The main reason to keep the frequency as low as possible is that there is less switching loss.
It should be apparent that the more times you have to switch the current the higher the total loss.
The upper limit would likely be due to switching and other parasitic losses, long before you reach any LED limitations at those currents.

I was going to post this question on the original thread, but then I thought that maybe it deserves a thread of its own... and besides, I ain't no hijacker...

It is definitely not apparent to me why there are losses when one PWMs an LED. So I did my homework before posting, and found several interesting sources of info. The file attached is the one I found to be most useful at explaining things a bit.

I found this paragraph rather interesting:

For the diode to transition from the conducting to the non-conducting state, the charge distribution must change. This can only happen with a movement of charge, which is a flow of current. In some cases, such as a silicon carbide diode, the charge distribution difference is caused solely by the junction capacitance: again a movement of charge occurs when moving from the conducting to the non-conducting state.

So it seems that yes, the diode's capacitance is partly responsible for these losses, but the problem is not as simple as it seems (at least to me). It also seems that the diode's reverse recovery current plays a role, but also what the diode is being switched with:

The important conclusion is that the use of a soft recovery diode will in troduce more switch-on losses
in the diode itself, but save additional losses in the semiconductor switch. When evaluating the performance of a new diode, it is therefore necessary to look at both the diode and semiconductor switch performance, not just the diode performance.

That last observation makes it understanding it even more complicated for me... Is there a simpler way of explaining it? Or is this as simple as it gets?

 

May be off base here, but does the paper in the link apply to your question? In a PWM situation for controlling led's where does the reverse current on the mosfet come from? Wouldn't it just be a "pulsing" low side switch? The paper was geared more to motor drive type circuits where there is a reverse EMF on the diodes in question, isn't it?

 

I have to admit I chuckeled a bit when I saw the tag about switching losse on a PWM LED. Yeah, they may be some but it is not anyting significant.

Switching losses come about when the switching device (transistor, FET) turns off while there is current. Until the device is fully off there is a period of time where the voltage across it is rising while the current is falling. Same is true as it turns on. These two short intervals account for the large majority of losses in a switching power converter.

With a LED... not so much. You don't have predominating inductive currents to contend with.

The one product I designed with a blinkie Led for PWM dimming ran at about 200 Hz. That was as slow as I could get the PWM in my PIC to run and had no issues.

Additionally, I am one of those people who can percieve fast blink rates: I see taillights on cars on the road as a series of flashes. But a 200Hz blink rate looks very stable to even me.

 

Switching losses aren't significant and are just a cost of doing business. You're saving in other areas.

 

Additionally, I am one of those people who can percieve fast blink rates: I see taillights on cars on the road as a series of flashes. But a 200Hz blink rate looks very stable to even me.

Those LED taillights are one of my pet peeves. The fast on/off are annoying. It baffles me why that wasn't a design consideration. I'm surprised it doesn't trigger seizures in people with epilepsy (as some video games have been shown to do).

WRT to being able to perceive fast blink rates, how are you able to watch TV or movies or be around any line powered lights? It would be awful to be able to perceive 30 or 60Hz flicker.

 

With a LED... not so much. You don't have predominating inductive currents to contend with.

That's what I thought at first, but then @crutschow's comment threw me off base a little bit, since I'm just an amateur and I have the highest regard for his opinion.... then again, maybe those very small switching losses might be significant in battery powered devices?
BTW, I was going to ask you the same thing as @dl324... it must be very annoying for you not only watching those tail lights, but also some fluorescent lighting and even the strobe effect on tv

 

WRT to being able to perceive fast blink rates, how are you able to watch TV or movies or be around any line powered lights? It would be awful to be able to perceive 30 or 60Hz flicker.

I can't see lights blink, TV looks just fine too. I never see the stroboscopic effect I see on tail lights.

It is worst on highway driving: if I scan the road well ahead of me from one side of the road to another (something I highly recommend doing) a new car will get sensed at a very primitive level of my brain as MOVEMENT and of course guide my attention back to just another car doing the same speed as the rest of us are doing.

It’s annoying to fairly regularly getting a startle warning over absolutely nothing requiring my attention that does indeed draw my attention to places it need not be. At least I'm learning what the cause is and not looking to see something I saw and now lost track of.

I only have one brain and I want it focused at the problem at hand, not looking for the damn blinkie lights!

 

... a new car will get sensed at a very primitive level of my brain as MOVEMENT and of course guide my attention back to just another car doing the same speed as the rest of us are doing.

I have a similar problem as yours... I also sense nonexistent motion sometimes. But it's due to "floaters" inside my eyes that cast shadows rather than lights. I've been told by my eye doctor that they're a normal thing on people my age.

 

Switching losses are generally low for a well designed circuit. It obviously does go up with frequency but can still be acceptably low at higher frequencies (such as for tens of KHz switching).
My answer was just an observation of that effect, not as to how significant (or not) it may be in a particular application, and to note that there's no advantage to pushing the switching frequency above that needed to avoid any flicker effects.
The exception to that is, if you are using an inductor to smooth the PWM and/or improve the efficiency, then a higher frequency has the advantage of reducing the size of the inductor for a given ripple.