100Hz flicker in lighting

[ocset]

I’ve still not found conclusive proof that 100Hz flicker in lighting does any harm eg headaches, eye strain, tiredness, nauseousness. Does anybody have any?

By 100Hz flicker, I am referring only to those which have less than 1ms of dead time..(ie no light regions  a maximum of  1ms out of every 10ms)

Malfunctioning T8 tubes (flu or led) sometimes actually flicker at 50Hz …which is bad.
Old skool Triac dimmed incandescants had massive 100Hz flicker but never had any complaints despite in the 70’s and 80’s millions of people used them.
Also, driving on a busy motorway in the dark when there’s fence posts in the central reservation causes  headlight flicker in the frequency range of epilepsy, but there are no reported crashes from it.

Derlichtpeter  says  100Hz flicker is bad…but doesn’t list the proof.
https://www.derlichtpeter.de/en/

The proof, as  you know, would involve 500 random people being made to live in 100Hz flicker , and another 500 to live in “no  flicker” lighting…for a week……….we would then see who had the most complaints…if any.
Also, 100Hz flicker lighting could be sneaked into offices, where “no flicker” lighting had previously been installed…to see if there were complaints.
Also, 100 people would be invited to go into 10 rooms, each with a light that either flickered at 100Hz or didn’t…………..they would be asked to state if there was flicker or not.

The test results should not be compiled by reps associated with large lighting co’s, or  those associated with electrolytic capacitor sales, etc etc.

This already got discussed here, but there was no definitive proof given….the site warns  and says start again a new thread since its 120 days old.
https://www.eevblog.com/forum/projects/100hz-flicker-in-lighting-is-a-problem/

[mcovington]

This is related to the old question of whether watching TV (on a black-and-white CRT, 1950s-style) is harmful to the eyes.  There used to be a lot of concern... CRTs in those days flickered.  Color CRTs later on had longer persistence, and as I understand it, digital flat screens do not flicker.

A test instrument to measure the flicker of lighting would be handy to have.  Anybody care to create one?

[james_s]

This has been discussed before. Harm? I don't know, but headaches and discomfort? It very much causes that for me, I find the strobing extremely annoying. What more proof do you need? I can't let you inside my head but if you turn on a light in front of me I'll tell you with 100% accuracy whether it's flickering from 50/60Hz power or steady on DC. You don't need hundreds of people, if just one can see the flicker and is bothered by it while 499 can't that doesn't mean the flicker is not a problem, it just means it's not a problem for a majority of the people you tested and that's not a license to ignore the discomfort it causes the one. Just a few weeks ago I was at my mom's place and she didn't understand what I was talking about when I said her LED christmas lights flickered, so I whipped out my mobile phone and took a slow-mo video of them in which it was plainly visible that they were strobing, she was amazed to see that and still couldn't see the flicker at all looking at the lights while I could.

As far as a tool to measure this, it's called an oscilloscope and has existed for decades. You can connect a photodiode directly to an oscilloscope with no additional hardware needed at all if you're just looking at mains frequency flicker. I've been tempted to pick up one of those $25 toy DSOs specifically for this purpose.

[kjr18]

As far as a tool to measure this, it's called an oscilloscope and has existed for decades. You can connect a photodiode directly to an oscilloscope with no additional hardware needed at all if you're just looking at mains frequency flicker. I've been tempted to pick up one of those $25 toy DSOs specifically for this purpose.

Actually I made something like that. I used a solar cell from solar "recharged" key-chain flashlight (recharged witch quote marks because inside was only two 2016 batteries and real solar cell not connected to anything) with some resistor laying around as a load. Connected both sides to DSO150 and you got portable flicker and pwm frequency meter. Works on mains led lamps, works on torches (if you want to know frequency and duty cycle on lower power settings). Obviously has some limitations but for this purpose is great.

[tooki]

For me, while it doesn’t cause headaches, I do find it exceedingly distracting (and thus somewhat fatiguing, especially for fine work), especially when dimmed (causing the duty cycle to drop), because quick eye movement (saccades) causes point light sources and specular (point) reflections to become visible as trails of dots of light. This is called the “phantom array” effect, and since eye motion is involved, not just persistence of vision, it’s been calculated that fully eliminating it in all situations requires a PWM frequency of 30KHz (though I’ve found that 3KHz is enough to fix it for home/office lighting for me).

Another key phrase to look up is the “flicker fusion threshold”, and just FYI, I can already tell you it’s way, way higher than the 60Hz often claimed (which really only applies to moving pictures, and even then only when looked at straight on. Our peripheral vision, on the other hand, is far more sensitive than that. As someone who is very sensitive to flicker, I can readily perceive the flicker of a 100Hz CRT or even a 450Hz dimmed LED when in my peripheral vision.)

[tooki]

As far as a tool to measure this, it's called an oscilloscope and has existed for decades. You can connect a photodiode directly to an oscilloscope with no additional hardware needed at all if you're just looking at mains frequency flicker. I've been tempted to pick up one of those $25 toy DSOs specifically for this purpose.

Actually I made something like that. I used a solar cell from solar "recharged" key-chain flashlight (recharged witch quote marks because inside was only two 2016 batteries and real solar cell not connected to anything) with some resistor laying around as a load. Connected both sides to DSO150 and you got portable flicker and pwm frequency meter. Works on mains led lamps, works on torches (if you want to know frequency and duty cycle on lower power settings). Obviously has some limitations but for this purpose is great.

Yeah!! I actually bought a small solar cell for exactly this purpose, just hooked up to my DS1054Z instead. It’s neat to see how different light sources behave.

[ocset]

Thanks, I have even seen 100Hz flicker myself……but it was on a 100Hz  source that had some 9ms of dead time, and the  adjacent 10ms pulses were of slightly different LED current magnitudes…..ie, they alternated from slightly higher in peak for 10ms , to slightly lower in peak for 10ms , continuously……….so it was really 50Hz flicker that I was actually  seeing due to the poor design of the product.

We probably all know the essentially 50Hz flicker of  worn out flu tubes which are dying and so conduct more in 1 direction than the other…….even I wouldn’t “choose”  to work in that…..though I never got sick from it. At my old college we used to work in loads of rooms that had that though, without problem.

The question for the 100Hz flicker  see’ers, would be, what  depth of flicker is it that can be detected by human eye, and what depth is it that causes sickness.

I worked under deep 100Hz flicker for 2 years at one place. I never got sick and only knew it flickered cuz of the phone cam.
.....But there was just one time when I felt  very slightly nauseous  and strangely , could see like a kind of weird kind of  flickering in my peripheral vision…..and a kind of mistyness in my overall vision. –But I put this down to the super strength coffee that I had just drunk.
I stopped working, took deep breaths, drunk a load of water, and felt quite a lot better, and the flickering went…I then went  for a big mac meal at lunchtime, and felt fine, and had no further problems therafter.

[StillTrying]

"The question for the 100Hz flicker  see’ers, would be, what  depth of flicker is it that can be detected by human eye."

Under this 3W LED lamp I can easily detect the deep 100Hz light ripple on the edges of anything that moves, but it's not toooo bad to work under, compared with some others, such as square PWMed.
https://www.eevblog.com/forum/chat/20w-halogen-bulb-viewed-by-a-photodiode/?action=dlattach;attach=357367;image

Although there's no light level flicker, I can easily see the 25 frames/sec of a TV picture.

[mzzj]

As far as a tool to measure this, it's called an oscilloscope and has existed for decades. You can connect a photodiode directly to an oscilloscope with no additional hardware needed at all if you're just looking at mains frequency flicker. I've been tempted to pick up one of those $25 toy DSOs specifically for this purpose.

Actually I made something like that. I used a solar cell from solar "recharged" key-chain flashlight (recharged witch quote marks because inside was only two 2016 batteries and real solar cell not connected to anything) with some resistor laying around as a load. Connected both sides to DSO150 and you got portable flicker and pwm frequency meter. Works on mains led lamps, works on torches (if you want to know frequency and duty cycle on lower power settings). Obviously has some limitations but for this purpose is great.

You could also use smartphone and one of those ”oscilloscope” apps.

[SiliconWizard]

The only real effect I'm aware of is this: https://en.wikipedia.org/wiki/Photosensitive_epilepsy
not everyone of course is affected by it, and for those who are, there's of course a large variety of trigger stimuli and responses.

It's also a known fact that some visual patterns can trigger sleepiness, usually at much lower rates than 100Hz or even 50Hz (I think more like a few Hz), so the usual flickering from ligthing and monitors is not likely to trigger this IMO. Specific visual patterns on screen could, though (regardless of refresh rate here of course).

[trevatxtal]

Benign paroxysmal positional vertigo (BPPV), labyrinthitis and vestibular neuritis )
Are severely effected by any type of strobe effect whether by lighting  or paling/fencing carpets tiling and so on.
Can result in falls and in extreme cases  passing out, besides headache, defocusing of the eyes confusion and losing awareness.
There is a vast amount of medical evidence to this.
You can spend hours looking it up on search engines

[Mr Evil]

I believe you had plenty of informative responses in the previous thread. Nonetheless, the question prompted me to do a quick, totally unscientific test.

I have some decorative LED lights with variable brightness acheived by varying the PWM duty cycle. They were made out of spare parts, and the only thing I had suitable was a microcontroller with no PWM output, and I could only make it do software PWM at 120Hz, so the flicker is quite visible. I varied the brightness and looked to see if I could detect the flicker by waving my hand underneath. At <50% duty or less, it's very noticable, like a strobe light. 50-75% it's still visible, but looks just odd rather than flickery. >75%, it's hard to tell if I'm really seeing it or just imagining it.

I attached a photo of how I measured the duty, with a photodiode attached to a handheld oscilloscope.

I'm quite sensitive to flicker, and can easily detect LEDs driven at 100Hz, which is unfortunately quite a lot of them. This is also not a scientific test, but I have so far been 100% correct on seeing if the LED lights we buy are driven by PWM before checking them with my 'scope. Most other people can't see it, even if I tell them it's there and how to look for it. It doesn't cause any physical symptoms, but it's really, really annoying. I find that anything <200Hz is visible. Above that may be detectable in certain specific circumstances, but it's not bothersome.

[Prehistoricman]

as I understand it, digital flat screens do not flicker.

Not true! LCDs do not flicker, but some of their backlights do. I have a rather modern 1440p screen that has no backlight flicker, an older one with cold-cathode tubes that has a fair amount of flicker, and a laptop screen with hard 100% depth PWM dimming. I don't notice any of their flickering.

A test instrument to measure the flicker of lighting would be handy to have.  Anybody care to create one?

I connected a solar panel direct to an audio jack so you can listen to the various flickery things (7-segment displays, decorative (maybe RGB) lighting, screens, lights).

[ocset]

Thanks,
It seems that  EN12464 has banned anything with CFD >50%...so thats all PWM'd LEDs at 100Hz, which Mr Evil kindly confirms in effect.

CFD = compact flicker degree

please see six light waveform  diagrams here....
https://www.derlichtpeter.de/en/light-flicker/cfd/

...But it seems that EN12464 does not give any advice on whether or not  {12% >CFD <50%}should be banned?

It would be very interesting to find out if the USA/Australia/NZ etc have an equivalent flicker regulation to EN12464?

[StillTrying]

I attached a photo of how I measured the duty, with a photodiode attached to a handheld oscilloscope.

You probably need a load resistor across the photodiode to discharge the scope's and PD's capacitance to improve the fall times. For only 100Hz 50k to 5k would do.


I don't think photocells are fast enough to see an accurate real shape of even 100Hz flicker, a photo transistor could be ~50 X faster, and a photodiode ~50 X faster again.

It seems that  EN12464 has banned anything with CFD >50%...so thats all PWM'd LEDs at 100Hz, which evil kindly confirms in effect.

Have you not made yourself some sort of light shape viewer to visually compare things, and even estimate p-p light ripple.

[Mr Evil]

You probably need a load resistor across the photodiode to discharge the scope's and PD's capacitance to improve the fall times. For only 100Hz 50k to 5k would do.

Thanks! I was able to go down to 7.5k before it started noticably affecting the sensitivity, and now the output looks much sharper.

[Prehistoricman]

I attached a photo of how I measured the duty, with a photodiode attached to a handheld oscilloscope.

You probably need a load resistor across the photodiode to discharge the scope's and PD's capacitance to improve the fall times. For only 100Hz 50k to 5k would do.


I don't think photocells are fast enough to see an accurate real shape of even 100Hz flicker, a photo transistor could be ~50 X faster, and a photodiode ~50 X faster again.

Regarding photocells, it's the same story as photodiodes. I just ran a test with a PWMed LED captured with a 1"x6" solar panel and loading it with a 5k-10k resistor allowed you to see up to 2kHz with relatively square edges.

[NiHaoMike]

Perhaps this "light to sound" kit would make a great design to adapt?

[StillTrying]

Regarding photocells, it's the same story as photodiodes. I just run a test with a PWMed LED captured with a 1"x6" solar panel and loading it with a 5k-10k resistor allowed you to see up to 2kHz with relatively square edges.

I think I meant photovoltaic or solar cell rather than photocell which could be a photoresistor, but still I consider them to be all similar bandwidth, with enough light and loading you can get some sort of audio frequencies through them, Which is nothing compared with the linearity and speed got from a properly biased and loaded photodiode! I'm surprised you got a 2kHz square wave from a solar cell, I didn't know they could work that fast.

Lots a light shape viewing here, - which were mostly just interesting diversions from phosphor LEDs on off shape viewing!
Here's the light from an outside street lamp.
https://www.eevblog.com/forum/chat/20w-halogen-bulb-viewed-by-a-photodiode/msg1760738/#msg1760738

[tooki]

Solar cells are silicon semiconductors, so they are fundamentally as capable of speeds as any other semiconductor.

Photocells (photoresistors/LDRs) are an entirely different matter.

[james_s]

Solar cells are HUGE though compared to photodiodes, as far as I know, it's the large capacitance that limits their bandwidth.

Still a modest sized photocell is fast enough to capture the flicker of lights up to several kHz.

[StillTrying]

Solar cells are HUGE though compared to photodiodes, as far as I know, it's the large capacitance that limits their bandwidth.

Yes that's what I was thinking, maybe they're OK for curved ~100Hz light shapes, I don't know anything about how linear they'd be.

I wish someone else would try some high speed linear light experiments I've done enough! Sometimes we see an op amp TIA but don't get to see it's response to a square wave light input.

[Prehistoricman]

Just for the sake of giving an example, I took a couple of measurements of my small solar panel.
C = 13nF
Rinternal = 87k

Therefore the bandwidth with a 5k load is 2.5kHz. Of course it's a bad detector for this kind of stuff with the relatively huge capacitance to internal resistance but, as said before with enough loading, you can tame it.

[james_s]

Yes that's what I was thinking, maybe they're OK for curved ~100Hz light shapes, I don't know anything about how linear they'd be.

I wish someone else would try some high speed linear light experiments I've done enough! Sometimes we see an op amp TIA but don't get to see it's response to a square wave light input.

There is a design wizard that allows you to simulate various designs. I haven't verified the accuracy but it's Analog Devices so I'd expect it to be trustworthy.

https://www.analog.com/designtools/en/photodiode/

[StillTrying]

https://www.analog.com/designtools/en/photodiode/

That looks OK, although I don't have any plan to try a 4GHz op amp.

I could do with a light source with known <10ns rise/fall times for testing, next time I play with it I'll try doubling or more the 200mA through 5mm blue and green LEDs to see if I can improve a bit on the best I got ~18ns rise/fall times.

[mzzj]

I was running around the house with battery powered oscilloscope and small solar cell loaded with 1k resistor (BW tested to be sufficient for 1khz square wave)
Classic iron ballast 2x36W  T8 fluorescent:


T5 2x54W electric ballast fluorescent:


Cheapo 10W E27 led bulb in kitchen:


T8 36W countertop fluorescent:


230v GU10 led spotlights in bedroom:



None of the lights flicker in my eyes but GU10 spots in bedroom are downright nasty otherwise. Any movement is really jerky. Even when reading a book and turning pages the movement jerks.
Bathroom GU10 led spotlights were bit worse than typical iron ballast 36W tube, motion in there is slightly jerky but not to the point of dizziness.
All of the E27 led bulbs currently in use had very low ripple.

Conclusion based on comprehensive one subject objective study: Keep the ripple amplitude below 30% and it should be "good enough"   

[tooki]

Nice measurements!

That strobe effect is exactly a manifestation of the “phantom array” effect I mentioned earlier. I know exactly what you mean about it being nasty.

[SiliconWizard]

Some flicker from lighting can also make some displays annoying to look at when the refresh frequency is not in sync (typically a 30Hz/60Hz display when you have 50Hz mains...)
It's particularly noticeable on some LCD displays (not talking about modern LCD panels for monitors, but rather the small LCD modules with no active matrix). I have a calculator for instance which display you can see flicker under such lighthing. A bit annoying.

[StillTrying]

T5 2x54W electric ballast fluorescent:

I had trouble viewing the small amount of light ripple from an electronic fluorescent in among all the EMI, it was a case of using the scope trigger on AC LINE and using ~32 X waveform averaging to extract the 100Hz light ripple from the noise. It was the same for the dim light from an indicator neon.

230v GU10 led spotlights in bedroom:

That looks like worse ripple than my 3W LED GU10 in Reply #7. Which doesn't give me a headache during the small amount of time I work under it, it just annoys me when I notice large movements are done in steps.

[bap2703]

Back from the graveyard.

I just bought an LED based light to replace a defunct florescent tube.

That's a 1/25 s exposure photo on 50 Hz line.


I am sensitive to flicker so I immediately noticed it.
I expected that in 2021 buying from a reputable brand would prevent that issue to show up, but no :/

[Marco]

I expected that in 2021 buying from a reputable brand would prevent that issue to show up, but no :/

An AC-DC converter is expensive and inside a lamp unreliable, this is as true in 2021 as it was before. A CYT3000A or similar remains attractive, especially if it has to be dimmable.

I wish the EU would just create a new standard for low voltage DC LED lights and kicked off the market for it (with a standardized way to encode a logic signal on the power rail for addressable dimming while they are at it). GU5.3 has to deal with all the 12V AC legacy bullshit. Should have been done a decade ago, but better late than never.

[DeanA]

There is a standard that covers this stuff:
IEEE 1789-2015 - IEEE Recommended Practices for Modulating Current in High-Brightness LEDs for Mitigating Health Risks to Viewers
https://standards.ieee.org/standard/1789-2015.html

[Marco]

That doesn't remove the problem of AC-DC converters inside lamps.

We need a 48V DC consumer standard (industrial and railed LED lighting already seem to be settling on 48V DC too). Hell, you could even reuse existing AC mains lamp bases and just block the mains voltage and refuse to operate in when not powered by 48V DC (though it would still make sense to make some pure 48V DC sockets). Exchange the mains switch/dimmer with a 48VDC lamp switch with integrated AC-DC converter and you can use the 48V DC lamps.

[james_s]

It seems to be an already solved problem, I have lots of LED lamps that don't flicker and have been lasting well. I see little reason to switch to using a separate converter for most applications.

[Marco]

Then you buy the next one and it's bad again,  because no led manufacturer advertises with some guarantueed level of "stroboscopic" effect. Philips Eye comfort for instance only advertise low "flicker" as measured by pstlm but they don't consider 100Hz flicker.

With DC power, true near zero flicker is cheap. No more need for them to lie with semantic bullshit. It would make doing the right thing cheap.

[richard.cs]

Solar cells are HUGE though compared to photodiodes, as far as I know, it's the large capacitance that limits their bandwidth.

Given a photodiode is just a smaller area of the same, you should be able to achieve the same bandwidth with a solar cell, but the load impedance would have to be scaled down in proportion to the area increase. That may or may not be low enough to be difficult in practice.

For me, I find 100 Hz flicker slightly annoying because of noticeable strobing with moving objects, but it doesn't give me headaches, etc.

[wraper]

This is related to the old question of whether watching TV (on a black-and-white CRT, 1950s-style) is harmful to the eyes.  There used to be a lot of concern... CRTs in those days flickered.  Color CRTs later on had longer persistence, and as I understand it, digital flat screens do not flicker.

A test instrument to measure the flicker of lighting would be handy to have.  Anybody care to create one?

Yes they do but not all and not always. Backlight is more often than not PWMed to adjust brightness, sometimes only in lower brightness range. Frequency can be as low as 200Hz. Some gaming monitors have special mode where backlight turns off during image transition time between frames, thus reducing image smearing.

[Tomorokoshi]

I ran across this when looking for LED replacements for fluorescent bulbs:

https://www.visosystems.com/tutorials/

Find the "Flicker and stroboscopic effects (23:29)" video. A PDF is here:

https://www.visosystems.com/media/TLA%20Tutorial%20EU.pdf

[tooki]

Solar cells are HUGE though compared to photodiodes, as far as I know, it's the large capacitance that limits their bandwidth.

Given a photodiode is just a smaller area of the same, you should be able to achieve the same bandwidth with a solar cell, but the load impedance would have to be scaled down in proportion to the area increase. That may or may not be low enough to be difficult in practice.

I have used a small solar cell (perhaps 3x6cm) to measure PWM frequencies, and it had no trouble clearly showing the 4+kHz of the LEDs being tested.

[rsjsouza]

Flickering annoys the heck out of me as well, even if it is imperceptible - headaches and trouble focusing on the task at hand (due to the ghosting effects) are my worst enemies.

One aspect that used to compound to this problem was the intermodulation product of the close frequency between fluorescent lights and a monitor's refresh rate - we used to tweak the frequency to 59 or 61Hz to get a clearer image than at 60Hz. 

[james_s]

Lately I mostly notice it with LED Christmas lights. They're everywhere, and most of them are dim and flicker, the handful of houses still using incandescent holiday lights always look so warm and cheery in comparison. It's a shame the LED strings are such crap still, LED lights for general illumination have gotten really good. Some of them are still crap and flicker but at least you can easily buy good ones that don't.