I started a thread similar to this one over in SEC and didn't really get anywhere. Let's try it here and see how it goes.
We have a 5 volt source from which we want to drive an LED. For the sake of discussion, let's presume that with 20 mA into the LED that it drops 2.0 volts. Calculating the resistor isn't difficult -- 150 ohms ought to work. Calculating the power consumed by the diode isn't any more difficult -- 40 mW.
Now let's put a one-shot in between the source and the LED so that the duty cycle is 25% at some reasonable frequency above the eye's flicker rate. Again for discussion, we'll presume 100 Hz.. Assuming that the LED can take the current, we can pump 80 mA into the LED with this duty cycle, at which point the LED's voltage rises to 2.5 volts (actual data from the HP data book). Resistor? 33 ohms ought to work. Power? 200 mW peak, 50 mW average.
With my limited biotechnical knowledge of the human eye, it would seem as though the second LED should be brighter. Yet in (admittedly few) bench observations the second LED was a bit dimmer, if anything. The LED wasn't getting overly warm, nor did the voltage and current change as time went on, indicating some moderate thermal stability.
The HP apps note on the subject (AN-1005) goes into a long song and dance as to HOW to achieve pulse driving an LED, but doesn't come right out and say that it will be brighter if you do this. But why would they take the time to show how to do it if it didn't give any benefit?
I'm not understanding something.
Jim