The Shower Test - can you spot PWM?

Oh, okay. I was reading it as 2 ms for the entire display, not 2 ms per grid square. So, yeah, that’s very slow PWM. :slight_smile:

Nope, no rolling shutter. I avoid those. Everything in the pictures was a single moment in time.

The larger dotted lines were caused by water droplets changing shape, not by camera artifacts. That’s one of the main reasons I don’t recommend the shower test — even if the light doesn’t pulse, the water droplets kinda do. And it can be hard to tell which is which.

Sounds like a terrible testing method then! I wonder if there’s a good way to positively identify PWM that doesn’t involve a shower or high end equipment… maybe a homemade Phenakistiscope? Would just need to get a frequency that’s some multiple of the PWM and it’d flicker more distinctly, I think.

Both PWM and CC can produce visible flicker and can produce dots in falling water, and dots when waving the light.

And both can produce distortion in a video or photo.

The question is not whether a light has PWM or CC, the question is whether the flicker is visible to the naked eye.

Waving a white index card through the beam works well. For fast PWM it requires waving the card quickly though.

Quickly waving a light usually makes bad PWM very obvious to me.

I can also see it by moving my eyes instead.

Early LED tail lights and cheap LED holiday lights drive me crazy. :stuck_out_tongue:

I was thinking (never a good sign) about this PWM frequency stuff and how to get a look at it without having to use a DMM with a frequency function or a scope.

Several methods have been put forward about converting photons into volts, most of which cost almost nothing other than some tinkering. What’s needed some way to convert that output into something useful. Something free maybe, as we are the BLF…

Since most PWM switching takes place in the audio ranges, why not try some audio analyzer software?

RoomEQ Wizard is free multi-platform audio analysis software widely used by audio junkies to measure acoustic properties of rooms and loudspeakers. It’s lots of fun to play with and there are plenty of videos and forums on how to use it. Find it here:

Instead of a microphone as an input, use the output of whatever you can cobble together to convert light to volts. Then use the real time spectrum analyzer to look at the frequency of the signal. There should be a sharp peak at the PWM frequency. I would expect to see harmonics from the square wave or other noise, but I think this would work.

The only got’ya would be matching the voltage output of your source with the input of the computers sound card. Most audio line voltage is in the 1ish volt range. Microphones run lots lower, but I’m not sure what the computers inputs can handle.

REW (nickname for the software) can also act as a signal generator. One could run the output of the sound card through a simple switching transistor or a small amp to provide the correct voltage to drive an LED. Then you have an infinatly adjustable (well 10Hz to 15K-20KHz depending your sound card) to experiment with PWM.

You could even hookup a speaker to your photocell and put the speaker next to the computers microphone to get a signal in that way.

I’ll see if I can do this sometime and report back.
All the Best,
Jeff

Jeff, I’m looking forward to seeing the results you get. IIRC, there has been some experimenting with using a sound card like this in the past on BLF. A quick search would probably pull up an old post or two about it. I think there are even apps designed for it, or there used to be. Or maybe I’m remembering wrongly.

It works. I’ve used a guitar tuner app (PitchLab) and a mic to measure PWM. It works particularly on medium-to-high modes when pointed at black fabric with very small fibers, thanks to the photoacoustic effect.

The hard part is getting the signal into the app. It either needs to be amplified enough to be heard by a microphone, or it needs to be otherwise routed in more directly.

Selfbuilt on CPF used a sound card frequency analyzer to measure PWM in his reviews. He made one of the largest and most respected collections of flashlight reviews ever created.

However, even that can’t detect anything faster than the Nyquist rate, or about 22 kHz on most devices. And even below that, there’s typically still a lot of aliasing until at least one octave lower, or ~11 kHz. When reviewing a BLF light with fast PWM, he claimed it had no PWM, just “noise” of some sort in the signal.

So as soon as I got a DMM with the right features, I switched to the LED-as-sensor method.

Well, most sound cards now can do at least 48khz and there are sound cards capable of 192khz for professional use. They don’t look or act like what most people think of “sound cards” but still… :smiley:

Right, but before I buy a 192KHz sound card, I'm buying a $20 DMM and a photodiode or LED or solar cell (or all three)

Okay actually I found a $40 sound card that claims something to do with 192KHz, but that was also admittedly a very quick google search.

Well, “Pro” sound cards capable of 192khz have been around long enough that I wouldn’t doubt you could get a “consumer” grade one now with that high frequency. I haven’t tried looking. But since mobos come with 48khz, 7.1 channel, surround sound on-board, the standalone cards would need to have some feature that the mobo chips don’t yet have. It would make perfect sense.

The main features in add-on sound cards are a better signal-to-noise ratio, more recording channels, studio monitor ports, more types of plugs, mixer controls, phantom power for mics, line-level adjustments, preamps, lower latency, and things otherwise intended to make music production easier or higher-quality. Usually they’re called audio interfaces.

For example, the Arturia Audiofuse is popular lately. Another nice one is the Sound Devices MixPre-10M.

Audio spectrum analyzers work fine for figuring out the PWM frequency. Here’s two programs I regularly use: REW and ARTA.

I used an M-Audio audio interface connected to a Thorlabs photosensor.

Funny pics.I never look at such big drop to see such oscillations. Very interesting effect. Thanks for the information.
Usually I look at many small drops far from origin (1-1.5m) in such way PWM up to 20-25KHz can be seen. Low PWM frequency <5kHz , I can see “directly” by eye movement.

For me minimum comfort frequency which I don’t notice at real using is about 10KHz.

I’ve done some tests with a function generator and 8kHz is usually fast enough for me. The limit depends on the duty cycle, but 10kHz seems a safe minimum for me too.

It may not be obvious from the picture, but the drops aren’t big. The big one on the left with a dotted line next to a streak was only about a millimeter in size. The smaller one at the top of the right image, also showing a dotted line effect, was maybe a tenth of a millimeter wide. This was taken very very close to the stream. The visible part of the shower head is only about 1cm. Most of the drops are moving faster than terminal velocity, and they slow down as they fall.

Same here. I find PWM distracting during use if it’s slower than 10 kHz. Above that is comfortable though.

Oh thanks,I see. It is not big , but close to shower head. For small drops “relaxation paths” should be smaller, than for big one.

I will occasionally hear PWM that I can't see, so I'd like to keep it above ~18KHz just in case.

So as not to hijack this thread,
I did a post about using Audio software to measure PWM Frequency:

I just bought one of this cheap multimeters because you mentioned the ZT102 for duty cycle measurements, but there are more interesting ones.

Like the AN8009

Has a µA range,
ca. $25
Safety is not good because of a wrong CAT rating, like all cheap ones.

I sniffed around and found the
ANENG AN870
This has a 10mA range (the AN8009 miss this)
Has everything was the smaller ones have and runs on AA
$26 in the Ali-App and a coupon brought it down to $22
And suddenly my credit card was charged.
Damn! I need a new meter as much I need a new light.
https://m.aliexpress.com/item/33004463675.html

HKJ tested a lot of nice cheap DMMs
Most of the new ZT Zotek or AN Aneng have a duty cycle.
http://lygte-info.dk/info/DMMReviews.html

A nice geek DMM is the 121GW from the eevBlog. It costs a bit more.
https://lygte-info.dk/review/DMMEEVBlog%20121GW%20UK.html