Common pack design mistakes, how to avoid?

[tsourorf]

Oh, that was for that pack? I hate to say but you've missed a few things there.

First of all, the 20mOhm IR is rather optimistic but I wouldn't care about it as it doesn't affect the differences between cell current contributions anyway.

The most important thing is that you haven't put any resistors for the nickel connecting the 7 cell row. I don't know the thickness of that, so, I wouldn't suggest a number for the resistance to put between them.

Then, there is the copper wire resistance. It's 1mm wide. That makes the surface approximately 3.14mm2 which, according to the known Matador table gives us roughly 5.3mOhm per meter. That translates to 95.4uOhm for the sections you have 40uOhm. Then, there are the lengths or copper at the end... From the pictures, I can see that the middle wire is approximately 15mm long and the side wires (until they reach the first cell) are approximately 15+18=33mm long. I would use 80uOhm for the middle section and 175uOhm for the side ones.

I'll be honest, I haven't changed your model to include the things above, so, I don't know what the impact would be. All in all, I expect a small change in the variance of cell current contribution but, because of the use of copper, I wouldn't expect any huge differences.

[Punx0r]

The cell IR does make a difference - the lower it is, the greater the effect of the interconnect resistance on current sharing.

[tsourorf]

If it's the same for all the cells, it wouldn't make any difference regarding the current contribution of each cell to the pack. To make what I mean obvious, if in a 5P pack, each of the cells contribute the following to the total current: 100%, 95%, 87%, 75%, 60% (Of their maximum contribution capability)... the percentages wouldn't change if the IR of each cell changed from 20mOhm to 100mOhm.

[Buk___]

The most important thing is that you haven't put any resistors for the nickel connecting the 7 cell row. I don't know the thickness of that, so, I wouldn't suggest a number for the resistance to put between them.

Look again. Each of the 6 gaps between the 7 cells has a nickel strip connection, but that is overlain by the copper cable and a large blob of solder.

If you consider the anodes to start with. In the middle of the cells where the curvatures of two adjacent cells come together, you have an approximate 2mm gap between the edges of the nickel-coated steel cans, bridged by 1mm^2 of copper, surrounded by solder:

NickelCopperSolderBridge.jpg (3.96 KiB) Viewed 4198 times

There is a somewhat bigger gap at the cathodes.

The math gets complicated as both sides of the can recede from the center on both sides (plan view), but in every case, when I calculated it (using a small sample of the resistivities of various solder mixes), the total resistance of the bridge was substantially less that that of the nickel-coated steel can. So, whilst not zero, relatively negligible.

Then, there is the copper wire resistance. It's 1mm wide. That makes the surface approximately 3.14mm2 which, according to the known Matador table gives us roughly 5.3mOhm per meter. That translates to 95.4uOhm for the sections you have 40uOhm.

Joachim's description says 8awg wire, which has a resistivity of 2.061mOhms/m. That translates to 0.002061/1000*20mm = 0.00004122 or 41.2u.

By all means, adjust to your judgement of the lengths involved. That's why I posted the model.

[Joachim]

The fuse&bms can handle max 80A.
I wil use this pack on a hobby motor-compressor that take maximum 45Amps.

[Buk___]

The fuse&bms can handle max 80A.
I wil use this pack on a motor-compressor that take maximum 45Amps.

I like the way building a pack + coper serie bars and no soldering on top of the cells....this is the way i can make a save strong pack ,cells in holders.The coper wire are on top of the serie nickel and not in the midel of the bulb.
Test the pack this weekend en check voltdrop/temps on the cells and serie connections.

Seems like it works brilliantly.

[Joachim]

Test the pack low temps 5 degrees....@18 degrees wil perform better.
Peak 45A sag 0,6V and continuous 20A 0,2Volt sag.
No heat on the series after 15minuts wurks fine for me

[Headrc]

Wow, nice build there. What spot welder are you using to weld copper?

[Joachim]

Wow, nice build there. What spot welder are you using to weld copper?

not copper...0,20x8mm nickel for a part of the parallel connections.

[Headrc]

Ok, thanks. I would still like to know what spot welder you are using.

[Joachim]

Ok, thanks. I would still like to know what spot welder you are using.

i use cheap welder

[Headrc]

Looks like a nice setup, I have not seen that welder before I don't think. What is the manufacturer name?

[Swe]

Can I make a Y-split with my 10 awg cable like in my picture? It is not exactly 50/50 divided, just what my eye thought was okay.
Or must I solder two 10awg together to make a thicker Y? I will run short peak of 65A. But maybe more like 30A cont. Since they are connected with each other again by the bolts I was thinking you could still see it as one cable so the current should not be too much even if I make a 60/40% split instead of 50/50. But if you read here it seems that every small extra resistance (the connectiobs between the bolts) makes huge different.
Under the black cap there is a nickelplated copper bus bar as wide as the black cap/cover.

E7E120E6-1D42-498E-90CE-531CF5FB386A.jpeg (53.22 KiB) Viewed 3997 times

[garolittle]

....Then he holds the hot tip onto the negative electrode as an example for about 20 seconds to get a solid solder connection.

I literally cringed as I watched that part.

[Lurkin]

I'm trying to put my first pack together this weekend. Any chance of some feed back re current sharing etc? the framing represents the cell spacers I have and I will be using nickel strip spot welded to each cell. I'm still debating the number of layers of nickel strip vs soldering on copper wire or attaching a bus bar - not sure what the current limits are on this.

This pack will be made with NCR18650GA cells, supposedly rated for max 10a per cell with 8 in parallel - so supposedly good for 80a. I chose these cells for their capacity at the time, not their discharge rate - they will never see that level of discharge. More likely to be in the order of drawing 10 - 20a from this pack at any given time.

Any feedback would be much appreciated.

[hemo]

This is nice Buk! Basically, you are using copper to minimize resistance and trying to keep equal distances from root to each parallel cell. I say it's perfect enough for pragmatic applications.

After the last complicated 16p pack I've made, I though I should try a much easier and possibly better way of connecting my cells. I have ordered a 0.9mm thick C101 copper sheet to cut to pieces that cover all the parallel ends. Then spot weld 2cm long 7x0.15mm nickel strips on the cell ends. Cut through the copper right where the nickel strip is, to let the nickel strip go through the cut. Then fold the remaining nickel strip (Approximately 1cm long) on top of the copper that lies on top of the parallel cells and spot weld again! This should practically give me no resistance (ok, it's just minimal) from the wire to each cell because the current will go through the whole copper until it reaches a very small portion of nickel and then go though the cell.

Once I receive the copper sheet I've ordered and try it, I'll post pictures.

Similar approach to the mock up MC applied on spot welding thread;
viewtopic.php?f=14&t=68005&start=325

Did you got through and finalise the build design ?

[hemo]

A UK 5 year old supplied China Dolphin 09 battery pack, another poor series connected pack,
Only a 10s 5p of 29E but only 5 series buss connector on each side marked with a dot.
All buss are 2 layers of nickel 0.15 0r 0.20mm.

DSCF0928_LI.jpg (53.15 KiB) Viewed 2852 times

[derekisastro]

This pack good for 80A max discharge 1mm pure copper wire for the series 6x
The pack:3s 28P 10.8V 81Ah 8AWG wiring no direct heat to the cells on soldering the copper serie/parallels

I can't see Joachim's build ... did he edit it out?

[derekisastro]

So, the configuration I am going forward with is the 13S16P with the cells connected as shown in the following images:

Perfect 16P battery Top.png

Perfect 16P battery Bottom.png

The connections between cells next to each other will be done with spot welding 7mm wide 0.15mm thick pure nickel strip. Thus, all cells will be connected with the nickel strips in pairs in the first place. Then, the 3.6cm length connections, connecting second with fourth or first with third will be done with spot welding a 10mm wide and 0.1mm thick copper strip that will be split right in the middle for the spot welder to do its job. Finally, the quadrants (rows in the images above) will be connected using a 12AWG wire and old fashion flux welding.

The final result should be a 48V 13S16P battery pack with each cell facing 2.073mOhms of resistance because of the above connections. You can see the circuit in practice using http://www.falstad.com/circuit/circuitjs.html with the text here: Perfect 13S16P pack.txt

I am looking for both feedback on my attempt and your own numbers to compare if possible.

Feeling excited now! Looking forward to actually finishing the pack and see the practical results.

And, in somewhat simple terms, it seems packs with parallel connections between cells of 2^n ...

So 2p, 4p, 8p, 16p and so on? It seems the 'perfect' connection is easiest to achieve when you just continually pair up cells, then pair up pairs, and so on?

[dukestar]

Hello, got a question on nickel strips...based on the ampacity chart a .15mm 8mm wide strip can handle ~5A...if I use a single wide strip of .15mm nickel along the entire length of p cells instead of 8mm strip to connect in series how do I calculate the max A the series connection can handle?

If 21700 cells would it be as simple as approx 21mm/8mm x 5A so around ~12.5A per p cell?

Thanks

[fechter]

If 21700 cells would it be as simple as approx 21mm/8mm x 5A so around ~12.5A per p cell?
Thanks

Yep, that would be my guess. You might need to derate it a little. Using a sheet instead of individual strips really speeds up construction too.

[pwd]

Hello pack experts. I'm attempting to design an odd shaped pack of 21700's in a 14S7P configuration and I could use some feedback. I need to build it to handle peaks of 140A. Here is what I have come up with so far
Black = parallel group
Red = series connection on side A
Blue = series connection on side B

Design 0.1d:

attempt 1

pack_layout.png (87.7 KiB) Viewed 2201 times

I feel good about the last four parallel groups since they are in a simple straight line but what can I do about the odd shaped groups to the left? I'm not sure how to attach the main positive and negative for optimal current sharing either.

Blank Slate:

blank slate

Screenshot_20190816_093904.png (79.79 KiB) Viewed 2201 times

Thanks

[knox1138]

Can we sticky this thread? Cause i've been trying to design a battery build and had a bunch of questions i couldn't find answers for til i went through this thread. Seriously, this thread should be required reading.

[kdog]

Pwd, that layout has many problems
Complexity, very poor current sharing, insufficient area for series connects.
Something like this would be better :

IMG_20190823_33441.jpg (87.93 KiB) Viewed 2091 times

Then add as many series connects as possible, one for each cell is ideal.. In you previous version several groups had only 3 series connections ~47amps each! Yikes! Thats going to seriously hurt the2 joining P groups
This was just a quick doodle, there is probly improvements that can be made to that layout.
To pull the power out, Iike to take several smaller gauge wires soldered to tabs which are spotwelded on. Ideally one for each cell but as many as possible. Bring them all together to join the main lead out.

[kdog]

Power leads like so:

IMG_20190823_40991.jpg (95 KiB) Viewed 2087 times