Re: My curiosity got the better of me: ordered Photon Rex
Taking apart the Rex is easy. Just use a small screwdriver to undo the little clips on the bottom and the two halves come apart. Then slide off the GITD rubber cover, and you get at the guts of the light.
(the picture is clickable for a full-res version)
The LiIon cell is a tabbed LIR2450, rated at 140mAh (I didn't read the rating on the Rex's cell, but every other LIR2450 I can find with a Google search is rated at 140, so the Rex's should be too). Unsoldering one of the tabs from the circuit board allowed me to solder test wires and take measurements.
Draw on full power is 210mA or so from a freshly charged cell, but drops to 180 or so in a short while. I then set it to about the same level of a DX fauxton with fresh cells (judged very scientifically by shining both at a sheet of paper, with the Rex closer as it's more floody), and the draw was approximately 50mA. At minimum power (starting from off; for some reason, the minimum when lowered from full power is brighter) it only draws half a milliamp.
The Rex is a microprocessor-controlled light that is never truly completely off; the good news, however, is that the parasitic draw necessary to keep the microprocessor alive is only 4 microamps, with extremely brief spikes of 15 or so (my meter registers a spike but it's too short for it to accurately report it) every few seconds.
Charge current to the cell is about 90mA when the Rex is connected with the charging doodad to a freshly charged Hybrio NiMH. I imagine the current might change as both cells' charge levels change, but I'm not motivated enough to do a proper graph (and I don't have a computer-connectable meter, anyway).
We already know that full power battery life is about 18 minutes to 50%; this isn't what the numbers suggest, but it probably happens because the LiIon cell is so small. LiIon cells are usually good for a 2C discharge rating, but that's for ones of decent size. My theory is that smaller ones such as LIR2450s start to seriously sweat at 1C, and the ~1.4C discharge current of the Rex at full power beats them pretty seriously.
Lower modes shouldn't tax the cell so much, so we should expect fauxton-level brightness to last about two and a half hours. The lowest low should last more than ten
days. If kept off, a fully charged cell would be drained by the parasitic draw in something like three and a half years
if my calculations are correct.
Edit:
The LiIon cell might not actually have a 140mAh rating; I've searched more extensively, and it would appear more serious (and more expensive) enterprises rate them at 90-100mAh. This probably means the ones I saw before are overrated, as is typical with cheap manufacturers.
If that's the case then the full power load is approximately 2C, or maybe even more, which would make a lot more sense with the full-power runtimes we know. Accordingly, medium power should run for about two hours, and low for 8 days.
I might just unsolder the cell entirely and see if it has a rating; as it is it's hard to read what's printed on it because you need to tilt the board too much, and you risk breaking the wires that connect it to the charging terminals. (end edit)
There is some bad news too, though.
The Rex overcharges the cell. After a full charge on a Hybrio NiMH I let the cell rest, then measured cell voltage, and found it to be alarmingly high: 4.26V. While not enough to blow the cell, this is 0.06V over the safe margin, and would shorten its life considerably if allowed to happen at every charge cycle.
Even more alarming is that the charging voltage while the cell charges gets as high as 4.34V! This is
very bad for it!
Also, I noticed that when the voltage is so high the Rex starts flashing irregularly. It flashes twice or three times, then stays off for a few flashes, then again flashes a few times, then off, and so on. I'm not sure why it does this.
And yes, I did rule out a bad multimeter by measuring with three of them. They all agreed.
In conclusion: the interface is great, and it puts out an amazing amount of light for the size, but the charging system sucks. I cannot possibly allow the cell to get so drastically overcharged; I'll solder a JST connector to the tabs and I'll charge the cell with my outboard charger, but I'm very disappointed that I can't use the nifty magnet thing to charge it anywhere. Based on these findings, I obviously cannot recommend the Rex to anyone.
Edit: I'm contacting the people at LRI, maybe they can suggest some way of repairing the existing charging system.