Effect of changing PWM frequency of a light source on perceived brightness

Question

First and foremost, I'm not entirely sure if this is the proper place to post this question, so I apologize if there's a better section for this.

I am looking to determine the effect of changing the PWM frequency of a light source, LED for example, has on the perceived brightness to the human eye.

As an example, say you have an LED operating at 80% duty cycle at a frequency of 200 Hz. Keeping the duty cycle at 80%, but increasing the frequency to 400 Hz, does the human eye perceive these two situations as the same brightness? What about a larger increase, to say 1 kHz or 10 kHz?

I have have tested this using an ATmega328p and two LEDs. I had one LED at 80% duty cycle at 122 Hz. The other LED starts at 80% duty cycle at 30 Hz (yes there is a very noticeable flicker), then with the press of a button, it changes the frequency. The values used are 30, 122, 245, 490, 980, 3920, and 31370 Hz.

While it wasn't exactly what I was after, I took light measurements using a BTS256-LED from Gigahertz Optik. Aside from 30 Hz, which I attributed to the flickering, it appeared changing the frequency had no effect on the amount of lumens that the LED put out.

Then I tested in both a light and dark environment, and I was not able to discern any difference while changing the frequency.

I have done a lot of searching and reading, and I've learned a ton about PWM, LEDs, flicker, heat dissipation, sensitivity to light color/brightness, and much more. Unfortunately none of it is what I'm looking for.

Looking for any kind of insight here. Perhaps this is a bad question, and I'm barking up a tree that doesn't exist. Or perhaps the answer is either very obvious, indicating my findings from my test were accurate, or far more complex than I realize.

I appreciate any and all responses, as I'm out of ideas/topics to look for more information.

Answer 1

I think your observations are correct. Once the frequency is above the lower limit where flickering is observed the perceived brightness should be independent of frequency at the same duty factor.

At very high frequencies, if the time required to actually turn the LED on or off (probably nanoseconds) becomes comparable to the pulse duration, then I could believe that the perceived brightness might change.

Answer 2

Is there a difference in human eye perception of a LED operating at 80% duty cycle at a frequency of 200 Hz, versus 80% duty cycle at a frequency to 400 Hz?

Short answer, NO.

But the answer only applies to a single LED or a group of LEDs fed by the SAME PWM signal.

If you have multiple LEDs being fed by a similar but not identical PWM signal (ie Two Arduinos programmed to use say 200Hz on separate LEDs or groups of LEDs) then the difference between the frequencies becomes important.
The difference between the PWM frame rates can easily be within the human perception range. This is very easy to duplicate with a couple of Arduinos and get strings of LEDs that seem to slowly brighten and dim even though the PWM frame rate is not changing on either controller.

You might find this a helpful read.

To reduce the potential for beating between LEDs, most of the industrial implementations put considerable capacitance across the LED or LED string and charge pump this with the controller. While this complicates the PWM/Light output ratio, it reduces the chance of flicker when running separate LEDs or strings on separate controllers.

Answer 3

Keeping the duty cycle at 80%, but increasing the frequency to 400 Hz, does the human eye perceive these two situations as the same brightness?

Assuming the LED is being switched on an off with a perfect square wave, yes. The human eye can't see much over 24Hz and the brightness is perceived the same. As long as the power is equal, the brightness will be also, and also the 'integrated' brightness by the human eye.

There may be problems if there is significant capacitance that changes the waveform, this would make 200Hz not the same power as 400Hz, so if you switch make sure the rise time on the LED does not change significantly from 200Hz to 400Hz.

If a detector can't tell the difference, there's a good bet that an eye can't either.