Low voltage cutoff for small lipo - where do you set yours and why?

Low side was the easier path for me when I built a similar cutoff for a trail camera pack. The detector chip wants its ground on battery negative anyway, so putting the load FET between the load's ground return and the negative terminal meant it could drive things directly with no extra parts. The catch is your load no longer shares a common ground with the battery, wich bit me later when I tried to add a charge port. High side keeps ground common but then you need that second FET doing the gate driving, which sounds like what you've allready got, so I'd say stick with your arrangement and just be careful where the grounds land on the detector side of the divider. On the trip point I lean gentler too. Mine cuts somewhere just above 3 volts under load, and the cell bounces back to a comfortable resting spot after cutoff. On a light load the runtime you sacrifice is barely noticible, whereas sagging deep on a small cell is a one-way street for capacity.

Hysteresis is not overkill, it's required. The instant you disconnect, the sag vanishes and the voltage climbs right back over your threshold, so the load FET re-engages and you get buzzing until the cell settles. Mine chattered for several seconds before I added feedback from the output side. And before you pot it, leave yourself a pair of test pads, epoxy is forever.

I ended up going high side on mine and I'd push back a bit on the low side route. I built a cutoff for a bait boat battery a couple summers back, and the thing that sold me on switching the positive rail was that the load keeps its ground no matter what. When the board cuts off, everything on the load side still sits at the same reference, so if you ever probe it or hang a second gadget off the same pack, you're not chasing wierd readings through a floating ground. I used a P-channel FET up top with a little NPN pulling its gate down, driven off the same detector idea you described. A few more parts, sure, but the behaviour was cleaner. On the trip point, for what it counts, at around an amp on a 1S pack the sag is small, so a cutoff up near 3 volts barely costs you runtime. If the pack mostly idles between uses like you say, I'd rather leave the margin than squeeze out the last few percent. Hysteresis turned out to be necessary, not a luxury. Without it mine sat right at the threshold and toggled the load on and off, which is worse for the cell than a slightly low cutoff. I fed a bit of the output back into the divider through a resistor so once it trips, it stays off until the cell recovers a bit above the trip point. Just enough gap that it can't chatter.

One caution on the epoxy plan: pot it after you've confirmed the trip and recovery behaviour, because once it's in resin, nothing is getting adjusted.

Watch that divider current draw once it's potted. I built one of these back in the late 90s with a plain zener and a comparator, and the reference divider sapped the pack flat over a winter of storage. If the thing sits idle between uses like yours does, that bleed matters more than the trip point. Mine draws so little now you'd need a meter to find it. Also on the potting, I used to wax my boards first, made the epoxy easier to chip off if I ever had to get back in there. Learned that trick rewinding motor armatures, of all places.

This is helpful, I was picturing the cutoff thing sitting between the battery and the light like a switch in the middle of a wire, didn't realize there were two spots it could go. The epoxy part scares me a bit now, I was going to cover mine completly since it looked messy. What do these test pads look like exactly? Are they just little bare spots of metal you can touch a meter probe to after it's all sealed up?