Stuff used

DeWalt DC9360 36V 2.4Ah battery pack - around £90 on ebay.co.uk. These contain 10 A123 26650 M1 Lithium cells.

From rswww.com:

Antex TCS230 50 W temperature controlled iron - www.antex.co.uk - look in standard soldering irons

101mm heavy-duty heatshrink (1151-018) - technobots.co.uk - look in Cable & Connectors / Heatshrink

Summary

I will describe below how to remove the cells from the case, remove the welded steel tabs and solder the cells together. The reason we remove the case and BMS is to save weight. The case and BMS weigh almost 400 g, adding around 50% to the weight of the cells, and after all, the main reason we are using these cells is for the weight saving.
The reason I remove the welded steel tabs is to reduce the electrical resistance of the pack as much as possible (copper is around 10 more conductive than steel), reducing the temperature rise in the pack when drawing high currents. This is probably only necessary in high current applications, where you are regularly drawing more than 50 amps from a pack. For less demanding applications, you can just leave the cells connected with the welded steel tabs.

Build

You should take care when soldering to these cells, as they can be damaged by excess heat. On NiCads we usually just use a simple large iron (75 Watts or more), but these generally are not temperature controlled and so get very hot - you can often see them glowing red hot...
So I decided to make my own, based on the Antex 50W temperature controlled iron. I made a large copper bit for it out of some 6mm copper plate that I had. Set the temperature control to 350 to 400ºC, to ensure you can melt the solder nice and quick. Works well.
Weller also do a nice looking 200 Watt temperature controlled iron.
Soldering iron


I decided to go for a fully soldered pack, for maximum current handling, so the first job was to remove all the tabs from the DeWalt cells.
NOTE Be careful when working on these packs. It is very easy to short circuit them which can easily blow a hole in the cell case. Always tape the ends of the cells that you are not working on.
Also, A123 say that although it is possible to rotate the button (negative end) on the cell, this should be avoided. Here's why:

copper connector to button

The copper strap from the cell contents is riveted to the underside of the button. This should survive very small rotations, but if you overdo it, you risk damaging the strap.

The tab on the positive (flat) end of the cell is spot-welded onto a small steel disc. This disc is soldered onto the cell.
Contrary to some stories, there is no vent in this end of the cell. The base of the aluminium can is solid with no holes.
The tab on the negative end of the cell (button with blue dot) is spot welded directly onto the steel button. The blue dot is where the cell is filled during manufacture.
I start by de-soldering the positive ends. Slide a scalpel under the tab and apply the iron to the positive end of the cell. You may need to feed a small amount of solder between the tab and the disc to improve the heat flow into the steel disc. After a short while the solder between the steel disc and cell should melt and you can lift it off with the scalpel. Although the case of the cell is aluminium, the steel disc is attached with normal solder. The end-cap must be plated with another metal to enable easy soldering. So no need for special aluminium solder.

tabs


You can then break the tab off the negative end of the cell with a pair of fine-nosed pliers. The steel button is quite robust, though as mentioned above you should take care not to rotate it.
I have taped the cells together with glass fibre reinforced packing tape. You could also use a tough flexible adhesive to fix the cells together e.g. a polyurethane adhesive like B&Q Rubber Glue or Bison Liquid Sole

bare cells


I connected the cells with doubled-up 14 SWG solid copper wire. An alternative would be to use copper braid for a more flexible connection.
A good video of the sort of time it should take to solder each cell. If it is taking much longer than this, then you need to use a bigger/hotter iron.
You are probably still going to lose some capacity by soldering to the cell, due to heat damage of the internal components, but it should only be around 0.1 Ah.
You should get some spare cells so that the odd cell in a pack can be replaced if it is going flat significantly before the others.
cell connections


Then connect up the leads for the balancing plug. I decided to provide connections to all cells (11 wires), so that a standalone balancer could be connected with just one plug.
Use plenty of Kapton tape to separate the wires from other cells, and avoid the wires crossing over each other, so that even if the cells get very hot and the insulation on the wires starts to melt, the Kapton should stop bad things happening.
The balancing wires inside the pack should all be the same length, so that the voltage drop along the wires when balancing is the same for each cell. If they are different of lengths, it makes trying to balance the cells more difficult.

balancing wires



Finished pack, ready for heatshrink. I wrapped Kapton around the high-current leads too.

pack


heatshrinked


Finished pack weight: 875 g.