Pulsing can make use of eye response and persistence to make LED appear brighter. There is a lot of factors to consider with regard to initial brightness, color, and angle of entry at the eye.
**broken link removed**
I believe that is correct, pulsing *can* make use of eye response and persistence.
The operative word in there however is, "can". The problem is, most circuits that are used for lighting can not take advantage of this because the period between pulses would have to be too long for a human being to be comfortable with. In other words, nobody wants to use a flashlight like they did when flashlights historically first came about and batteries would run down quick, where they had to literally "flash" the light on (then the battery voltage would die down and the light would dim out) then turn it off and wait a few seconds, then turn it back on to get another flash.
Most people these days want a light that at least appears to be of constant intensity like modern flashlights or home lighting. They dont care if it is pulsing, as long as those pulses are close enough together to make it appear to the eye that the light is of constant intensity. The problem with this is that when the LED is pulsed fast enough to appear constant, the eye actually responds as if it were really constant and so it integrates as usual. This of course means that the LED curve is the main limiting factor yet again, even though the LED is flashing.
There is one exception to this rule that i can think of, and that is for someone lost in a cave and their flashlight battery is running down. By pulsing the LED once every five seconds they can probably see where they are going and conserve battery power and actually get some benefit from the light that might appear to be brighter than usual and thus get more overall light from the same battery power which would seem like an increase in efficiency. As noted however the pulses have to be quite far apart in time, on the order of several seconds.
This isnt to say that there are some applications that can't benefit from fast pulsing (dimmers for example) but the benefits will never include higher efficiency as the overall efficiency will either be exactly the same for some apps or much lower for other apps. Thus, pulsing never increases efficiency.
This isnt too hard to prove looking at the manufacturers data sheet for say the Nichia 20ma white LED. The light output vs input current goes DOWN as the current goes up, meaning that for example double the current produces LESS THAN double the light output.
This isnt too hard to prove experimentally either...
Experiment To Prove It:
Take two LEDs, check their light output with a constant current source to make sure they are about the same, pulse one at 15ma average at 25 percent duty cycle at 1kHz and drive the other at a constant 15ma. Since the 15ma 25 percent duty cycle will require a 60ma pulse at 25 percent duty cycle, we will be driving the pulsed LED at four times the current level for one quarter of the time period. Now both LEDs have the same average current flow, yet one is pulsed and one is not. Point both LEDs at a white wall (the target) keeping them the same distance from the wall and note that the pulsed LED light output is quite a bit lower than the constant current driven LED. It's not as low as half as bright, but will look around 25 to 35 percent lower than the other LED, which is equivalent to the constant current driven LED being farther away from the target. This also means that we can illuminate the same target to the same level of brightness from a farther distance with the constant current driven LED. The other thing to note is that there will be a color shift, toward the blue or purple, at the higher current level thus the pulsed LED color will change somewhat also.
Lastly, to prove that both LEDs are really the same under pulsed conditions, swap them and note the same results.
Finally, if we look at the voltage vs current for the LED we find that the voltage goes up as the current goes up, and since we are driving the LED with pulses of 60ma that means the voltage goes up. It doesnt go up by that much, but it does in fact go up, which means we loose efficiency. The loss isnt that much, on the order of 10 percent, but it's still a loss. That's for the LED itself though. If we dont use a switching circuit the loss will stay the same, but we have to remember that the pulsed LED will look around 30 percent dimmer which still means a loss in efficiency. If we want to keep the light output the same we'd have to pulse at 85ma and then we would be exceeding the manufacturers recommendations for pulsing vs current as well as consuming a lot more power.