I don't know where the law of diminishing returns kicks in for your LEDs,
I'd suggest you look into led forward voltage vs. current. Knowing that
only Vj goes into light the rest becomes heat in Ri and Rbulk. Thus you
can set yourself a goal, f.ex. 80% electrical efficiency (meaning 20% of
the input power turns into heat on Ri and Rbulk etc), this will give you a
peak pulse current you can use for the efficiency you are looking for.
There should be some reference on Ri or Rs or something and a Vj/Ij curve
in the data sheet. You can estimate Ri from that curve (take Vj to be Vf
at nominal current and use the slope of the curve from there towards
higher current to find Rs - this is wrong from the academic pov. and close
enough from the real life one. grin.).

Otherwise IR LED's rated 25mA continuous are often pulsed at 1Amp and more
in remote control transmitters and they do not seem to mind (there are
billions of those around). That would be 40x overrating (but the duty
cycle is very low over time). The only thing I have noticed about them is,
that the *may* have a larger cathode electrode (better heatsink ?)
molded in the plastic.

So maybe they would be brighter (or you could force them more) if they
would blink, also with a low duty cycle, like 1:10, 3Hz, in addition to
the high frequency low duty cycle drive. I do not know what this would do
thermal stress-wise. In any case I would plan for some sort of safe
fail-over scheme, it would be a pity if a single led failure would take
out the whole device.

Peter

--
http://www.piclist.com hint: To leave the PICList
mailto:piclist-unsubscribe-request@mitvma.mit.edu