"Copper/nickel sandwich" buses for series connections

Since I'm building a 14s6p pack with 21700 cells (Molicel P45B) that can deliver up to 180 A, I'm trying to find the best solution for welding. On my CNC mill I can cut almost any shape and so I made holes and slots into the copper.
IMG_2432.JPG
Sandwich welding the slots works good with 0.15 mm copper and 0.15 mm nickel, where 100 joule are necessary to get a strong connection.
IMG_2427.JPG
But now I try building the busbars the other way. First weld the nickel strips to the copper, then turn it around and weld the nickel directly to the cell. Much less energy is needed, 20 joule or less per point. This should work with thicker copper, too. Another advantage is, you can't miss the pole when welding.

IMG_2431.JPGIMG_2430.JPG
Has anyone experience with building this kind of busbars by themself and the currents they can handle? I had contact with a Chinese producer, but manufacturing and shipping is quite expensive, because of my custom layout, so I'd like to build it by myself.
 
Last edited:
Looks good...
IMG_2436.JPG
0.15 mm copper (= 20 mm²) + 0.2 mm nickel/steel. Hope this is enough for 150 A.
Welded with 50 J (kWeld + kCap).
 
Last edited:
I thought getting them already made was pretty cheap. These prices are for 270 cells with .3mm copper and including shipping. Came quickly.
 

Attachments

  • IMG_0438.png
    IMG_0438.png
    290.9 KB · Views: 31
Maybe I've asked the wrong seller, but the answers came very slowly and they did not understand well, what I wanted. Finally the prices were much higher ($90 for 84 cells without shipping).
The battery I'm building currently is a sample. If it works and I know the final layout, I will try to order the busbars, because then I'd like to build a few more batteries.
 
Last edited:
Has anyone experience with building this kind of busbars by themself and the currents they can handle? I had contact with a Chinese producer, but manufacturing and shipping is quite expensive, because of my custom layout, so I'd like to build it by myself.
Yes: Fiido Q1S Upgrade; 200A, VESC, Copper-Nickel 52v Molicel

I also did a CNC version as well, but I don't seem to have a thread about it here.

fiido-batter.jpg

As to how much it can handle -- a lot. Most of the time I see people throw out numbers it's typically 3-4x what you'd get out of nickel, and considering you are laying down a sheet instead of strips, it's going to be even higher.

Given the 6p config it looks like you're using, I'd usually rate a nickel config at a max of 40-60A if you had the 6 nickel strips connecting across the series. If we work off a 50A estimate, then multiply that by 4 -- somewhere in the realm of 200A. The limit is, in this case, far more likely to be what your cells can handle outputting -- cells which could each handle a 33A output are pretty much only the top tier cells these days.
 
200 A, 11.2 kW, for a few seconds, no problem. The cells and the copper don't get significantly warm. The next tests will take a little more time, but before them I have to build a case.

201W.jpg
 
200 A, 11.2 kW, for a few seconds, no problem.
Not only for a few seconds. Yesterday, two flights with the battery were very promising. 8 min. total motor running time, always enough power and finally the cells had a voltage of 3.4 V each and the temperature was 50°C. They would have worked a litte longer. But this was my first real life usage and I didn't want to stress them too much. Better than expected for a battery that has been built for learning how to weld and gathering experience.
 
Did you cut that shape yourself
The copper and the HDPE frame are constructed with Fusion 360 and cut with my CNC router, so they perfectly fit together and with the diameter of the cells.
These will be the parts for another battery (14s5p Molicel P45B):
Copper+Frame.png
 
Last edited:
Hi all, I just joined so please forgive me if my etiquette isn't correct.
I've taken on quite the battery project for a motocross bike I'm building and wanted some help with the connection design, as well as gut checking my math and thought process.
I've already learned a lot from reading through this forum so thank you dearly.
Here are the specs:
-76v 90ah
-21s18p
-(378)New Samsung 50s cells (5AH, 25amp continuous, 45amp peak)
-450amp total continuous, 810amp peak output ability (461amp max drawn from 35kw qs138 v3 motor) (controller can take 50kw so I'd like the battery to be able to take that in case a more powerful motor that size is released at some point)
-Ant BMS with 420amp continuous, 1050amp peak output (can balance and limit output to a set amperage)

I purchased a Kweld for this project and can run it off of 2 12v truck batteries in parallel to reach the full 2000amps the kweld can (but shouldn't) take.

The picture shown below is a quick model I drew up color coded by parallel connections how they make sense to me. One has to be split across the 2 halves. (I have a lot more CAD experience than electrical so it helps me understand things)
As I see it there are 2 main options for me:
-Parallel connections using spot welded nickel strips. Big copper busses soldered to the nickel strips for series connections
-Copper plates used as 2 parallel and 1 series connection each (copper nickel sandwich)
Big questions:
-which method would work best?
-What sizes of material would I need for optimal results?
 

Attachments

  • Screenshot 2024-10-28 153941.png
    Screenshot 2024-10-28 153941.png
    341.5 KB · Views: 4
I would not do the first, solder, method. Too much heat into the cells. Even doing it super careful with max temp iron and super skills, you will still cook some cells making the whole thing unbalanced.
The nickel sandwich will work well for this application. Just make sure you size it correctly. You need more series mass than parallel. In fact keeping the parallel smaller on purpose can help it 'blow' a fuse. This is what Tesla does with it's thin wire to each cell connections seen on many earlier packs.

Also, you're design looks a bit funky....have you split one of the P groups? Avoid that if possible! If you absolutely have to do that, then that is the only time you would want more parallel mass to ensure good balance inside that P group.

Cheers
 
I would not do the first, solder, method. Too much heat into the cells. Even doing it super careful with max temp iron and super skills, you will still cook some cells making the whole thing unbalanced.
The nickel sandwich will work well for this application. Just make sure you size it correctly. You need more series mass than parallel. In fact keeping the parallel smaller on purpose can help it 'blow' a fuse. This is what Tesla does with it's thin wire to each cell connections seen on many earlier packs.

Also, you're design looks a bit funky....have you split one of the P groups? Avoid that if possible! If you absolutely have to do that, then that is the only time you would want more parallel mass to ensure good balance inside that P group.

Cheers
Thanks for the reply, I did split a P group. Once the BMS shows up I will see if I can nest it on one side and put the full P group on the opposite side. Otherwise I will make sure to beef up that P connection. The controller Im using limits me to 76v so I need to have an odd number of P groups :(

I'm having a hard time finding info on sizing the connections correctly. Thats something I'm hoping to get help with here.

Also, for the "fuse" approach to the parallel connections, I've seen a few ways to do that on here, what would be the best way to handle that for serviceability? If it does blow a connection I would want to be able to fix the battery as easily as possible.
 
General rule I like to follow is oversize your series connections as far as reasonable/practicable. I would recommend at least 50% more than you expect continuous. You also need to factor in the width of the links. If they are super wide, you won't need as much mass.
I actually kinda just answered this over in this thread, so will cross post from there:

As an example, I am currently building a 20S10P pack with double stacked rows for the parallel cells. I'm using Samsung 50S cells capable of '25A' (real world 10-15A) and the draw will typically be under 50A with potential peaks up to 65-75A.
I will be using 0.2x8mm nickel and I typically just say it has a max amp draw of 10A to keep it simple. If using 0.15 I say 7A.
If layering nickel I typically assume a loss of capability of about 2A per layer.

Since I will have 5 series 'links' per parallel group I simply intend to have 2 x 0.2 x 8mm layers of nickel. This gives me a conservative value of 18A per link or 90A total. This could theoretically peak to over 150A without issue. The point is, I've added a healthy margin, so I'm not worried about it!

In the past for different pack layouts where there are less series links or higher amp draw required I've simply added more nickel layers (sometimes up to 4 layers), or switched to a copper/nickel sandwich. Both work well!

For fusing, you need to make a call. Keep it simple and don't do proper fusing, or figure out some uber complex way to fuse. Doing it DIY will be a compromise either way.
As for replacing a cell, good luck! Basically, like above, it's super hard to make possible. Best bet is to use cell holders, but you're still going to end up having to tear off padding, heat-shrink, barely paper and rip of spot welds. Sucks no matter what.

Cheers
 
That is very helpful, thank you.
Based on that, if I did a copper nickel sandwich with .15x10mm copper (30 amps) and .15x8mm nickel (7 amps) I can pull 37 amps per link which would get me 333 amps in the best case scenario which is 9 links.
-Can I double up the sandwich to get 665?
-My not-ideal connections only have 2 cells in direct contact, 5 layer sandwich?
 
Apart from that time I ate a pounder I've never heard of a double stacked sandwich being done like that :p

From memory I've successfully layered nickel up to 4-5 layers high....as you get higher you have to increase the power of the spot welder. By the time I hit 5 layers I'm maxing out my JP welder. You can add more, but the welds will only hold to the few layers beneath, not all the way down to the cell.

As for layering multiple copper sheets, I've never seen it done. I honestly don't know if it would/could work. I'm guessing it would be a challenge for the welder to keep up. At the very least you would want to be very careful to align everything. Miss the cell center, or for get to place probes together and you'll have a fairly big blowout!

What you probably could do more easily is the standard copper/nickel sandwich, then layer more nickel on top of that. You could probably add 40-50 extra amps with a 4-5 nickel layers on top of the copper. Not sure if you could get more on there.

Cheers
 
Are you the titanium bike frame maker Blacksheep? Just curious
I am not, though that does sound fun. I am a mechanical/industrial design lead with a specialty in injection molded parts and electro-mechanical design. (I use CAD a whole lot)
If that knowledge is helpful to anyone here, I'd be happy to return the favor.
Please get/use a thermal imaging camera to capture/share the results.

Cheers
For sure,
I will also start a build thread for the bike and post detailed testing and progress so I dont clog up this thread more than is helpful
 
Back
Top