Eye safety of TSAL 6100 Infra Red Emiting Diode

I'm thinking of creating my own laser tag system and am a bit concerned about the eye safety aspect. Most of the homebrew system I've seen use the TSAL 6100 diode(TSAL 6100 data sheet).

From what I've gathered from hobbyist forums it should be totally eye safe. However the math doesn't add up, my math could be wrong, read probably is, but I couldn't find any conclusive data that would disprove it. Also I cannot manage to factor in what effects lenses would have, which are a must to keep the beam from having to large a spread. Here are my calculations.

First off using the inverse suqare law and a TSAL running at 1Amps (from what I've read most players run it at 1.5 Amps) at the minimum distance of 200mm we get the following result:

Ee = 1.5W/sr / (0.2)^2m = 37.5W/m^2 (the 1.5 W/sr is from the TSALs datasheet at 1A)

The irriadiance of 37.5 is a bit high, but it's still under the 100W/m^2 safety limit for the cornea.

The same equation done at a distance of 100mm however yields an Ee of 150 W/m^2 which is above safety levels. Assuming you don't point a tagger in your face and start firing of shoots you'll be safe.

What I'm curious about is how a lens affects the safety characteristics. Assuming a lens with a focal length of 100mm. Won't the irradiance be the same with the irradiance at a distance of 100mm from the IRLED.

My math and undesrtanding of the principals could be wrong, but I was curious if anyone could put my mind at ease.

I've also made a rough calculation of the retinal hazzard. R(lamb) = 0.33 rad (angular subtense) alpha= 5mm/200mm = 0.025 rad

We'll use an alphaeffective = 0.011rad and that gives us an emission limit for the radiance of 545mw/mm^2/sr

and our iriadiance is Lir = 1500mw/sr * 0.33 / (5mm)^2 = 99mw/mm2/sr

that's way below the limit, but if we modify that active light source to 1mm it becomes 495, which is skirting the line.

(the tec doc doesn't define an active light source size, at least I couldn't find it so I used 5mm )

Again the above calculations are for a source without a concentrating lens. What would happen if we added a lens in the mix? Would it stay in the safe zone? How does pulsing the diode effect intensity? I'm guessing it lowers it, but if anyone could manage it, I'd like a numerical demonstration that it brings it closer to safer bounds.

The above calculations are based on the folowing OSRAM paper (OSRAM Eye safety calculations), which explains the principles involved in the calculations in a bit more detail.