As my first LED upgrade, I modded a Magcharger for a friend to take an SSC-P7.
The clearances inside the head are tiny, and having seen how much heat gets produced from shorted NiCds, I wanted some bombproof shortcircuit protection.
The user is totally non-technical, so I realised a standard fuse could be a problem, as well as being hard to fit.
I then realised the Polyfuse was the ideal solution - small, little voltage drop in normal operation and self-resetting once the fault is cleared.
There was no point putting it in the head, because there was a chance of a short before the fuse. Then I realised that with a bit of careful work, I could fit it in the tail, where it would protect against a shortcircuit anywhere.
I used a piece of very thin Double-Sided Printed Circuit Board to interrupt the connection between the battery and the spring. I ground away the copper near the edges to make sure contact with the case wouldn't short out the Polyfuse.
To keep the side against the battery flat, I ground away through the copper and fibreglass so I could solder the Polyfuse to the inside surface of the copper that's in contact with the battery.
Since the torch would work perfectly without it, I put a clear explanation in place for the owner.
If there is a short circuit, the Polyfuse will stay hot until the short is removed, so you have to make sure there are no heat-affected parts near it - you don't want it melting through wire insulation.
I was just wondering about these resettable fuses and incorporating them into LED flashlights. My understanding is that when tripped, they do still let some current pass through. I take it you haven't encountered a short circuit and don't know from first hand experience if this is enough to still get some light output, but then again I haven't gotten a chance to really look at the specs of these things too closely yet, so I have no idea what the leakage current is yet.
I was actually wondering if this has any potential for use as part of a cheapo driver for maybe a small backup light with with a 1W LED or something that you wouldn't necessarily care about and wouldn't be too mindful of what voltage you threw at it. These are supposed to expand when overheated/tripped and take awhile to reduce in size and resistance. Maybe have them next to a heatsink so when they expand, they hit the heatsink, cooling them enough to possibly maintain a certain level of resistance. Actually, the nature of the tripped state probably takes long enough to cool down than having it directly attached to a heatsink might work, selecting the size of the heatsink for a generally inexact range of resistance.
In no way practical, I know, but just a thought. Since I have none of these laying around, I can't really speak for how the may or may not behave.
I've tested the Voltage-Current characteristic of these Polyfuses. They have such a strong positive temperature coefficient of resistance that you can't use them as a constant-current source. As soon as they start heating up, the voltage and therefore the power dissipated in them goes up so quickly that the current drops to a low level.
There is no physical change in their size when they heat up. They're covered in a hard epoxy coating which would crack if there was any expansion.
You could use them as a thermal cutout if you placed in contact with the heatsink. It doesn't matter if the heating is internally or from an external heat input - the resistance will go up. So if the heatsink can't get rid of the heat, the Polyfuse will get hot and increase in resistance, thereby increasing the internally generated heat and so reducing current.
Well, at least it should make a good fuse. No one wants to replace a blown fuse in a flashlight when they can't see what they're doing.
Yours may just seem hard because it's in the un-tripped state and the epoxy coating might be able to expand under heat, or maybe I've just been looking at a different type of resettable fuse, which is filled with a crystalline polymer embedded with carbon to make it conductive. When it heats up, the polymer expands, becoming amorphous, and creates gaps and therefore resistance in the carbon pathways. When it cools, it recrystallizes. I don't think it really expands much at all, but it's enough that if you're going to put it in an epoxy cast, they recommend putting a small box over it so it doesn't get embedded in epoxy.
...Which makes me wonder if casting it in epoxy to fix its size might regulate the current it passes when tripped. It would probably still heat up and the polymer would become amorphous, but without being able to expand, it wouldn't be able to fully transition into its tripped state. It's possible that it would even pass erratic amounts of current, I don't know. That might actually be worth an experiment though.