#21700

You only find out how many pitfalls there are once you build your own FPV drone packs: the negative side of tabless 21700s is really current-hungry — I had to push the UF20B to gear 52 to get full penetration. Paired it with a 40A BMS and measured every cell's internal resistance before welding; max spread 2mΩ. Flew three packs, and the shell wasn't hot when I touched it after landing. Happy. The recipe is attached to the project. Judges, please go easy on me.

The wires in v1 were too thin and got hot, so this time I went with 2S 21700 plus a 30A BMS, and double-layer nickel strip. This machine is seriously small — it sits in a corner of the bench without taking up space, recharges in a few seconds, and I welded 40 spots in a row without ever having to wait. Best part when it was done: not a single spot needed re-welding.

Capacity grading, matching, welding, fitting the BMS, heat-shrink wrapping — five steps. For the welding step I used the UK1, 0.15 nickel-plated at gear 60, just right for full penetration. The one thing I most want to say: matching matters ten times more than welding. If cells from the same batch differ by more than 5mΩ in internal resistance, no amount of good welding afterwards will save it. The most-asked question in the comments, "how much per pack?", is answered too — in step 4 of the project.

Four 4S packs, under 60 apiece. Welding was the easiest step; getting the balance lead wiring order right was what really strained my eyes. My palms were soaked with sweat on the first flight. Crashed once and broke a prop, but the pack came through fine. To anyone wanting to get into it: learn spot welding first, and everything after that is just saving money.