LED Lights

Lamp flicker: Hidden in plain sight

Most people are probably not aware that when they switch on almost any lamp, they are witnessing an optical illusion. They think that the lamp light is of constant brightness, even though it is not. The lamp is actually continually cycling between bright and dim, so rapidly that they think that they can't see it, even though their eyes do see it.

When a light "flickers" it rapidly switches between bright and dim light. Sometimes flicker can be obviously visible, like that of a malfunctioning fluorescent or LED light that switches on and off a couple times a second. However, the normal light flicker of incandescent, fluorescent, and most LED lights is hidden because it happens so rapidly that we don't know that we see it. Sometimes flickering light makes people sick, causing seizures, headaches, eye strain, nausea, spatial disorientation, and other neurological symptoms (see Background: Health Effects of LED Lights and Screens).

LED lights present both a risk to human health and an opportunity for better health. Some LED lights flicker more seriously or in a different way than incandescent or fluorescent lights, creating the potential to harm the health of sensitive individuals. However, LED lights can instead be engineered to be completely flicker-free, creating light that doesn't harm sensitive individuals and that is potentially better for human health than incandescent or fluorescent light.

While glare from LED headlights or streetlights is also a significant public safety concern due to loss of visibility on roads, a problem beginning to be recognized even by the lighting industry (see a Cree whitepaper), this website will focus on the potential health effects resulting from light flicker. 

LED Light Bulbs

What makes some lights flicker?

LED lights can be engineered to either not flicker at all or to have varying degrees of flicker. The flicker is so fast that it's difficult to impossible for people to know that they see the individual flashes of light. 

Unless specifically engineered not to do so, any light bulbs, including incandescent lights, flicker to some extent. This is due to the 60 Hz AC electricity supply in North America or 50 Hz in Europe and other parts of the world. Alternating current (AC) means that electrons alternately flow in one direction through a circuit and then switch to flow in the opposite direction. A bulb lights each time the electrons flow through the circuit, regardless of the direction. Since 60 Hz AC means that the whole forwards/backwards electron cycle repeats 60 times a second, the light bulb flickers cyclically from bright to dim 120 times a second (120 Hz; 100 Hz in Europe). In contrast, If is a bulb is powered by direct current, such as in a simple circuit powered by a battery (like in a battery-powered incandescent flash light), the bulb doesn't flicker at all - it will be constantly on at the same brightness as long as the battery is on.

The flicker of incandescent lights tends not to bother people, perhaps because when the bulb dims, the recently-burning bulb filament dims slowly enough that it may only dim a little before it flashes back on again. An average incandescent light bulb has about 6.6% flicker (IEEE std 1789, 2015, Figure 2). My own experience and widespread anecdotal reports on the LEDStrain.org forum suggest that sensitive people can tolerate incandescent flicker much better than LED flicker that has similar brightness flicker. Please see the discussion of how a possible explanation for this might be the greater color flicker of many LEDs compared to no color flicker of incandescents in relation to my 2023 measurements of flicker (see Testing LEDs and Screens).

The flicker of fluorescent lamps varies, depending on their engineering. Older fluorescent lights with magnetic ballasts have about 40% flicker (Poplawski and Miller, 2011), largely at 120 Hz / 100 Hz, but may also produce 60 Hz / 50 Hz flicker, especially as they age (Brundrett, 1974). Decades ago, there were health complaints about the flicker of magnetic ballast fluorescent lights and newer electronic ballast fluorescent lights are engineered differently so that they produce less flicker. Electronic ballasts introduce >20,000 Hz flicker, which reduces, but does not eliminate, the magnitude of the 120 Hz flicker caused by the AC power. The residual 120 Hz flicker tends to be similar to that of incandescent lights (Poplawski and Miller, 2011).

In contrast to incandescent or fluorescent lights, if they are not deliberately engineered to do otherwise, LEDs will turn completely off during their dim phase, giving them 100% flicker (IEEE std 1789, 2015, Figure 3). 

When the current supplied to an LED stops, the light immediately shuts off, creating 100% flicker, as the bulb is alternately on and completely off. Some of the LED bulbs sold at least in the US have 100% flicker. However, circuit engineering strategies exist for reducing LED flicker or for making LEDs essentially flicker-free (IEEE std 1789, 2015). 

Given the known adverse health effects of flicker from magnetically ballasted fluorescent lights in the 1980s and early 1990s (reviewed in IEEE std 1789, 2015 and in Background: Health Effects of LED Lights and Screens), LED flicker might produce even higher risks. However, completely flicker-free LED lights might be even healthier than prior artificial lights.

The figure below shows graphs of light output vs. time which display the degree of 120 Hz flicker of several kinds of light bulbs. Each of the graphs has the same average light output, meaning that each of the lights would appear to be the same brightness. The nervous system tends to perceive the average light output when flicker is fast. The Talbot-Plateau Law is the concept that the apparent brightness of intermittent light is proportional to the fraction of time that the light is observed (proposed by William Henry Fox Talbot in 1834 as a result of experiments where a viewer's gaze is fixed on a single point on a rotating disk with white and black sectors or on the intermittent reflection of sunlight through a slit; refined further mathematically in quantitative experiments by Joseph Plateau in 1835 using an observer's fixed gaze on the edge of a rotating white- and black-sectored disk). 

Incandescent lights always have a waveform shaped like a sine wave (a) due to the slow brightening and slow dimming of the burning bulb filament. Notice that the shape of the electronic ballast fluorescent waveform (b) has a different shape. Depending on the circuit engineering, an LED light bulb could have any of several different waveform shapes, including (d) sine wave and (e) square wave. There are many other LED waveform shapes as well, some of which are shown in Poplawski and Miller, 2011. Importantly, LED bulbs that have been engineered to be completely  flicker-free (c) have constant light output, a completely flat line in the graph, corresponding to 0% flicker.

Flicker percent = 100% x (max-min)/(max + min) where max and min refer to the maximum and minimum light output, respectively. Notice that the flicker percent does not indicate the shape of the waveform. For example, (e) and (f) both have 100% flicker, but have different waveform shapes.

Flicker of incandescent, fluorescent (CFL) and LED light bulbs and constant light output of completely flicker-free LED light bulbs. The above figure was drawn by J. Hackett and may be reproduced for any purpose with appropriate citation of this website as the source.

Industrial LED Lighting

Industrial-style commercial LED lights are now being used for new classroom, heathcare, or other workplace lighting. Often, these are long LED strips with individual LEDs spaced about every third of an inch. The strip is housed within a long rectangular fixture with a diffuser panel over the lights. Other kinds of industrial lighting might be 2-dimensional grids of LEDs, often behind diffuser panels.

Between the wall light switch and the lamp fixture, is a driver (also called a power supply or ballast) that usually includes a dimming function (reviewed in IEEE std 1789, 2015). The power supply controls the direct current (DC) output to the LED strip - all LEDs use DC. Depending on how it is engineered, a driver may or may not eliminate all of the 120 Hz flicker from the AC supply. If it outputs constant DC, without any residual 120 Hz ripple, the LEDs will be flicker-free. If it outputs pulsing DC, the LEDs will flicker. Many LEDs use pulsing DC at full brightness, usually at about 500 Hz, which has the advantage of increasing the life of the LEDs compared to constant DC.

In addition, drivers that are also dimmers have a strategy for reducing the total light output of the lamp fixture during dimming. Theoretically, a driver could achieve this in a completely flicker-free way by always outputting constant DC current at any given dimming setting, but altering what that constant DC current is if the user changes the dimming setting. However, this constant current strategy has the disadvantages of reducing the longevity of the LEDs and resulting in altered color temperature when dimming the lights.

In a different strategy, a driver can achieve dimming by a strategy called pulse-width modification (PWM) which lengthens the “off” time of the light as it flickers so that the light appears to be dimmer, even though it is still fully bright when it is on. Dimming can make flicker more obvious to the nervous system. Some drivers use a combination of reducing the current and PWM to achieve dimming. Some switch between the two dimming strategies depending on how much the light is dimmed. Just because a driver may reduce the current to achieve dimming doesn’t mean that the LEDs are flicker-free at full brightness – often they are programmed to have flicker.

All of the possibilities for how drivers may work can be confusing. "PWM" refers only to a dimming strategy; it does not indicate whether or not there is flicker at full brightness. The graphs and table below summarize the possibilities. Flicker-free light at full brightness is shown in (a) and (c). Square waves at full brightness are shown in (b) and (d). LED waveform shapes other than the square waves shown in the graphs are also possible. Constant current reduction dimming is shown in (a) and (b) while PWM dimming is shown in (c) and (d). Notice how PWM dimming does not decrease the maximum light brightness, it increases the time that the light is off, thus making the light's flicker more obvious to the nervous system as the light is dimmed.

Drivers for LED strip lighting can produce completely flicker-free light (a) or can introduce flicker by always pulsing the direct current from the driver at any brightness (b and d) and/or by increasing the amount of time the light is off when dimming via pulsed width modification (c and d). The above figure was drawn by J. Hackett and may be reproduced for any purpose with appropriate citation of this website as the source.

Theoretically, a driver that creates flicker at full brightness and/or that uses PWM dimming could potentially be safe for humans  if the flicker is too fast for the nervous system to detect and/or respond to in an unhealthy way. It isn’t clear how fast that needs to be since no scientific study has identified a safe level of flicker for the population as a whole, including the segment of the population whose health is most affected by flicker. 

There can be another source of confusion for average consumers when reading product literature for LED drivers because the mechanism by which the light switch communicates with the driver can also be either by reducing a current or through a PWM signal, and that mechanism is unrelated to the mechanism by which current is delivered from the driver to the LEDs. The overlapping language and non-specificity in product literature make it incredibly difficult for a non-engineer like me to try to sort out how any particular driver works based only on basic online descriptions.

 

Note that the possibilities shown in the above figure are also the possibilities for how LED backlighting might function for LCD screens. These screens have a grid of white LEDs behind the LCD layer of the screen. OLED screens use PWM of individual red, green, and blue LED pixels (see Background: LED Screens).

 

How is color controlled in LED lights?

White light is a mixture of multiple colors of light. One way to achieve this is to have blue or violet LED light excite a phosphor coating inside the covering of LED bulbs. The phosphors in turn emit other colors of light of lower energy, creating a fairly full spectrum of light overall. An alternate strategy is for a mixture of red, green and blue LEDs to combine to appear white. Some LED fixtures use a mixture of both strategies. The US Department of Energy summarizes these strategies for creating white LED light. The phosphors in LED bulbs each have unique rates for absorbing energy and for emitting light. The colors produced by LEDs can be slightly to greatly out of sync with each other due to differences in the timing of light production from phosphors compared to each other and compared to the direct production of any visible light from the diode. Many LED lights have significant color-to-color flicker, while incandescent lights do not have any color-to-color flicker (Testing LEDs and Screens).

Some LEDs achieve various colors of the rainbow by combining red, green and blue LEDs in variable ratios. Similar to these full color-tunable light fixtures, other "white-tunable" LED fixtures allow for the tuning of the shade of white light by altering the ratio of warmer white and cooler white LEDs in the fixture. The US Department of Energy also reports on color-tunable LEDs, describing that they work by allowing flexible control of the relative dimming of multiple colors of LED lights. While it is theoretically possible to control ratios of light output from the 2 or 3 kinds of LEDs in a white-tunable or color-tunable light fixture using constant current dimming, it is most likely that tunable light fixtures will use flicker to alter the amount of time that each of the 2 or 3 colors is on. This not only increases brightness flicker, but also increases color-to color flicker. The use of flicker instead of constant-current dimming is particularly likely in this scenario because constant-current dimming of LEDs also alters the color temperature of the individual LEDs (see below). PWM, which is a kind of flickering, is a common mechanism used to dim LED lights (see explanation of PWM and other dimming strategies above).

Are there limits on LED Light Flicker?

Metrics for quantifying the amount of flicker

Flicker metrics: Methods for numerically describing the light flicker waveform. The IEEE report (IEEE std 1789, 2015) describes flicker frequency, flicker percent, and flicker index and discusses the rationale for initial recommendations for limiting flicker in LED lights. The flicker portion of Pacific Northwest National Laboratory Naomi Miller’s webinar on flicker and glare (after minute 30) provides an update on more recently-developed flicker metrics and a technical review of LED flicker (Miller, 2019). Also see CIE TN-006, and a US Department of Energy graph of recommended limits on flicker. 

Synonyms for flicker percent: modulation depth, % modulation, PF


No Observable Effect Level (NOEL): Proposed limit on LED flicker in IEEE std 1789 (2015) that combines flicker percent and frequency. Below 90 Hz, the NOEL recommendation is that flicker percent should be less than 0.01 x frequency. Above 90 Hz, the NOEL recommendation is that flicker percent should be less than 0.0333 x frequency. See Background: Heath Effects of ≥100 Hz flicker for an analysis of the basis for the NOEL recommendation.



None of the above metrics consider color-to-color flicker.

Testing LED light flicker


LEDbenchmark.com describes flicker metrics and how flicker is tested.

These websites provide independent evaluations of LED bulb flicker:

Evaluated many LED bulbs between 2012 and 2015 in terms of flicker and several other characteristics, but stopped due to a lack of funding. Based in Australia and the United States. (See company information). 

Electrical Engineer Peter Erwin tests LED bulbs for flicker and has developed a novel flicker metric. Based in Germany. (Learn more about Der Lichtpeter).


A flicker meter or oscilloscope is needed to quantify flicker. This US Department of Energy website devoted to lighting flicker includes multiple publications and videos addressing instruments for assessing flicker. Currently-available flicker meters only detect changes in light brightness over time and cannot detect how the color of the light might flicker.


Without specialized equipment, if the flicker is severe, such as 100% flicker at 120 Hz, the flicker can be detected by observing a fidget spinner. When spun, the stroboscopic effect makes the blade pattern appears to move alternately both forwards and backwards as the flickering light intermittently illuminates the blades. The direction of spin of the pattern depends on how the rotational speed interacts with the flicker frequency. A fidget spinner is insufficient to detect most less dramatic flicker (personal observation). 


A high speed camera can also be used to detect flicker. Flicker may appear as alternating brightness in each frame or as lines across the image as the flicker interacts with the camera's panning capture of the image. Even a smartphone slow motion video that records 240 frames per second (fps) is good enough to detect more subtle flicker than a fidget spinner. It can be used to estimate the flicker frequency (see Testing LEDs and Screens). A smartphone 240 fps video is insufficient to detect flicker with fairly low percent flicker and/or with high frequency, but people may still experience health effects (see Testing LEDs and Screens). 


Early recommendations for limiting LED flicker

The 2015 IEEE std 1789 recommendations (see IEEE figure 18) proposed strict limits for <90 Hz flicker that might cause epileptic seizures and proposed more modest limits for >90 Hz flicker. Studies of the effects of fluorescent lights had already indicated such >90 Hz flicker could cause headaches, eyestrain, possible behavioral changes in autistic children, and difficulties in concentration (reviewed in IEEE std 1789, 2015).

 

The 2015 IEEE std 1789 recommendations for limiting flicker were only estimations based on studies of the biological effects of fluorescent light and on other studies of the biological effects of <90 Hz flicker. They were not based on any data about the biological effects of >90 Hz LED flicker. There is still very little such data today (see Background: Health Effects of LED Lights and Screens).

How much flicker do manufacturers allow in LED bulbs?

The first mass-market LED bulbs were quite variable in terms of flicker. In 2011, testing by PNNL showed that there was a variety of LED bulbs on the market, with 45% being either completely flicker-free or having very low flicker (flicker index ≤ 0.05 for 42/93), 23% having very high flicker (flicker index ≥ 0.40 for 21/93), and the rest having intermediate flicker (Poplawski and Miller, 2011). Flicker index is calculated from a graph like those shown above, by dividing the area under the curve for the portion of the graph that is above the average by the entire area under the curve. A larger flicker index means there is more flicker. Additionally, LEDbenchmark.com reviewed the flicker of many LED bulbs between 2012 and 2015 and their results can be sorted by flicker on their website. The flicker-free bulbs reviewed on this site include these nearly flicker-free Philips bulbs that I happen to own and have used without issue since 2013 

I didn’t realize until this year (2021), that major U.S.manufacturers seem not to make flicker-free LED bulbs anymore, at least not that I can find. The only flicker-free bulbs that I have found after extensive web searches and personal communication with Philips are a subset of the bulbs at Waveform Lighting. 2024 Update: in testing I found that Waveform bulbs are not actually flicker free (Testing LEDs and Screens). Their low flicker is in a pattern that I can tolerate when the bulbs are new, but they start to flicker dramatically and tend to injure me as the bulbs age. It would be interesting to see a repetition of Poplawski and Miller’s study for the bulbs available today.

Lighting companies, with the exception of Waveform, do not currently report flicker metrics for their bulbs to the public, so there isn’t a way to evaluate them before purchase. Some companies are now misleadingly marketing their bulbs as “flicker-free” on Amazon.com if the bulbs lack <90 Hz flicker, but still have >90 Hz flicker. Der LichtPeter lists many LED bulbs that are falsely marketed as being flicker-free on Amazon. I have also recently noticed other LED bulbs falsely marketed as "flicker-free" in the US on Amazon.com. When I have contacted the manufacturers, they admit that the bulbs are not flicker-free.

I have not found any way to evaluate before use whether LED bulbs will fail in a way that increases flicker or even produces visible, slow blinking on and off.

In the U.S., there are no >90 Hz flicker requirements for LED lights, with the exception of California's Title 24 requiring less than 30% flicker. Additionally, LED bulbs with <30% flicker at less than 200 Hz or any flicker percent at more than 200 Hz may have a UL certification. 

As discussed above, 2015 IEEE std 1789 provides recommendations for limiting flicker (see Background: Health Effects ≥100 Hz Flicker for graph and analysis). However, since then there has been a trend in the lighting industry to forego the IEEE recommendations and to instead assume that if people can't knowingly see the flicker then the lights must be healthy, including in this Philips Eye-Comfort white paper. In recent years, the IEEE 1789 recommendations have been superseded by much less stringent flicker metrics (Pst, that measures <80 Hz visible flicker only, and SVM, that is much less conservative than the IEEE 1789 recommendations for the 80 Hz to 2000 Hz range; see Naomi Miller’s 2019 webinar on flicker and glare).

I question whether the IEEE recommendations themselves are stringent enough for sensitive individuals - I know they are not close to stringent enough to protect me. I have personally not yet found any LED lights with flicker that do not trigger symptoms for me within a few seconds to a few minutes of exposure, including multiple bulbs with much less than 30% flicker and/or >200 Hz flicker that fall well within the IEEE "no observable effect" range (see Testing LEDs and Screens). The only LED lights that I have personally found to not trigger symptoms are those that are completely free of all flicker (on constant DC) or nearly flicker-free (on AC) with rare, unique waveforms. To my knowledge, there has not yet been any relevant scientific study of the health effects of LED flicker for sensitive individuals.

Given that the technology to make flicker-free LEDs exists, why would manufacturers choose to produce bulbs that flicker? 

Nearly flicker-free LED lights were relatively common in the past, but are now exceedingly rare, at least in the United States.

One issue is that the circuitry needed for flicker-free operation very slightly reduces the power efficiency of the bulb. A “power factor” less than one means that some of the energy entering the circuitry is lost as heat rather than being used directly by the light-emitting diode to produce light. LED bulbs need to have a power factor ≥ 0.7 to be sold in California (effective Jan. 1, 2018, California Energy Commission Title 20). This requirement also partially aligns with requirements to receive an Energy Star rating (Energy Star eligibility criteria v2.1, 2017). Nearly flicker-free LEDs have seemed to have power factors that hover around this threshold (see data at LEDbenchmark.com).

In addition to the trade-off in energy efficiency, depending on the circuitry design, limiting flicker has the theoretical potential to limit the longevity of the bulb or driver, limit the features (especially achieving warm light upon dimming or allowing tunable color), or increase the cost. Some of these concerns can be avoided or mitigated.

Many early possibilities for circuits driving LEDs, including several possible strategies for creating a constant current, are reviewed in Arias, M., Aitor, V., Javier, S. An Overview of the AC-DC and DC-DC Converters for LED Lighting Applications. Automatika‒Journal for Control, Measurement, Electronics, Computing and Communications 53 (2), 156-172 (2012). The 2015 IEEE std 1789 recommendations also describe several possibilities for circuits driving LEDs and an explanation of how to create a circuit that produces flicker-free light output. CIE TN-006, International Commission on Illumination (2016) discusses design advantages of flickering lights.

References

Arias, M., Aitor, V., Javier, S. An Overview of the AC-DC and DC-DC Converters for LED Lighting Applications. Automatika‒Journal for Control, Measurement, Electronics, Computing and Communications 53 (2), 156-172 (2012). https://www.researchgate.net/profile/Javier-Sebastian/publication/265360097_An_overview_of_the_AC-DC_and_DC-DC_converters_for_LED_lighting_applications/links/55a4e86308aef604aa040a82/An-overview-of-the-AC-DC-and-DC-DC-converters-for-LED-lighting-applications.pdf

Brundrett, G. W. Human sensitivity to flicker. Lighting Research & Technology. 6, 127-143 (1974). https://journals.sagepub.com/doi/abs/10.1177/096032717400600302?journalCode=lrtb

California Energy Commission. 2019 Title 20, Public Utilities and Energy, Division 2. State Energy Resources Conservation and Development Commission, Chapter 4. Energy Conservation, Article 4. Appliance Efficiency Regulations, section 1605.3. State Standards for Non-Federally-Regulated Appliances. (k) Lamps. https://govt.westlaw.com/calregs/Document/IEEDE2D64EF7B4F168C0E85379828A8C2?viewType=FullText&originationContext=documenttoc&transitionType=CategoryPageItem&contextData=(sc.Default)

California Energy Commission. 2016 Title 24, Part 6. Building Energy Efficiency Standards for Residential and Nonresidential Buildings. http://www.energy.ca.gov/title24/2016standards/

CREE Lighting. Is street lighting damaging our health? 2022. https://online.flippingbook.com/view/702884488

ENERGY STAR® Program Requirements Product Specification for Lamps (Light Bulbs): Eligibility Criteria Version 2.1. 2017. https://www.energystar.gov/sites/default/files/ENERGY%20STAR%20Lamps%20V2.1%20Final%20Specification_1.pdf

International Commission on Illumination. Technical note: Visual aspects of time-modulated lighting systems - definitions and measurement models. CIE TN 006:2016. CIE, 2016. http://files.cie.co.at/883_CIE_TN_006-2016.pdf

The Institute of Electrical and Electronics Engineers, Inc. IEEE Std 1789™-2015: IEEE Recommended Practices for Modulating Current in High-Brightness LEDs for Mitigating Health Risks to Viewers. 2015. http://www.bio-licht.org/02_resources/info_ieee_2015_standards-1789.pdf

Miller, Naomi. Metrics in Motion: Flicker and Glare. July 11, 2019 webinar. https://www.ies.org/lighting-education-facility-showcase/metrics-in-motion-flicker-and-glare/

Perz, M. et al., Modeling the Visibility of the Stroboscopic Effect Occurring in Temporally Modulated Light Systems. Lighting Research and Technology, 47, 281-300 (2014). https://doi.org/10.1177/1477153514534945

Plateau, H. F. Experiments on Light. The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science, Series 3, 5, 321-334 (1834). https://doi.org/10.1080/14786443408648474

Talbot, J. Sur un principe de photometrie. Bulletins de l'Académie royale des sciences, des lettres et des beaux-arts de Belgique. 2, 52-29 (1835). https://www.biodiversitylibrary.org/item/21688#page/64/mode/1up

U.S. Department of Energy. Solid-State Lighting: Color tunable products. https://www.energy.gov/eere/ssl/led-color-tunable-products 

U.S. Department of Energy. Solid-State Lighting: Flicker. https://www.energy.gov/eere/ssl/flicker

U.S. Department of Energy. Solid-State Lighting: LED basics. https://www.energy.gov/eere/ssl/led-basics

WIlkins, A. Fear of light: On the cause and remediation of photophobia. Lighting Research & Technology. 53, 395-404 (2021). https://doi.org/10.1177%2F1477153521998415