for Batrium for quirky faults and poor compatibility with older Batrium hardware.

I don't know. Orion was near the top of my list, but initial configurations and voltage limitations on the Jr turned me away. It would have worked in this config.

FINAL HOLIDAY WEEKEND UPDATE:

No fire.

Yep. It took quadruple checking of the cable, but I had the CAN high and low reversed. Talking to the GX now.

Batrium requires the use of Victron ESS to exert control. Unfortunately, ESS is designed for grid interaction and is a whole bag of worms that has all kinds of things that need to be configured. At this time, I can't charge with the generator. It will power loads, but it won't charge. More learning to do.

My 3kW array was experiencing some cloud edge effects and ended up putting out about 3400W for 5 minutes... popped the 30A breaker into the MPPT that was a first.

I'm not confident the BMS can control the GX. Today it hit the very conservative voltage limit and kept going (3.97V/cell). I didn't stick around to find out and just disabled the MPPT. I even tried lowering the voltage below current to force a trip, and it didn't happen.

I can change the configuration remotely, so there's hope I can get it sorted, and a friendly neighbor that can be hands-on if I screw things up.

Overall grade: D

Why?

1. I'm basically running unprotected. I'll have to watch it daily to ensure it doesn't go over limit. I am the BMS.
2. I'm also concerned about #3. Rather than closing the gap, it may be widening. 0.02V is noise at this point, but if I don't see that slowly close with active balancers AND the Batrium passively balancing all other cells, I'm going to replace that module.

 

[curiouscarbon]

thank you for update. positive hopeful wishes for this install!!

 

[sunshine_eggo]

Still no FIRE!!!

I've made incremental progress, and I've establised that there is at least some communiation between the WM5 and the MPPT.

I can't remember the specifics, but it required changes to the default ESS parameters on the GX (ESS is a system used for grid interaction (AC coupled PV, peak shaving, etc) and external control). It also required DVCC to be enabled.

Having been an avid user of DVCC, I had always had that enabled for other benefits, but once ESS was installed, DVCC ON shut my inverter down. Bastid! I traced that back to one of the ESS settings. Once DVCC was enabled, the MPPT started reacting to different BMS states.

I changed the peak cell charge value to below the current peak cell, and I observed the MPPT pull back to zero charging. The cool part was it didn't cut the MPPT entirely. It keep the MPPT running powering the loads, though this may be undesirable in a situtation where you want to bleed the battery down off the cliff...

Since I splurged for the expansion board, I am going to install control wires to the MPPT to do a hard cut-off on charging above a certain point. I just have to figure out how to do that next trip.

In addition to configuring the system to operate within the desired operating range (20-80% SoC, 3.5-3.92V), since the BMS doesn't cut out like a FET BMS, I'm going to program the BMS to be actively involved in charge management... it's pretty much designed to do that. It even has charge termination criteria based on balancing, i.e., it can be setup to stop charging/balancing based on multiple criteria.


Speaking of balancing, the gap has narrowed to .07-.08V. I've been actively adjusting the threshold of balancing to ensure the 13 higher cells are being hit with the burn. Fortunately, I had a 5V USB fan that I was able to power via the expansion board's +5V output. It's cooling the heat sink, and it's prevented the balancers from shutting down at 75°C. I don't think I've seen anythign higher than 65°C.

I know I mentioned it earlier, but it records how much has been bled off in balancing... Here's snapshot cell data showing total mAh bled off from each in a little more than two days:




I accidentally left the voltage too low during charging and managed to also bleed off almost 1Ah from the low cell, but as you can see, it's balanced 22-25Ah out of the other cells. That suggests I'm at least 5% out of balance plus whatever it takes to get everything bled down to #3.

This doesn't take into account any of the active balancing, but I doubt they're transferring more than about 0.2-0.4A between cells, and they'll stop trying at 0.01V difference.

I'm still not convinced #3 isn't losing charge due to a faulty cell. I won't be until they've top balanced, and I can cycle them without the Batrium balancers.

Anyway, thanks for coming along. once I get everything sorted, I'm going to post comprehensive instructions on what this took to happen.

 

[toms]

Thanks for taking the time and effort to share, well done ?

 

[sunshine_eggo]

To finalize the answer, yes, I would choose the Batrium again. I may change my mind after trying to get the second half online with the MM8, but I doubt it.

My problems were not at all associated with the Batrium but with the configuration of Victron ESS for off-grid operation (almost never used for off-grid) and Victron DVCC. I will fault Batrium for poor documentation, but I hope to supplement that if they'll accept some revisions.

I'm getting more confident in the condition of #3. Using linear assumptions and approximations for SoC vs. voltage, each .05V represents about 7% variation in SoC or about 31Ah.

As it stands, #3 lags behind the high cell by 0.045 to 0.05V, and it's been consistently better each day. Here's what the Batrium reports for mAh burned off in balancing since 7/4 sorted from low voltage to high:



Note that the "FinalBypass" flags are cycling through all but the first cell (#3), and they are applying 0.49A continuous drain on each cell. The snapshot just shows what was displayed in that moment.

That's about 9% of the total 441Ah capacity.

Looks like I'm around 2/3 of the way there.

These are just sanity checks and are likely confirmation bias as much as anything else, but it does make me hopeful. I'll not consider #3 healthy until it keeps pace with the others after balancing at 3.92V. Once balanced, I plan to stress test it with my absorption fridge running on AC overnight. That should pull it down nicely and let me know how unstable the balance may be.

I did get an interesting piece of info on the active balancers. Even when strung together, they really don't work except between adjacent cells. Furthermore, they will only output their 4A rating when the difference is a full 1V, and it's directly proportional. I was seeing about 1A when I first hooked it up with a 0.24V difference. Once things got down to 0.12V, I was seeing right around 0.4A.

At this point, the balancers aren't doing much of anything - maybe 0.1-0.2A at most.

I'm pretty tickled with the performance. I can hit it with a 500W load, and the battery voltage barely budges. The MPPT is almost sluggish in its response. With the old lead acid, the drop was immediate and the MPPT response was immediate to compensate. Now it's like there's a MPPT ramp up, lagged peak and then a ramp down that finishes several minutes after the load is removed.

Next time I'm on site, I'm going to implement remote switching of the MPPT and Inverter with the expansion board relays for another layer of protection.

 

[sunshine_eggo]

Dang it... VRM just keeps giving me reasons to love it.

The BMS passes a few bits to the GX that get recorded on VRM. This is very encouraging to see it graphically:



The high peaks are coming down and the low peaks are coming up.

This hasn't been a "set it and forget it" situation. I've been changing the balancing voltage to ensure all but #3 are getting burned off as much as possible. I adjust it 5+ times/day as they move up and down. There's usually a point at night when all cells drop below the balance voltage, and I have to adjust it when I wake up. Then during charging, I have to adjust it up to keep from accidentally burning off #3 too. I've missed it a few times as you can see #3 has burned off > 1Ah in the post above.

I'm hoping that a couple days from now I can just leave balancing at 3.92V and call it good. After a couple days/cycles of balanced operation, I'll disable balancing and see if it drifts.

FWIW, I used to think about 0.05V was an acceptable deviation on 3.7V chemistry. I still think it's "okay" provided you're not operating anywhere near voltage limits, but between this exercise and other experiences with hybrid vehicle Lithium batteries, that number has narrowed to 0.02V. Roughly speaking, that's about 2-3% difference in SoC.

 

[curiouscarbon]

nice graph! ?