Reading smartphone reviews, you might notice such a phenomenon as AMOLED screen flicker or PWM (Pulse-Width Modulation). Some reviewers cite the flicker frequency and report whether or not this screen is dangerous to the eyes.
You can also find many questions about the eyestrain due to the AMOLED screen flicker. And a lot of people suffer from this issue.
After the first iPhone with an OLED display was released, many similar complaints appeared on the Internet: headaches, eye strain, etc. And it does make sense.
After all, sunlight and the light from an IPS display or eBook reader with a backlight are mostly emitted in continuous flow. At the same time, PWM technology endlessly flips the switch like a child, turning the AMOLED screen on and off hundreds of times per second!
Indeed, there are displays with PWM brightness control operating at low frequencies and large modulation depth (more on this later).
But does it affect eyesight or mental health in any way? What are the risks of flicker (PWM), and why do most AMOLED screens flicker? What exactly happens when you look at AMOLED display, and why do only a few people feel its harmful effect? Is it harmful at all?
Often, the Internet talks about it in broad strokes, but this topic requires a serious study and explanation. Actually, that’s what this article is about.
Just one more thing. If you understand well why OLED screens flicker and want to know how it affects your health, go straight to part two.
Part 1. What is PWM, and why do AMOLED screens flicker and IPS don’t?
So what’s the fundamental difference between the light emission of IPS and AMOLED screens? After all, LEDs are used everywhere today, including both of these screen types. Even e-ink readers sometimes use LED backlighting.
As you know, IPS screen does not emit light directly, and if there were no light source underneath it, we would not see an image.
All modern smartphones with IPS displays have light-emitting diodes (LEDs) as a backlight. These diodes shine “white” light, and each pixel is covered by a color filter – a glass one of the three primary colors: red, green and blue.
When the white light from LED passes through the green glass (green color filter), the dot on the screen appears green, and when it passes through the red one, we see a red pixel:
To display an orange dot on the screen, we need to allow the maximum amount of white light through the red glass, almost half as much light through the green glass and just a little bit of light through the blue filter.
Since these individual dots are very close to each other, we will not be able to distinguish individual colors and they will be merged into a single orange color on the retina:
If we just want to reduce the screen’s overall brightness, we reduce the amount of light emitted by the whole backlight.
And the main question is how exactly this brightness is reduced. Can we just drive less current to the LED? Sure! You drop the voltage, and the LED gets dimmer.
In fact, this is how brightness control works on the vast majority of smartphones with an IPS screen. But OLED displays work differently.
OLED doesn’t need color filters or a white backlight. Each pixel here is a separate organic LED, which emits one of the three primary colors on its own. Thus, if we want to display the orange dot, we need to turn the brightness of the red LED up to the max, slightly decrease the brightness of the green LED, and turn off the blue one:
And that’s where we are in for a surprise. It’s not possible to reduce current flow through the organic diode without consequences, as we did in the case of the IPS screen.
Unfortunately, the organic LED changes its hue depending on the current supplied.
Take a green LED, for example. When you drive a LED with 20 mA current, it starts to emit light with a wavelength of about 525 nanometers (remember that the color perceived by the eye depends only on the wavelength). But if we reduce brightness to 20% (5 times), we get a different shade of green with a wavelength of ~532 nm1:
So how can we achieve accurate color reproduction with OLED? If we can’t change the brightness of the LED by adjusting the voltage or current, we have to come up with another way.
And the solution turned out to be very simple! We just need to supply a current of 20 mA, no more and no less. In this case, all the organic diodes will work at full brightness and not change their shade. But we have to “chop it up” into discrete parts. Say, we can divide one second by 100 and get 10 milliseconds – this will be the duration of each “part”.
During this time (10 ms), we will drive the LED with a 20 mA current to work at full brightness. Then we turn off the LED and wait for the next 10 ms. During this time, LED emits no light at all. Then we turn the LED to the max (20 mA) again for the next 10 ms, and so on.
If we want the red LED to be as bright as possible, we won’t turn it off at all during each 10-ms cycle. It would be something like this:
And if we want to reduce the brightness of the LED down to 20% (5 times), we just need to reduce the time the LED emits light (at max brightness) during each cycle by a factor of 5:
So the red LED turns on at max brightness (20 mA current) and stays on for 2 milliseconds, then turns off and stays off for 8 ms. Once the 10ms cycle is over, we turn the LED back on for 2 ms and off for 8 ms. In other words, the LED emits light for only 20% of the total time.
This trick is possible due to an interesting phenomenon known as critical flicker frequency (CFF). The human eye is able to notice individual light flickers if they occur at a frequency below some value, that is, when the light turns on and off less than, say, 100 times per second2.
But when the on/off frequency exceeds that number, we don’t feel flicker. The screen seems to emit continuous and steady light, only dimmer.
For example, the iPhone 13 Pro Max screen has a resolution of 1170 x 2532 pixels, which is more than 7 million individual organic LEDs! And a microprocessor controls pixels brightness by regulating each on/off cycle’s time and duration.
Each of these LEDs must support hundreds or even thousands of brightness levels (controlled by pulsation) to display all sorts of colors and shades.
It’s more than just the frequency
As logic dictates, frequency is not the only parameter that affects the flicker level. It also matters how long the LED stays on during one cycle.
There will be no flicker at all at full brightness, even if the frequency is low. After all, in this case, the LED is on 100% of the cycle time. However, the LED may only be on for 10% of the cycle time at low brightness. It turns out that most of the time the screen is off, and separate flashes of light are becoming more noticeable:
The manufacturer can also combine brightness control by modulation (flicker) and reducing the current. For example, the brightness is dimmed by the current to a value at which color distortion could be noticeable to the naked eye. Then modulation (PWM) is turned on, i.e. the display starts to turn on and off quickly.
The difference between the max and min LED luminance during on/off cycles, known as modulation depth, plays an important role also. In some cases (such as LED backlighting for IPS screens using phosphors), the brightness may not go down to zero during off time due to the phosphor afterglow.
If we don’t want to consider all these variables (flicker frequency, on/off-state duration, min brightness during off state), we can use a single metric that represents the flicker level. This metric can be calculated using a very simple formula3:
100 * (Emax – Emin) / 2 * Emean
Where Еmax is the max brightness of the LED (or the whole screen), Еmin is the minimum brightness (when LEDs are off), Еmean is the average brightness value during one cycle.
Now let’s calculate the flicker level for a screen with such parameters:
- PWM frequency is 100 Hz, i.e., we have 100 on/off cycles per second. So each cycle lasts 10 ms since there is 1000 ms in one second.
- Max brightness is 400 nits (Emax).
- Min brightness is 0 nits (Emin).
- We set smartphone screen brightness at 50%, i.e., LED emits light for 5 ms (50% of the cycle) and then off for another 5 ms.
- So the average screen brightness during one cycle is (400 + 0) / 2 or 200 nits (Emean).
Here’s what we get in this case:
100 * (400 – 0) / 2 * 200 = 100%
This example shows a flicker level of 100%. It means that the screen is simply off half of the time. Such flicker is considered very high and dangerous. But if the flicker level is 0%, then there is no flicker at all, i.e., the screen is emitting light continuously. Almost all smartphones with IPS displays and many eBook readers are flicker-free.
In many cases, this value can easily exceed 100%. For instance, if LEDs turn on for 2 ms and then turn off for 8 ms during a 10-ms cycle, the flicker level is 500%.
Well, this knowledge is enough to get to the heart of the problem.
Part 2. Is AMOLED screen flicker (PWM) harmful, and why?
Many people believe the flickering of AMOLED screens is nothing more than just a mental problem. Like, people with a high level of anxiety read some scary stories on the Internet and then see flickering everywhere.
Actually, this misconception is due to the fact that one does not understand the difference between conscious and unconscious perception. And there are a lot of things that the body perceives, but we do not consciously feel it.
Almost everyone notices flicker if its frequency is not very high (on average, no higher than 60-70 Hz). Such perceptible flicker causes discomfort and, in rare cases, photosensitive epilepsy. Thousands of brain neurons instantly begin to activate in sync with a specific frequency, resulting in an epileptic seizure4. You can read in detail about neuronal syncing in our article on sleep tracking with smartwatches.
Anyway, AMOLED screens operate at frequencies 5 times the flicker fusion threshold (or critical flicker frequency), so it must be about something else.
It’s not about pupil!
An interesting myth on the Internet states that the human pupil expands and contracts to the beat of the AMOLED screen flicker. Such unfounded statements can be found even on the official website of the monitor manufacturer ViewSonic and the DxOMark test lab.
And it is this lightning-fast ongoing pupil size fluctuation leads to rapid eye strain and fatigue, according to ViewSonic and DxOMark.
This is an interesting theory, but the human pupil actually reacts to light changes with a relatively long delay (at least 180 milliseconds5 or 6 Hz frequency)
But we are talking about frequencies 30-50 times higher! That is, the pupil should react to the light change within one millisecond, which is, obviously, impossible, taking into account the 180 ms delay.
One study6 even measured the pupil response to a light flickering. This is what it looked like when flickering at a frequency of 0.7 Hz (the light turns on and off every second and a half):
The vertical lines in the graph show when the screen turns on (solid line) and turns off (dashed line). We can see how the pupil contracts and expands as the light flickers (pupil size is shown in relative units, not millimeters).
And here is what happens when the screen flickers at a frequency of 1 Hz:
We can see that the pupil’s reaction does not precisely correspond to the moments of switching the screen on and off (there is a slight delay). Nevertheless, the pupil size fluctuations are clearly noticeable.
Pupil size fluctuations in the beat of light flickering are clearly observed up to 2.3 Hz, with barely perceptible up to 3 Hz:
Obviously, the pupil can’t react to the flicker frequency of the screen at 240 Hz, which is 70 times the threshold of barely detectable fluctuations. Apparently, it is about something else again.
Important note: the pupil’s diameter is indeed related to the discomfort of reading from a screen in the dark. But it doesn’t relate to flicker. Depending on the polarity of the contrast (black text on a white background or white text on black background), more light can enter the eye than needed, which will cause pain.
We found nothing…
If we read the latest official report7 of the Scientific Committee on Health, Environmental and Emerging Risks (dated June 5, 2018), we see a very interesting conclusion regarding flickering light at high frequency:
Although no published case-studies were identified, there are claims that a small number of people are very sensitive to temporal light modulation at about 100 Hz, triggering symptoms such as headaches, migraine and general malaise.European Commission, SCHEER, 5-6 June, 2018
And from the same report:
It is possible that some of the susceptibility to high frequency (100 Hz and above) temporal light modulation may be due to the phantom array, even if the array is not perceived.European Commission, SCHEER, 5-6 June, 2018
It seems ironic that not a single study has been found on the problem the Internet has been talking about for years. Nevertheless, the European Commission is being a little deceitful, hinting at phantom arrays.
At the end of the last century, experiments8 proved that a person can distinguish individual flickers of light at a frequency of 200 Hz, that is, when PWM operates in cycles of 5 milliseconds (LED lights up for 1 ms every 5 ms).
The electroretinogram clearly demonstrates9 that the retina registers light modulation at a frequency of up to 200 Hz and processes this signal. Although it exceeds many times the limit we consciously perceive (60-70 Hz).
In other words, there is such a thing as an invisible but affecting high-frequency flicker of light in medicine.
Compared to the camera sensor human retina has a very low resolution, especially if we talk about color vision. But the “picture” we see looks crisp thanks to the fast unconscious coordinated eye movements called saccades.
Using these saccades, the brain constantly scans the image to project it onto a tiny central area of the retina with high resolution (only the center of the retina has a high density of cones).
You’ve probably heard of the so-called “pencil test” when you wave a pencil around very quickly in front of a light source (be it a smartphone screen or light bulb). If the pencil trace seems blurred, it means that the light is not flickering. But if we see pencil outlines (or a lot of separate pencils), it means that PWM regulates the backlight, i.e. the light is flickering:
The same happens when an object is stationary, and the eyes make saccades, scanning the surroundings in rapid jumps (for example, when reading text from an AMOLED screen). As a result, separate clear copies of the same image are built on the retina with a slight shift instead of the required motion blurring.
Not for the faint of heart
Take your eyes off the article for a second, look straight ahead, and then look away. Did anything seem strange to you?
Chances are, you didn’t notice any weirdness. But if you start recording video on your smartphone and quickly move the camera sideways, the result will look different.
The recording would show all the objects as you move the camera become blurry and fly away:
But why doesn’t this happen to the eyesight?
Frankly, the reason for this is somewhat creepy. At the moment of gaze shifting when your eyes make saccades, you literally go blind for a fraction of a second. This is due to saccadic masking10, a special biological mechanism that turns off retinal image processing when it becomes completely blurred.
As a result, it looks to you like all objects didn’t move, and the whole scene is sharp. But in reality, the brain is seamlessly stitching two scenes, the one before the eye movement and the other after the saccade ends. Everything in between is simply cut out.
If you don’t believe me, then go to the mirror, open your eyes wide, and start looking from one eye to another. You will be surprised, but no matter how hard you try, your eyes will be motionless all the time, as if your gaze is frozen at one point.
Your eyes will actually move, but you will go blind before each movement. Once the eye movement is over, you’ll get your sight again.
I hope you understand now how ridiculous it is to stick to the “if I don’t feel something, it doesn’t exist” approach having such a brain. Your brain brings only a fraction of the signals it receives to your consciousness.
The phantom array
When a flickering light bulb illuminates the scene or when we look at a flickering AMOLED screen during saccades, the sharp image occurs on the retina instead of a blurry trail (see the pencil test above). This phenomenon is called a phantom array.
And this is where the fun part begins, especially if you are looking at the flickering screen in complete darkness. In this case all you see is flickering light, and your brain stands no chance to mask those sharp images that appear during the saccades. Similar effects can occur if light flicker with a frequency of up to 2000 Hz11.
But, again, it depends directly on the flickering intensity. The screen flickering will not cause any biological reactions at a certain level. And this level is easy to calculate11:
- No harm: Flicker Frequency * 0.0333
- Low-risk: Flicker Frequency * 0.08
For example, if the iPhone 12 mini screen is PWM at 250 Hz, flicker intensity up to 8% (250*0.0333) will have no biological effect. And a flicker up to 20% (250*0.08) may pose little risk.
Here is what the actual flicker intensity looks like on the iPhone 12 mini as a function of brightness:
|Screen Brightness||Flicker level|
Here we see that at brightness above 75%, the screen has no effect on health, but the impact is quite significant at low brightness. What’s worse, many smartphones (even the most expensive ones) have flicker intensity up to 150% at 50% brightness.
Getting back to eye saccades and phantom array, I would like to show the results of an interesting study that analyzed the effect of screen flicker frequency on eye movement. Specifically, researchers compared reading from a screen flickering at 100 Hz to reading under flicker-free lighting.
Here are the findings of that study12:
The results are consistent with the view that flicker has two distinct effects on reading, both of which are potentially disruptive. The first relates to an increase in the number of prematurely triggered saccades, which are, as a result, less accurate. The second is an increase in the number of saccades perturbed in flight, which land short of their intended target.PMID: 2017572
Such disruptions cause the brain to make more corrective saccades, which apparently causes additional strain and fatigue.
Anyway, our body can perceive light flickering at a high frequency. Yet, science doesn’t know why only a few people feel this effect. Or maybe it just doesn’t want to know.
Can high-frequency flicker damage eyesight? That’s not the point since the problem of high-frequency flicker is not related to eye disease. Neither observation nor theory supports such causation.
However, flickering light can cause eye fatigue or headache, which some researchers attribute to saccadic eye movements and phantom images on the retina that need to be processed. Also, the brain continuously reacts to the light signal itself.
I’ve not found any information on whether high-frequency flickering synchronizes the brain’s electrical activity. Still, low-frequency light flickering (in particular, at 40 Hz) induces the gamma brain rhythm at frequencies 38-42 Hz, which you can see in the electroencephalogram13.
Let me remind you gamma wave (frequency ranges from 25 to 140 Hz) activity is most prominent during alert, attentive wakefulness14.
No one has any data on whether your memory will be impaired or if you will become short-tempered and more irritable after five years of actively using flickering AMOLED screens. But it’s better to follow recommendations and not to expose yourself to any light source with a high-intensity flicker for a long time.
Alex Salo, Tech Longreads founder
If you enjoyed this post, please share it! Also, consider to support Tech Longreads on Patreon and get bonus materials!
- Texas Instruments Incorporated [source]
- Flicker fusion threshold [source]
- GOST standard R 54945-20124
- IEEE standard PAR1789
- Eyeing up the Future of the Pupillary Light Reflex in Neurodiagnostics [PMID: 29534018]
- Tracking the allocation of attention using human pupillary oscillations [PMID: 24368904]
- Potential risks to human health of LEDs Final Opinion, SCHEER [source]
- The Phantom Array: A Perisaccadic Illusion of Visual Direction [DOI]
- Recommending practices for modulating current in High Brightness LEDs for mitigating health risks to viewers [source]
- Saccadic masking [source]
- Designing to Mitigate Effects of Flicker in LED Lighting: Reducing risks to health and safety [DOI]
- The effects of flicker on eye movement control [PMID: 2017572]
- Gamma Band Light Stimulation in Human Case Studies: Groundwork for Potential Alzheimer’s Disease Treatment [PMID: 31156180]
- Gamma wave [source]
This is a very well researched and written post. With so much anecdotal information out there, it’s so rare to see an article with solid citations and rooted in strong scientific principles. Personally, I’m still undecided on the matter of PWM, and would like to see more studies on this in the future. But I am sure that your article will help point some budding researcher in the right directions!
Thank you for your kind words! I appreciate it!