[onephatz]

I've been working on a 1s30p battery holder that can be made using less expensive methods. the prototype is 3d printed and the cells are harvested from laptop batteries. the intent is to weld nickel tabs from the cells to the bus bar then cover the tabs with a piece of mica it insulate it.

If I use a ncr18650pf cell the specs are as follows
Vmin: 3.0v
Vnom: 3.6v
Vmax: 4.2V
Amax: 300A
Capacity: 291wh

the goal was to build a pack that could run a chevy volt motor at the same voltage levels as stock. I compared it to the calb cam72 and it has about the same energy density but twice the power density. and if you run up to a peak of 400V you only have to run 96 cells instead of 112 for lithium iron cells.

Looking for feedback on the design, capacity, layout, ect.
I've been considering producing these for sale but am not willing to invest in the molds unless some quantifiable demand. Let me know what you think

jesse

 

[Solarsail]

I think it is good that the rows of cells are staggered so the basic pattern is an equilateral triangle rather than a square. This would allow for a higher density of cells.

You need to have a fuse wire or very narrow tab (1mm?) from each cell to the bus. I believe one per cell is sufficient. The other cell terminal can be tabbed in parallel which is easier than tabbing to the bus. So unless you can run an 8mm tab in parallel let's say on the V- side between the cells, then this design requires double number of tabs to bus, which is somewhat more labour intensive.

The 1s30p module is too small, I think. You would need 96 of these to make a 96s30p pack. I am building the same idea to extend the range of a Leaf by 12 kWh, and the layout is 14s10p banks with two in series in the same enclosure to make a module, and then 3.5 modules in series - namely the 4th module will have a center tab and only one of the two banks will be used, with the other bank unused in the Leaf: 3.5s2s14s10p for an equivalent of 98s10p. The batteries are the NCR18650B, 3.4 Ah each. Placing and connecting 96 modules will not be easy, as compared to 4 modules (12kWh) or 7 modules (24kWh).

Each module will contain a charger, two 14s balancers and protection circuits, two cell group DVM so each of the 14 groups in a bank can be manually checked, and a module digital ammeter.

One nice feature of this design is that each module is nominal 50V (when the two banks inside a module are connected in parallel - 2p14s10p) which matches the 48V standard for propulsion motors. So the modules can easily double as 48V (or 96V or 144V) packs for non-Leaf applications.

Here is a thread and picture for it:
http://www.diyelectriccar.com/forums/showpost.php?p=909826&postcount=22

 

[brian_]

How would the cells in this holder be cooled? It appears that they are fully boxed in, so there can be no airflow, and of course there is no provision for liquid cooling.

How about mounting the holders? There are no flanges or tabs or threaded sockets to connect them to a structure.

I also wonder about the bus bars: only the edge is exposed for making connections to each cell, and even then it is flush with the housing, which might be awkward for tooling. It might be easier to understand the intent if at least a few adjacent cells were electrically connected in the prototype.

 

[brian_]

I think it is good that the rows of cells are staggered so the basic pattern is an equilateral triangle rather than a square. This would allow for a higher density of cells.

I agree. Properly arranged (which is called hexagonal packing) is certainly better than the classic "round pegs in square holes" approach.

 

[Solarsail]

(edit: I was not sure if brian's question were to me the original thread - but now I believe he was addressing onephatz)

My application is 25kW or 3.5kW per bank of 14s10p, so that will be 3.5kW/(14s x 3.6V) = 68A per group of 10p, or 6.8A (2C) per cell. At 50mO internal resistance, that would translate into 2.3W thermal generation per cell at high discharge power. I believe the cell will not have an issue dissipating at this rate - and it is momentary. Charging is at 0.5C which is only 0.6W thermal generation. Continuous discharge will also be about 0.5C.

There is 2mm separation between cells. The enclosure will be vented. At continuous charge/discharge of 0.5C, the module will generate 2x140x0.6 = 170W. I believe the vents to be sufficient and fans would not be required.

 

[Solarsail]

(edit: I was not sure if brian's question were to me the original thread - but now I believe he was addressing onephatz)

For mounting - as the pattern is a square mount, there is ample room in the space between 4 cells to pass an M5 threaded dowel vertically, parallel to the cell. The plastic spacers appear to be rigid enough that for a 14x10 pattern of cells, about 8 or 10 dowels are sufficient. The dowels begin at the bottom plate or enclosure, go through the first bank, then through the second bank, and terminate at the top plate. The walls of the enclosure will be about 5mm away from the plastic grid on all sides including top and bottom to allow for ventilation. Mica sheets will be used where necessary.

If you are referring to my prototype, the plastic spacers (ordered from eBay), allow for an 8mm nickel strip to run from one cell to the next. Therefore, in a group of 10p cells, one strip of 18cm can be tack welded to the V- of the ten cells. Then on that same face, a 1mm fuse strip will be welded between the V+ of each cell in the next group of cells to the above V- strip. There will be no need for a copper bus bar.

As there should be relatively little current flowing through the V- bus strip because the cells are balanced, I do not believe a copper bus bar is needed, except for the first and the last bus strip in the 14s10p grid.

 

[onephatz]

You need to have a fuse wire or very narrow tab (1mm?) from each cell to the bus. I believe one per cell is sufficient. The other cell terminal can be tabbed in parallel which is easier than tabbing to the bus. So unless you can run an 8mm tab in parallel let's say on the V- side between the cells, then this design requires double number of tabs to bus, which is somewhat more labour intensive.

The 1s30p module is too small, I think. You would need 96 of these to make a 96s30p pack. I am building the same idea to extend the range of a Leaf by 12 kWh, and the layout is 14s10p banks with two in series in the same enclosure to make a module, and then 3.5 modules in series - namely the 4th module will have a center tab and only one of the two banks will be used, with the other bank unused in the Leaf: 3.5s2s14s10p for an equivalent of 98s10p. The batteries are the NCR18650B, 3.4 Ah each. Placing and connecting 96 modules will not be easy, as compared to 4 modules (12kWh) or 7 modules (24kWh).

I wanted to build a holder that had enough current capacity to run a chevy volt motor at 400V. I considered trying to build one that was 6s and I still may but at this size a person can 3d print it at home. I also wanted to make a holder that a person could build a pack of any shape and size to fit their own application. I would agree that building in a cell level fuse is a good idea but don't know what cross section to get a reliable 15A fuse.

How would the cells in this holder be cooled? It appears that they are fully boxed in, so there can be no airflow, and of course there is no provision for liquid cooling.
How about mounting the holders? There are no flanges or tabs or threaded sockets to connect them to a structure.
I also wonder about the bus bars: only the edge is exposed for making connections to each cell, and even then it is flush with the housing, which might be awkward for tooling. It might be easier to understand the intent if at least a few adjacent cells were electrically connected in the prototype.

No, the pack is not actively cooled. I've seen a discharge vs temperature rise plot of the ncr18650pf cell and if you anything less than 10A you will not exceed the 65c limit of the cell. I will agree that liquid cooling is much preferred and I'll be working on that as soon as I get this lined out a bit more. I plan on open sourcing this design and part of the change can be to eliminate the side walls so force air cooling would be possible. your point on retention is well made, on a previous iteration (currently on rev5) I integrated a cut out that would allow it to be clamped to a structure.

I'll get the tabs welded on so you can better see what it's supposed to look like. a friend has a powerful tab welder that I'm going to use to weld the nickel strip to the copper bus bar. if I can't make that work I'm going to have to nickel plate the copper bar.

jesse

thanks for the feed back so far.

 

[brian_]

(edit: I was not sure if brian's question were to me the original thread - but now I believe he was addressing onephatz)

Sorry I wasn't clear: my questions (in post #3) were about the onephatz's design, from the original post of the thread.

 

[onephatz]

You need to have a fuse wire or very narrow tab (1mm?) from each cell to the bus. I believe one per cell is sufficient. The other cell terminal can be tabbed in parallel which is easier than tabbing to the bus. So unless you can run an 8mm tab in parallel let's say on the V- side between the cells, then this design requires double number of tabs to bus, which is somewhat more labour intensive.

The 1s30p module is too small, I think. You would need 96 of these to make a 96s30p pack. I am building the same idea to extend the range of a Leaf by 12 kWh, and the layout is 14s10p banks with two in series in the same enclosure to make a module, and then 3.5 modules in series - namely the 4th module will have a center tab and only one of the two banks will be used, with the other bank unused in the Leaf: 3.5s2s14s10p for an equivalent of 98s10p. The batteries are the NCR18650B, 3.4 Ah each. Placing and connecting 96 modules will not be easy, as compared to 4 modules (12kWh) or 7 modules (24kWh).

I think you are right about the cell level fuses. I've got to figure out what cross sectional area I need to fuse at around 15A. As far as not big enough, I considered building a 6s module but I wanted to build something that could be configured to fit a variety of applications and could be printed on a home 3d printer.

How would the cells in this holder be cooled? It appears that they are fully boxed in, so there can be no airflow, and of course there is no provision for liquid cooling.
How about mounting the holders? There are no flanges or tabs or threaded sockets to connect them to a structure.
I also wonder about the bus bars: only the edge is exposed for making connections to each cell, and even then it is flush with the housing, which might be awkward for tooling. It might be easier to understand the intent if at least a few adjacent cells were electrically connected in the prototype.

Yea, no active cooling here. I have seen a 10A time vs temperature curve for the ncr18650pf cell and really anything under 10A is going to be fine. My ideal is liquid cooling but I don't have a good way yet to get the channel inbetween the cells cheaply. I agree that there needs to be a attachment point and previous iterations did (this is rev5) and I expect that over the next few iterations I'll put them back on. What would you rather see, a flange or a wedge? I'm leaning towards wedge as it's easier to make/print. I'll get some tabs attached and take another picture to show how it's supposed to be attached.

jesse