The move from incandescent filament lamps to fluorescent, and then LED lighting over the last couple of decades has delivered immense benefits in terms of energy saving, but had brought with it problems for people sensitive to flicker or to too much of a particular set of wavelengths. It’s not always easy to quantify the propensity of a particular light for flickering. So [kk99] has produced an instrument returning a visual indication of its quality.
At its heart is an M5Stick ESP32 development platform, and a TSL250R light sensor hooked up to one of the ESP’s internal ADCs. The flicker waveform is displayed on the screen as a simple oscillograph, and a Fourier transform is performed to extract its frequency. The result is an extremely accessible and compact instrument, showing the suitability of the M5Stick form factor for such designs. So far we’ve only brought you an M5Stick in a password keeper, but we look forward to seeing more projects featuring it.
You can see the light flicker meter in action in the video below the break.
Noticing this a bit more with cheaper LED bulbs. The easy way to tell, though not really measure is to just waggle a pen or pencil back and forth, holding one end between your fingers and if you see 7 or 8 pencils instead of a blur, then yeah, you’ve got flicker.
Not a bad idea but depending on the light bulbs you can it’s hard on the eyes.
Or just look through camera of mobile phone. You will see easily, my guess is camera trying to adapt.
I was really caught out by LED flicker while testing a light dependent resistor. My first wrong assumptions were that the LDR did not have a fast enough response time to be producing a 100 Hz waveform, as most overviews of LDRs will mention response time of even 10″s of milliseconds. My second mistake was to discount mains powered lights because mains is at 50 Hz, right?! Well yes, but if the LED bulb has half its diodes installed one polarity and half the other polarity, then the flash happens twice per AC period. It was not fun at the time but changing the sampling rate to longer periods showed the same waveform but without meeting the nyquist requirement for the actual 100 Hz signal. It gave an intuitive way to understand how slower sampling can reveal faster signals, a bit like the time domain reflectometry stuff Ted Yapo has been working on, iirc. Just to be clear, I held tight to my erroneous thoughts until I happened to move the sensor near to the window and observed the amplitude of the oscillation almost reduce to nothing. That was the clue that set me off to review my previous assumptions.
LEDs don’t like reverse voltage, so the lamps don’t have half the diodes in the other way. They instead have a rectifying bridge. Result is same i.e. 100Hz (or 120Hz) flicker
Thanks for the clarification, another assumption busted!
from my experience lack of proper(or even any) filtering after rectifier bridge is a problem here.
Once I tried to use cheap spotlights to make ilumination for highspeed video and this is what i saw after filming it at 500fps:
https://cdn.hackaday.io/images/original/4575191569871971683.gif
This is how they look when teared apart:
https://cdn.hackaday.io/images/9271801569872778186.jpg
Obviously there’s any output filtering on output of rectifier, so I decided to give a try to put some big filtering capacitors(~500uF in this case) on output of rectifier bridge:
https://cdn.hackaday.io/images/6686761569873155123.jpg
and and result was:
https://cdn.hackaday.io/images/original/7529151569873758827.gif
I guess there’s mainly price factor why they do not put proper filtering on output as anyway you’ll not see such effect with naked eye, so no one is bothering about it.
This flickering is not visible for our perception, but I wonder can it have some affect affect on our health?
Yes. It can make spinning objects like circular saw blades or objects fixed to a lathe appear stationary. A workshop light like that is a safety hazard.
Yes they do. In most lights though the diode string isn’t used in a blocking configuration because they’re using a switching mode regulator to limit the current and reversing the diodes would cause the blocking voltage to rise to the breakdown region.
In the cheapest lights where there’s just a bunch of diodes in a string matched to the the line voltage, the diodes can be reversed. There’s a phosphorous coating that glows in the green-yellow spectrum, sometimes red if it’s a “warm white” diode, while the blue light is blinking at the line frequency.
To elaborate: a typical blue diode runs between 3-4 Volts forward biased, and 5-6 volts reversed before it goes into zener mode, so, you may have let’s say 30 of them in a string with a limiting resistor for up to 120 Volts. When the string is reversed, it tolerates up to 170-180 Volts.
Of course if you did build a bulb like that, it would blink really badly (hence the phosphor coatings), it would produce terrible EMI, and it would be destroyed quickly by transients in the electric grid, but you could sell them for $2.50 at the supermarket…
And there you have the first generation of “filament” type LED bulbs.
Really nice idea and project, I’m pretty sensitive to flicker. I usually use super slow Mo video to check the lights.
Nice project.
I am also sensitive to flicker, especially if the on/off intensity difference is high, it can make me queasy upto over 500Hz.
I usually use my phone camera to check bulbs – just hold it right up to the light and the rolling shutter will show me if there is low frequency flicker.
I also use a cheap digital tachometer (with the laser disabled) to determine the frequency (just divide RPM by 60 to get Hz).
I also fund this site (in Swedish) with measurements (flicker, colour temperature/rendering, spectrum, current) of lots of LED bulbs (note that IKEA changes the electronics in their bulbs without changing their names, thats why he only list their numbers as they are unique):
https://sle.se/led/
Any replacement for the TSL250R? They appear to be out of production.
You could use TSL257 or any similar light to voltage converter.
Yep, try an LDR – the two different types I have were both responsive enough to capture the 100Hz mains flicker of my LED bulbs when part of a voltage divider and fed into an ADC.
Flickering is not the only issue. The color spectrum, measured as color rendering index (CRI), not to be confused with the specified color temperature, is also a huge issue because an LED’s color distribution is so spiky/scant compared to incadescant and fluorescent lights. IMO all LED bulbs should be required to have a chart of their color distribution on the package.
Fluorescents were much worse still some time back. These days you seem to be able to find some quite good ones. LEDs too, of course.
Tried something like this a little while ago.
Used an LED as a photodiode, connected to my scope input.
Wish I hadn’t. The led lighting in my house is disgusting on the scope.
Though the waveform was a lot nicer when the led was outside pointed at the sun.
I’m relative sensitive to flicker, older TV’s (Tube-type) more modern Plasma-Flat-TV’s and especially the Beamer’s with DLP and rotating Filter were always very unpleasant.
These day’s the flickering rear lights from cars and some traffic lights are more worse than that! :(