Discussion about a new battery build

[brainwashed]

For context, I'm in the EU.

I have been running a DIY 16S 125Ah battery for one year now with no issues. It's a very simple build, just a particleboard base plate, two wooden planks at each end with two M8 threaded rods for compression. There's a 2cm layer of extruded styrene (the one used for floor insulation) all around, with the top being removable. All this can be sat on a welded base with wheels. It's ugly but works. It stays in the attic where it's never above 35C or below 7C.



The top insulation is placed over everything and the JK-BMS just sits freely on top. The base with wheels is not pictured as it's part of a bigger frame which also hold the inverter and breaker boxes.

Some mistakes:
- the finished battery was too heavy (60-70kg) to get it alone into the attic. The foldable staircase is very narrow and unstable.
- I bought B-grade cells but didn't think of ordering at least one spare. One cell was significantly worse, got replaced under warranty, but had only 50% of usable pack capacity for a while.
- ugly
- used a JK-BMS which has its own protocol and cannot talk Pylontech or Deye or anything else

I've seen that nkon now has some B-grade 280Ah cells which work out to around 16c per Ah. Looked a bit and found out that the JK RESS (which they also carry) has support for Deye. So was thinking of attempting a new battery, which will sit in parallel with the old one. Even an 80A BMS could work as the load can be shared between those two batteries, would it not? At most I will be pulling 125A from the system, the old battery has a 150A BMS.
I might forego the insulation, doesn't seem to be needed and I am in the process of insulating my attic anyway. Instead maybe some nice wooden panels with wood screws, or any other ideas?
Should I build it as a 8+8 pack instead of an in-line 16S pack? I think it would be quite long otherwise. I already have an issue with the old pack where some balancing wires barely reach the BMS, so it can only be left on top or dangling. Dangling works better because my Raspberry Pi can then connect from another room via BT.
Should I go with JK RESS or are there any other JK options on nkon.nl that are able to talk to Deye? Can they be daisy-chained so the inverter can see the total capacity? This is not really critical, for now my Deye just has a hardcoded capacity and it's usually within 10% of the BMS one, plus conservative voltage limits.
Maybe better to build two separate enclosures sitting on top of each other? They would obviously need some removable connectors for balancing plus the thick welding cable instead of a bus bar.
Recommended to buy one or two spare cells?
For the initial connection was thinking of wiring the new battery in parallel but through a high-power 5-10 ohm resistor or a light bulb or something, for a few days.

I've looked at battery enclosures and they end up costing >500, sometimes more if you include shipping. In a DIY build, the 280Ah battery would end up costing less than 1000. If I go for A-grade and smaller cells I'm at the same price. A commercial 100Ah battery with nice enclosure costs around 670.

 

[Robbert]

With regards to your JK-BMS, I think you selected a great BMS. I Have JK as well and due to the incorporated balancer, my batteries (one approx. 300 cycles and the other one a bit more), they stay balanced pretty well.

If you go for another BMS, you might need a separate balancer.

Closed loop communication is nice, but not really necessary, as long ad your inverter cuts off, or changes its state at the right voltages.

I use Solar Assistant and Solar Assistant allows you to change the state of your inverter, depending on the SOC as well. Don't know if SA works with Deye as well, but interesting to look into that.

Wrt the battery itself, I would build just a 16s battery and not 2 8s batteries. It's heavy, but you don't want to move it after you finished it

 

[brainwashed]

Thanks for the input.
I was not planning to build two batteries but rather have two rows instead of a single long row.
After some thinking, I thought a moving trolley (with handle) would be a good base to build on. It's pretty cheap, has a sturdy floor, wheels and a metal frame. I only need to move the battery around a bit while working in the attic as there is still some (a lot!) of finishing to do. The old 50kg (~100lbs) battery can still be easily lifted and moved to one side, the new one not so much.
I use Home Assistant with my current JK-BMS but would like to be able to integrate it with the inverter. There is a project on this forum using an ESP32 and some components but never got around to it. Was just hoping a newer model can save me this hassle. My inverter has a slight offset on the battery current reading (0.4-1A) and it becomes quite off after two days without a full charge.

 

[hwy17]

Does your existing battery have cell separator insulators? You should use those.

The batteries will share current but I would still use a 150A BMS so that either battery can carry the load if the other one is shut down.

I have a battery in two rows of 8 and there's no problems with it for me. It makes compression hard if you want to compress because most materials that are that long would just bend under any force. I was going to use 1/2" aluminum plate for compression but I gave up on it and built with no compression.

 

[brainwashed]

I did not know about cell separators before but will order them for the new build. Since the old batteries don't move or rub, I didn't see a need for one, the existing insulation seems to do the job.
Yes, I will probably go for a 125A or 150A BMS, no sense in saving a few bucks there.
I would compress them only along the rows, so four compression plates in total. Space is not an issue so the rows can be spaced a bit to make room for the threaded rods. I can lift the old battery just by the compression plates, which are just solid wood panels. The force seems pretty evenly distributed on the height even with just one rod per side, but the new cells might be taller and require two rods per side. I know that there are special kits but those add an unjustified cost. Compression should not be skipped, it prevents the cells from bulging and the terminals from receiving mechanical load.
There are machines that can tie PP or PET bands, like those used for wrapping heaving cardboard boxes, perhaps that's also an idea. It would still require a load spreader (plate) at the ends so the tape does not dig into the cells ( https://www.amazon.de/-/en/Manual-Strapping-Tensioner-9-19mm-Ratchet/dp/B07ZX1L27Q )

 

[hwy17]

Check if separators are included with the cells, they often are nowadays.

If you space the rows apart you will need a longer bus bar to span between them, and in theory this could affect the BMS voltage reading between cell 8 and 9 if it measures 16 all in one series. My BMS measures 8 and 8 so it wasn't a concern for me.

 

[brainwashed]

Thanks for the heads-up. AFAIK, the bus bars that come with the cells are already longer than needed when strapping side-by-side, but could only see that when they get here. If not, I could create one from copper or even galvanized steel. The thick 35sqmm (AWG 2) wire doesn't make much sense for such a short distance. JK-BMS measures all cells in parallel, at least according to my testing. That's also how it performs the active balancing, by temporarily charging a capacitor and discharging it across that single cell. At least that's my assumption.