Let's look at PWM from the other side

[Timothybil]

We have all seen PWM on scope traces through the efforts of Selfbuilt and others, but has anyone ever seen a trace of the actual light output of an LED under PWM? I know there has to be some latency between the voltage rise and the corresponding output. I think it would be interesting to record the output of a photocell responding to an LED being fed with PWM, and to be able to look at the PWM trace and the corresponding output trace side by side.

 

[maukka]

All of my measurements are done optically from the emitter, not from the driver. The LED driven by PWM responds to the sudden rise and fall in voltage very fast. The phosphor has enough time to pretty much completely dim with PWM even at hundreds of kilohertz.

 

[kpatz]

This appears to give a lot of information on LED latency:

https://electronics.stackexchange.com/questions/86717/what-is-the-latency-of-an-led

Phosphor LEDs, like the white ones in flashlights, are the slowest responding, typically in the hundreds of nanoseconds range. But most LEDs are quite fast.

The components driving the LED often add more latency than the LED itself, with parasitic and junction capacitance, trace inductance, MOSFET switching times, being the biggest factors. LEDs used for high-speed signaling are specially designed to minimize parasitic capacitance, resistance and inductance to improve switching times. LEDs used in flashlights don't need to be super fast to respond, and you wouldn't want them to anyway since it would make the flickering more apparent to sensitive individuals. Even so, they are fast enough to turn completely off and on on each PWM cycle, and it's our "persistence of vision" that makes it look like a steady light to our eyes.

In many designs, such as the commonly used low-side MOSFET switching, turn-off time is slower than turn-on time, so the LED fades out a bit slower at turn-off than it rises at turn-on.

 

[HKJ]

When I reviewed flashligts I used a photo sensor to record pwm.
Here a few examples from a Xtar Wk26 on medium and low:

Leds are very fast to turn on and off.

 

[stephenk]

Photo showing PWM in action in this review http://lightpaintingblog.com/flashlight-review-generic-cheap-zoom-light/
I would link some of my light painting photos using PWM for effect but I'm currently away from my PC. Maybe in a few days.

 

[DavisonDave]

You can see it when it's snowing. I've done a lot of winter backpacking. With my headlight on at less then full brightness I can see the falling snow flakes jumping down as they fall. The lower level you use the farther the snow flakes appear to jump. This is the only situation when I see the flickering from PWM.

 

[NoNotAgain]

You can see it when it's snowing. I've done a lot of winter backpacking. With my headlight on at less then full brightness I can see the falling snow flakes jumping down as they fall. The lower level you use the farther the snow flakes appear to jump. This is the only situation when I see the flickering from PWM.

If you're able to identify PWM in snow falling, you'd also see the same thing in rain.

The typical test used by a layperson is the light shining at a moving fan blade. If the blades appear to be moving stroboscopicly then the light has PWM.

 

[KITROBASKIN]

Adjust the rapid dripping from a water faucet to see flagrant Pulse Width Modulation.

 

[Timothybil]

Thanks for all the info. I didn't know if there was any kind of chemical type interaction in an LED firing, which is really what brought about the question. It's obvious to me from your answers that there is none, it's all state and electron shell jumping which is for all intents and purposes instantaneous. I never even thought about the response delays of the various components of the driver. I guess one learns something every day. Thanks again.

 

[archimedes]

I don't think it could necessarily be considered truly "instantaneous" due to the Stokes effect, but very fast for sure.

Nice basic summary may be found here ...

https://www.digikey.com/en/articles/techzone/2011/aug/whiter-brighter-leds

 

[Timothybil]

I don't think it could necessarily be considered truly "instantaneous" due to the Stokes effect, but very fast for sure./QUOTE]
That's why said 'for all intents and purposes'. Nothing is instantaneous. Sometimes the interval is so small as to be unmeasurable with current instrumentation, but not instantaneous.

 

[archimedes]

That's why said 'for all intents and purposes'. Nothing is instantaneous. Sometimes the interval is so small as to be unmeasurable with current instrumentation, but not instantaneous.

But I do think it should be measurable with current instrumentation .... In fact, it seems that the short (but apparently clearly evident) delay to switch "on" and switch "off" is seen in @HKJ post, here ...

https://www.candlepowerforums.com/posts/5174250

If it were "instantaneous" , or so nearly so as to be "unmeasurable with current instrumentation" , then the upstroke and downstroke in those graphs should be perfectly square, instead of slightly sloped.

 

[HKJ]

But I do think it should be measurable with current instrumentation .... In fact, it seems that the short (but apparently clearly evident) delay to switch "on" and switch "off" is seen in @HKJ post, here ...

https://www.candlepowerforums.com/posts/5174250

If it were "instantaneous" , or so nearly so as to be "unmeasurable with current instrumentation" , then the upstroke and downstroke in those graphs should be perfectly square, instead of slightly sloped.

Not really, there are 3 parameter than can slow the turn on/off:
1) The driver
2) The led
3) The measuring equipment

In this case I believe it is the driver.

 

[archimedes]

Not really, there are 3 parameter than can slow the turn on/off:
1) The driver
2) The led
3) The measuring equipment

In this case I believe it is the driver.

Yes, of course response time will be affected by all of those, but my point was that the Stokes effect takes "measurable" time.

I am not a materials scientist expert, but a brief websearch suggests it may occur on the order of tens of nanoseconds.

I would think that (even so brief) duration should be measurable, with proper test environment and appropriately sensitive equipment.

As I searched around on this, I found a fascinating article on optical autocorrelators used to infer timing of laser light pulses lasting on the order of femtosecond duration. It also mentioned that light flashes lasting picoseconds can be more easily and directly measured.

I'm not planning to try to split hairs further on the meaning of "instantaneous" , however very interesting and thought provoking concepts involved here ...