PWM - What is it, How does it work and how to detect it.

[Cataract]

Coool! That's a great visual!

I notice outddors because it rains alot, so any wet surface can be reflective..
Hate the pwm flicker on tail lights on the newer cars. Slow pwm led lights should have been banned on cars. The faster the car in front of you moves, the more visible the pwm gets. In every curve, or bump in the road, the flicker from the tail lights grabs my attention. The roads around here are made of bumps and curves only

...and also whenever you move your eyes around...ach! :duck: I do everything to avoid them; slow down, pass them or just take another street when possible. I do see less of them lately, but the age of darkness is upon us; what normally is a good time for flashaholism now has it's demon! I did notice that the worst ones (I prefer not to name the 2 specific brands here) seem to have remedied to that on newer models, but other makers are making the same mistake all over again. Hopefully there will be a law against that sometime soon... hopefully.

 

[Yagon]

Here's a picture I took of a 123² Tactical (R2) before I sent it back because the flickering was so bad.

On the left is "moonlight mode", middle is "low", and on the right is "medium". "High" and "max" showed no flickering.

After I got so fed up with the flickering I sent it back and got the S2 123² Turbo, which as I said, flickers on the "medium" level. My 123² Tactical warm white flickers on "low" and "medium", and my RGB neutral flickers on "moonlight".

I have a Mini AA and its PWM is much higher frequency than the flickering coming out of the Quarks.

From what I remember about the thread that discussed this issue, the flickering is not so much a sharp "on and off" cycle, but rather an "on and fade" cycle.

Nice picture! that's almost art, but what is shown here is definitely PWM. I have never seen such bad PWM on any of my 4sevens lights, but there are people who could say a lot more about 4sevens and the use of PWM.

Could that be a failed storage device (blown cap or inductor overheated and short cct) in the switching regulation ?

 

[Cataract]

Could that be a failed storage device (blown cap or inductor overheated and short cct) in the switching regulation ?

Yes, it could be a blown or defective component (also including frequency pulser), bad solder or cracked board whether the problem was from manufacturing, shipping/handling, overheat of even freeze. I dropped my 4 sevens lights many times with no such issue though, so I tend to think it was a defect from the manufacturing side. This kind of problem might be hard to catch during QC since the light generally behaves the way it should.

 

[Yagon]

Yes, it could be a blown or defective component (also including frequency pulser),

Do you mean the cmos switch ?
Thought that would either fail open or closed cct .
Doesn't have to mean a manufacturing error(of the flashlight) too , it could simply be a case of infant mortality of the component .

 

[Cataract]

Do you mean the cmos switch ?
Thought that would either fail open or closed cct .
Doesn't have to mean a manufacturing error(of the flashlight) too , it could simply be a case of infant mortality of the component .

Yes; I've seen CMOS and other types of clocks ticking at the wrong rate because of manufacturing defects or because the signal wasn't steady enough (because of a manufacturing defect again) for other circuits to function normally and even because of premature death of only part of the IC. Most of the time things are black and white with modern IC's, but not always...

And yes again, it could be premature death of some component. 4Sevens is not necessarily to blame on this, but a faulty solder or trace could possibly cause this type of problem too; I've seen weird stuff in my career. Without their blueprints on hand it is hard to rule out anything for sure, and even then... That's why manufacturers usually want to have the defective product back; so they can find out what went wrong and prevent it from happening again.

 

[agnelucio]

I don't think that dropping a light would cause significant damage to the electronics (except perhaps for a large inductor). Much more likely to be a dry/dodgy solder joint.

 

[agnelucio]

..the correct interpretation of your math is not really that it will look like it is flashing at 50Hz, but that it should be as easy to detect as if it was...

Yes, sorry, that's exactly what I meant. (and should have said).

 

[Cataract]

I don't think that dropping a light would cause significant damage to the electronics (except perhaps for a large inductor). Much more likely to be a dry/dodgy solder joint.

Very likely, but I've seen stuff happen. Since we do not know what was the actual cause, we can only propose theories which is not the scope of this thread (my bad for digressing so far out). Let's get back on track with PWM

Yes, sorry, that's exactly what I meant. (and should have said).

:thumbsup:. We are only humans using keyboards

 

[tattoosteve99]

Just read through this. Excellent write for me. I was wondering about this and now it's clear. Thanks so much!


Sent from my iPhone using Tapatalk

 

[Cataract]

Thanks Steve. Knowing I could make it clear for fellow CPFers makes my day :twothumbs

 

[hjdca]

:huh: PWM can be easily seen by pointing your light at your shower head, with water running of course! The water may look like separate drops depending on the PWM frequency. My Pelican 9410 does use PWM for the lower mode, but it is of sufficient frequency that it's only noticeable in the way described.

Try this with your lights, let me know what you see.

I tried this with 5 four lights: Draco (Flupic), Haiku (GDuP, 4 level), Mac's Tri-EDC, Jetbeam TC-R1, & P60 dropins for my TnC from Vinh and Nailbender... The Draco with Flupic had the worst PWM on low and user defined settings.. Very choppy... On high, there was no PWM... The Haiku with the GDuP driver had no perceptible PWM on any level except some slight "chaining" at 10 ma... At 200 ma all the way up to 1200 ma, there was no PWM... The TC-R1 had no PWM at any level, even the very, very low settings had no PWM... Mac's Tri-EDC and the P60 dropings from Vinh and Nailbender behaved exactly the same.. High quality, high frequency PWM at low and medium levels and no PWM on high... The PWM at these lower levels was excellent, just some high speed "chaining".... This experiment really makes me appreciate the GduP driver and the TC-R1 driver... Thanks for the tip !!

 

[hjdca]

This experiment really makes me appreciate the GduP driver and the TC-R1 driver... Thanks for the tip !!

Actually, I am more impressed with the GDuP than the TCR1 because of the buck, boost drivers... The GDuP will do its thing with a primary and a rechargeable... The TCR1 does not have this capability and you have to use a rechargeable... Other DC engines from China also seem to be missing buck drivers..., ie. you put a 4.2 volt rechargeable in it and all you get is high, like Fenix...

 

[Hondo]

Could that be a failed storage device (blown cap or inductor overheated and short cct) in the switching regulation ?

Check my post at the end here, along with others observations:

http://www.candlepowerforums.com/vb...Question-on-Quark-Tactical-and-PWM&highlight=

This is more than one or a few lights with bad components, and David considers it normal, and not PWM. It is apparantly "warble" in a constant current controlled circuit. It is interesting as it seems to be the only case of a non-PWM controlled light looking exactly like a PWM light, and at a fairly low frequency. The reduction to 30% brightness rather than 0% as in true PWM does not seem to lessen the effect noticeably. Just when you thought you were safe from flickering lights by ordering a 100% current controlled light, there comes an exception.

My interest in the other thread centers on the different flavors of this circuit. Most Pro interfaces seem steady, but I have one that all three of the lowest levels do it. Most reports are for the Tactical interface, and only on the low and medium. But there are also plenty of the same lights without the phenomenon.

 

[Cataract]

Thanks Hondo, I was just about to point to that resuscitated thread. Even though the flickering effect in Quarks is not supposed to be PWM per se, our eyes perceive it as the exact same phenomenon and the effects on perceived light intensity will obey the same law of average output. Usually, though, this flickering in Quarks is very high frequency (at least in all of my numerous Quarks) and very hard to see, as pictured in my original post:

Anything significantly noticeable sounds unusual to me and I too would find it very annoying.

 

[Hondo]

I managed to grab this image of three lights. The one on the left is my Quark 123^2 Tactical, wearing a filter to remove green and make it more neutral. The center is an ITP A2 Eos, known to be in the 2K+ Hz range. The last is the notorious 100 Hz Fenix L0D. They are all on their medium level.

I would give the Quark, based on the Fenix being 100 Hz, about 350 Hz. I somehow don't get the graph linked by cave dave in the other thread, as it shows a full cycle in what is labeled about 0.5 ms, which would be 2 kHz. Yet he says it supports his guess of about 200 Hz. That was measured by HKJ, and I am guessing it is the non-objectionable artifact that David speaks of in a light which does not exhibit this "PWM-like" behavior.

Here is another pic with just the Quark, in which I really cranked up the speed I swept the light. This emphasizes that the brightness is indeed not going to zero, just getting a lot dimmer, so this is not PWM. On the Fenix, it is clear that the light goes completely dark between cycles.

 

[reppans]

I get a similar thing with my D25A clicky on moonlight mode, but only with a low battery. However, I also have another copy that has this PWM-like behavior with fresh cells. Here's an old picture of it next to a Quark AAX... (FWIW, I have many LV head Quarks, and one HV, and have not noticed any PWM-like behavior.)

 

[inetdog]

FWIW, particularly in large LED replacements for incandescent in luminaires, the presence of a phosphor for CRI improvement can partially mask the PWM. The relaxation time of the phosphor can be greater than the period of the pulses.

Tapatalk...

 

[Cataract]

Great pics guys! I really like the idea of holding multiple lights together so you can have a fairly precise number on frequency based on a known one (provided you take the time to count all the bright spots)

FWIW, particularly in large LED replacements for incandescent in luminaires, the presence of a phosphor for CRI improvement can partially mask the PWM. The relaxation time of the phosphor can be greater than the period of the pulses.

Tapatalk...

Excellent to info. Never even thought of that.




Sent from somewhere in a tunnel, 42 minutes away from anywhere.

 

[KARADENIZ]

current-controlled, PWM nice explained thanks
My Sipik SK68 is not CC or PWM, is it possible to update it ?

 

[Cataract]

current-controlled, PWM nice explained thanks
My Sipik SK68 is not CC or PWM, is it possible to update it ?

:welcome:

Thanks! I will add one or two things to the original post eventually, but for now I'm still unboxing stuff.

I'm not familiar with the Sipik as I don't own one or ever seen one in person. Try in the "homemade and modified flashlight" section; Almost every flashlight has been modified by someone at some point