I have a couple of different LED flashlights. One of them has three different "modes" of brightness, and the way it controls it is via pulse width modulation (PWM). Here is a picture that illustrates how it works:

enter image description here

I know that this particular flashlight's PWM circuit operates at about 120Hz. What this means is that when you move something very quickly under it, when running a low duty cycle, it produces a strobe-light effect, where you'll see many "copies" of the moving object. It reminds me of the way video-games are rendered, and it actually creates a really neat effect because you only "see" very small time-slices. Watching running water this way is absolutely fascinating.

From empirical observation I determine that on the lowest setting it seems to be on about a 2-5% duty cycle. The ghost-images produced are remarkably sharp. On Medium, I reckon it's about 25 to 30%.

Something I've noticed very recently is that when I use this flashlight under dark-adjusted conditions, the brightness that is perceived definitely doesn't seem to scale linearly. I have no scientific light intensity equipment to perform measurements with, but I am more or less convinced that a 1000 lumen light on a 2% square-wave duty cycle appears brighter than a 20 lumen light, all else being equal (which would include incident light energy).

I think there may be some biological explanation for this. Is this an effect that has been observed by others?

  • $\begingroup$ It is very common knowledge, that most (all?) of our senses work with a logarithmic scale. For sound and light this is proven since Helmholtz times. $\endgroup$ – Georg Oct 5 '11 at 10:35
  • $\begingroup$ Yes I am aware of that. The variable I'm concerned with isnt the logarithmic response, it's the relationship between high (well, 120hz isn't that high) frequency modulations which contain the same average energy. $\endgroup$ – Steven Lu Oct 5 '11 at 10:38
  • $\begingroup$ Next thing is the integrating action of the perception. This is maybe different for dark adapted (b/w) and color vision. $\endgroup$ – Georg Oct 5 '11 at 11:02
  • $\begingroup$ How precisely do you know the 1000 lumen light really has 1000 lumen, the 20 lumen one 20 lumen, and the duty cycle is 2%? Under which circumstances did you compare them? What colors are the lights? $\endgroup$ – leftaroundabout Oct 5 '11 at 11:21
  • $\begingroup$ I think a lot of my questions could be completely answered if I just had the proper equipment for performing some scientific testing. An oscilloscope would allow me to figure out the duty cycle, and light power meters too for finding out intensity. $\endgroup$ – Steven Lu Oct 6 '11 at 14:37

Your eyes are very bad at estimating absolute brightness, especially a dark adapted eye looking at a bright light.

One way to measure the 'average' brightness is to try and read the same printed text under the equivalent illumination eg. 20lumens * 100% and 1000lumens * 2%

Remember though that LED flashlight makers are notorious liars when it comes to lumen figures.

  • $\begingroup$ That is definitely a good idea because it would take into account actual cognitive performance based on vision rather than relying purely on how it "feels". $\endgroup$ – Steven Lu Oct 6 '11 at 14:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.