Been a while since my last mod…
What to do if I want build something, but I am currently in the situation that I have bought and built everything I desired years ago already, and nothing much exciting has been happening in led innovation for years? It must be this: building a light of a emitting colour that you can hardly see and has no purpose (700nm), with the light produced indirectly via fluorescence which is more complicated and more inefficient compared to just using a 700nm led.
But there is a ruby in it 
I recently bought a couple of 10mm ball-shaped synthetic rubys on aliexpress, no real reason, just because I could. Sounds exotic but they are only a couple of dollars nowadays. They have the correct specific mass for corund, and the fluorescence from uv-light is around 700nm, so they indeed seem legit rubys to me.
Because of the fluorescence I thought of using one in a flashlight, and looking into my stash of hosts, the BLF348 seemed the easiest fit:
If I remove the reflector and lens, and replace that with the ruby ball, it should about fit. Because the pill sits on a short spring (that electrically contacts the driver with the body, like a P60) I have a bit of extra length to use. Only extra mod: the ball rests directly on the led dome if i do nothing, which is not elegant and could cause failure, so the led dome needs a collar for the ball to push against. Looking for something suitable in my hobby mess I found a small Omten switch that was faulty anyway, and inside I found the bit (the bit under left in the pic below) that I was looking for, needing just cutting to height.
So what led to use? The absortion spectrum of ruby shows a nice absorption peek at 400nm, that should cause a nice fluorescence, and looking through my leds I found a 400nm led that I once bought from Kaidomain.
So I swapped the lovely 219B V1 SW45 R9080 (sorry 
) for the 400nm led on the tiny pcb of the BLF348.
The ruby ball doubles as the front lens, and with the o-ring between ball and bezel it is even sealed just as fine as the stock light.
There you go, and it works alright.
I checked the light with my spectrometer, and there are two partly overlapping peaks around 700nm of which left one (around 700nm) is in all the literature about ruby spectra and the right one (around 710nm) seems unusual. I will not be bothered though, the light does what it was intended to do.
Apart from big peaks around 700nm, there is a low background of visible light going over the complete spectrum that barely shows up in the measured spectrum but is clearly visible by eye through the red output (reason is that my eyes hardly see 700nm, while the lower wavelenghts in the visible spectrum show up a hundred times better), I’m sure that is the weak white-ish stray light that every UV-led also emits besides their main peak, which is not absorbed by the ruby and shines through. It is not un-absorbed 400nm light btw, that seems all gone.
My integrating sphere measures 67 lumen, which seems not correct, by eye I estimate it 10 lumen the most. It could be the luxmeter (Mobilux classA) , although it is the most accurate portable luxmeter that you can get, it may struggle to correct well for 700nm. Or else my personal eyes are less sensitive to 700nm than the eyes from the people whose brightness perception of all wavelengths was used to define the lux back in the thirties.
In the beam at 5cm from the wall you can see the led white-blue-ish projected (the ruby ball acts as a lens). At longer distance the beam is spread out and more uniform.
That was fun building this today, I bet the only flashlight in the world that uses a ruby as a lens. 
