newsletter link
mondo arc

LED Flicker Safety Issues

Issue 63 Oct/Nov 2011

The advent of LED lighting has opened an age old debate about whether artificial lighting is healthy and safe from a human perspective. This month Dr Geoff Archenhold will attempt to get beneath the surface of the arguments about some of the factors that impact on human vision.

The use of artificial lighting has always sparked controversy about how healthy the emitted light is and whether that light is safe for humans. For example, warnings from the Health Protection Agency regarding single envelope CFL lamps and excess UV light exposure to the fears posed by the French Agency for Food, Environmental and Occupational Health & Safety (ANSES) study into a few LED lights that create a blue light hazard.

The key issue facing the lighting industry is should it put healthy and safe lighting high on its agenda when designing lighting schemes and manufacturing lighting products or should price and ignorance of the issue prevail!

The lack of knowledge on healthy and visually safe lighting is prevalent and the effects are rarely discussed outside academic circles so we have an awful long way to go to put quality lighting first. I will attempt to scratch the surface of some of the key health and safety issues with artificial lighting with the aid of several published documents reviewing the issues.

Optical Radiation
The use of artificial lighting has laid claim to a wide range of light sensitive symptoms that are aggravated by their use including:
•    xeroderma pigmentosum
•    lupus
•    migraine
•    epilepsy
•    myalgic encephalomyelitis (aka chronic fatigue syndrome)
•    Irlen-Meares syndrome (aka scotopic syndrome)
•    fibromyalgia
•    electrosensitivity
•    dyspraxia

•    autism/asperger syndrome

•    retinal diseases, such as age‐related macular degeneration (AMD)
•    chronic actinic dermatitis,
•    solar urticaria

It has been estimated by the Scientific Committee on Emerging and Newly Identified Health Risks (SCENIR) in a worst case scenario that approximately 250,000 individuals within the EU (some 0.05% of the population) might be at risk from increased levels of UV/blue light optical radiation. However, it is not just the few that suffer significantly from elevated optical radiation systems. Recent work has shown several light sources produce more harmful optical radiation than others.

The Basics
Visible light is defined as the electromagnetic radiation with wavelengths between 380nm and 750nm which is mostly detected by the human eye. Electromagnetic radiation exists from gamma rays right through to radio waves as shown in figure 1 with the visible wavelengths occupying a small section of the spectrum. In addition to wavelength, light can also be characterised quantitatively by its intensity.
The ultra-violet portion of the spectrum (fig 1) can be particularly dangerous to humans and is usually divided into three regions:
UVA (315nm – 400nm)
UVB (280nm – 315nm)
UVC (100nm – 280nm)


Thankfully, natural sunlight is attenuated as it travels through the earth’s atmosphere which means most of the radiation with a wavelength below 290nm is filtered out before it reaches the surface.
Each artificial light will have its own unique characteristic fingerprint and that is often referred to as the Power Spectral Density (PSD) or spectrum curve which identifies the amount of radiant energy at each wavelength. For example, in last month’s article we use the spectrum curves to determine the Correlated Colour Temperature (CCT) and Colour Rendering Index (CRI) of an LED light source.

Depending on the characteristics of the light emitting system, the emitted spectrum can be broad or it can have sharp ‘peaks’ at certain wavelengths; the former is the case for natural sunlight and the latter is for incandescent, halogen and certain types of LED lamps where the spectrum will contain peaks of radiant intensity at certain wavelengths as shown in figure 2.

White Light Generation - Fluorescent Lamps
A fluorescent lamp generates light from collisions in a hot gas (‘plasma’) usually containing mercury which emit photons at two UV emission lines (254nm and 185 nm). The created UV radiation is then converted into visible white light by UV excitation of a fluorescent coating on the inside of the glass tube. The chemical composition of this coating is selected to emit in a desired spectrum for example warm white lamps may use three part phosphors. For example, fluorescent lamps can be enriched for blue light (wavelengths 400-500 nm) in order to simulate daylight better in comparison to incandescent lamps. Like fluorescent lamps, CFL emit a higher proportion of blue light than incandescent lamps.

White Light Generation – LEDs
The vast majority of LEDs use a similar principle but instead use a blue LED coated in a phosphor material to generate the white light. The advantage of LEDs as a light source is the pump wavelength is around the 470nm wavelength +/-20nm and therefore does not contain UVA, UVB or UVC wavelengths that are harmful.

UV Radiation and Blue Light
With a few exceptions (most notably the formation of pre-vitamin D3), most photochemical reactions caused by UV radiation in the skin and eyes are detrimental as proteins and DNA can become damaged and dysfunctional if exposed in a prolonged manner.

There are internationally recognised exposure limits for the radiation (200-3000 nm) emitted from lamps and luminaires that are set to protect from photobiological hazards (International Electrotechnical Commission 2006). These limits also include radiation from CFLs.

The UV content of the emitted spectrum depends on both the phosphor and the glass envelope of the fluorescent lamp. The UV emission of incandescent lamps is limited by the temperature of the filament and the absorption of the glass. Some single-envelope CFLs emit UV-B and traces of UV-C radiation at wavelength of 254nm. Experimental data has shown that CFLs produce more UVA irradiance than a tungsten lamp. Furthermore, the amount of UVB irradiance produced from single-envelope CFLs, from the same distance of 20 cm, was about ten times higher than that irradiated by a tungsten lamp.

For example in 2008 the Health and Safety Executive studied the emissions from various CFL lamps and found that some energy saving compact fluorescent lights can emit ultraviolet radiation at levels that, under certain conditions of use, can result in exposures higher than guideline levels. The HSE’s view was that open (single envelope) CFLs should not be used where people are in close proximity - closer than 30 cm or 1 ft - to the bare light bulb for over 1 hour a day. The Agency advised that for such situations open CFLs should be replaced by the encapsulated (double envelope) type or alternatively, the lamp should be moved so that it is at least 30 cm or 1 ft away.

The photochemical blue light hazard (BLH) can be evaluated on the basis of the standard EN 62471. The latter classifies light sources into risk groups 0, 1, 2 and 3 (from 0 = no risk through to 3 = high risk). The sun would be classified as being in the highest risk group.

Risks can be allocated to the risk groups according to different measurement criteria:
•    Measure the distance at which an illuminance of 500 Lux is attained (a typical value for general lighting purposes). According to EN 62471, the 500 Lux criterion must be used for lamps intended for general lighting (including lamps for lighting offices, schools, homes, factories, roadways, or automobiles).
•    The second criteria measures photo biological safety from a distance of 200 millimetres. The 200 millimetre criterion is to be used for all other lamps (including for example lamps for such professional uses as film projection, reprographic processes, sun tanning, industrial processes, medical treatment and searchlight applications).
Table 1 compares the portions of the spectra of various light sources that overlap with the blue light hazard function and also the CIE photopic function. The ideal light source would want a low BLH value and a high CIE photopic value to be a truly efficacious (and least hazardous) source.

It is interesting to see the halogen lamp only generates 5.7% of its light within the BLH region and cool white (6500K) LEDs generate 24.5% of their light in the BLH region. Of course this is to be expected as halogen lamps are warm white and 6500K LEDs generate a colder bluer light.
Previously it has been difficult to measure the BLH radiation of lighting fixtures as it required the use of expensive laboratory equipment however Blueside Photonics have developed the Blue Light Hazard estimator PC software which only needs the use of a standard lux meter providing you know the spectrum of the light source within the fixture. 

If you then measure the illuminance of the fixture with the light meter one can instantly estimate the hazard (provided you are aware of the source size ‘limitations’ that apply to the test). Figure 3 shows the example whereby a cool white LED fixture with a 500 lux point at 2 metres (equating to a 2000 cd source) is not hazardous by the GLS criteria referred to in the BLH standard, the fractional exposure of the allowed limit (for all day viewing) is 0.477: a 3000 K incandescent source with the same intensity metric delivers 0.217 of the daily dose.   
Figure 4 shows a comparison of light sources against the blue light hazard as measured by CELMA and as shown the highest irradiance occurs with any light source as the CCT increases.

Focusing on LEDs we see that if they are measured using the 500 Lux criterion, none of the current LED products available (at any colour temperature) belong to risk group 2. However, a study by ANSES in 2010 found that three out of nine high‐output discrete LEDs tested could be classified into risk group 1 if the 500 Lux criterion of the IEC 62471 standard is applied.

ANSES caused a major concern by stating the intense blue-white light was a “toxic stress” on the retina, with a severe dazzling risk and that young people in particular are sensitive to this risk as their eyes are still developing and the lens is not capable of filtering out the light wavelengths. This issue didn’t however state that other artificial light sources cause a similar hazard and thus one could argue that objectivity was not prevalent in the report.

Indeed, one lighting industry association, CELMA, mentions that blue light exposure is important to human beings. Blue light with a peak around 460‐480nm regulates the biological clock, alertness and metabolic processes. In natural conditions, outdoor daylight fulfils this function. Yet, people spend most of the day indoors (offices etc.) and are often lacking the necessary blue light exposure. Blue and cool white light sources can be used to create lighting conditions such that people will receive their daily portion of blue light to keep their physiology in tune with the natural day‐night rhythm. Due to the highly flexible application possibilities, LED based light sources are particularly well suited for that purpose.

A further study undertaken in 2010 by Health Canada measuring the amount of UV radiation given off by Halogen, Incandescent and CFL lamps revealed that at a distance of 30cm from a 60W incandescent bulb (see AG60 in Figure 5), UVR damage to unprotected skin or eyes may result if the bulb is repeatedly used for longer than 3.4 hr/day.  Several single envelope CFL products required less time to achieve the recommended daily amount and the Halogen lamp required more than six hours of continuous exposure.

Effects of Light – Epilepsy
Five percent of the total world population has single seizures, and the annual incidence is 50 in 100.000 (WHO 2001). About five in 100 of epileptic people have photosensitive epilepsy. Photosensitive epilepsy is a form of epilepsy in which seizures are triggered by visual stimuli that form patterns in time or space, such as flashing lights, bold, regular patterns, or regular moving patterns.
The visual trigger for a seizure is generally cyclic, forming a regular pattern in time or space. Flashing or flickering lights or rapidly changing or alternating images are an example of patterns in time that can trigger seizures.
While photosensitivity of epileptics has scientifically been proven there are few studies which determine if the flicker frequency range > 120 Hz causes seizures.
All artificial lighting systems suffer from flicker however the frequency of the flicker will depend on the technology. For example incandescent lamps would have a frequency similar to that of the mains frequency however high frequency ballasts are now used on fluorescent tubes which have switching frequencies in the 20-90kHz range.

Effects of Light – Migraine
Migraine can be defined as an intense pulsing or throbbing pain in one area of the head. It is often accompanied by extreme sensitivity to light and sound, nausea, and vomiting. Migraine is three times more common in women than in men.

It is estimated that 14% of the adults in Europe have migraine (Stovner et al. 2006). According to self-reported information, certain visual patterns can reliably trigger a migraine attack, such as high contrast striped patterns or flickering lights. Fluorescent lamps can cause eye-strain and headache (Wilkins et al. 1991). Patients with migraine show somewhat lowered flicker fusion thresholds during migraine-free periods. In addition, photophobia, which is an abnormal perceptual sensitivity to light experienced by most patients with headache during and also between attacks, is documented in many studies.

Effects of Light – Irlen-Meares
Irlen-Meares manifests itself primarily as a difficulty with reading and spelling which may be improved by use of coloured lens or overlays. Self-reporting suggests that fluorescence lighting in contrast to incandescent light aggravate the symptoms of dyslexia. Probably the main problems are caused by UV radiation and blue light, emitted by cool white tubes.

Effects of Light – Autism/Aspergers Syndrome
Autism is a neuro-developmental disorder characterized by deficiencies in social interactions and communication skills, as well as repetitive and stereotyped patterns of behaviour. Recent epidemiological data show that autism is a frequent disorder, observed in one child in 500. A study undertaken by Colman et al. suggested that repetitive behaviour can be aggravated by the flickering nature of fluorescent illumination, had interpretative problems and could not be replicated.

Effects of Light – Myalgic Encephalomyelitis (Chronic Fatigue Syndrome)
Chronic fatigue syndrome is one of several names given to a potentially debilitating disorder characterized by profound fatigue which lasts for at least six months. It has a prevalence that varies from 0.2% to above 2%. According to self-reporting, about 52,500 people in the UK (= 21% of myalgic encephalomyelitis) have increased sensitivity to light.

Flickering of Artificial Light Sources
The dissemination and understanding of the effects of flickering of artificial light sources is in its infancy. However there are clearly some benefits for the lighting industry to take more notice especially with the opportunity afforded to it through the new technologies of LEDs and OLEDs. There are several aspects to flickering of the light output as follows:
•    Modulation Frequency
    - Occurrences, or cycles of a periodic signal per unit of time
•    Modulation amplitude, Modulation depth, depth of modulation
    - Peak-to-peak magnitude variation oF a periodic signal, sometimes referenced to signal mean
•    DC Component, DC Value, DC Offset, Offset
    - Mean value of a periodic signal
•    Flicker, flutter, shimmer
    - Light source modulation
    - Not associated with any periodic signal metric

Figure 6 generated by Northwest National Laboratory defines the key parameters of flicker.

Interestingly the flicker of a light source is defined within a standard EN12464-1:2002 which states that flicker causes distraction and may give rise to physiological effects such as headaches. Stroboscopic effects can lead to dangerous situations by changing the perceived motion of rotating or reciprocating machinery. Lighting systems should be designed to avoid flicker and stroboscopic effects. It is noted within the standard that this can usually be achieved for example by use of DC electrical supply for incandescent lamps, or by operating incandescent or discharge lamps at high frequencies (around 30 kHz).

Unfortunately the majority of LED drivers used today have very poor flicker characteristics and a typical constant current LED driver ripple current may be seen in figure 7. Here a nominal 700mA LED driver actually provides 652mA current out with a peak to peak ripple current of 142mA or 21.8% with a ripple frequency of 100Hz (2x Mains frequency). They key aspect for designing or specifying LED products is to ensure that the LED drivers are not overdriving the required maximum forward current specified by the LED manufacturer. Typical measurements of LED ripple current have been seen up to about 85% of nominal current so this could be a major cause of stress on the LED and wire bonds for some manufacturers so make sure you check!

Of course if your LED driver uses Pulse Width Modulation technology then your ripple current is actually 100% as the current through the LED varies between 0 and full current at a specific PWM frequency. If you require high resolution current control then unfortunately a major disadvantage with PWM is that the flicker frequency drops to only a few hundred Hertz at best.

A recent flicker study by the Lighting Research Center in New York State tested how 10 humans detected and more importantly accepted flicker using both the flicker frequency and also the degree of percentage flicker with very interesting results as shown in figures 8 and 9.

In figure 8, the lower the flicker frequency enabled the subjects to detect stroboscopic effects of the light source especially if the flicker percentage is high. For example, the majority (80-100%) of the subjects would be able to see stroboscopic effects of a PWM (100% flicker) driven LED system with a PWM frequency of up to 1KHz whereas an LED light source with a flicker frequency from 1KHz and only 5% flicker would not be seen by observers. For clarity an LED light source with 5% flicker is essentially a DC light source with 5% ripple current.
However, figure 9 shows that the majority of subjects in the study would accept light sources which give rise to the stroboscopic effects irrespective of the flicker percentage after a flicker frequency of approximately 1500Hz.

The advantages of LED systems is the electronics required to control them are advancing at a tremendous pace and therefore it is now possible to see driver technologies that implement DC driving which are capable of dimming the LED output down to less than 1% without using PWM to control the LED current with long life and low ripple current.

One such promising technology is the Digital Signal Processing technology incorporated into the iDrive range of LED drivers. The DSP technology directly controls the LED current at a switching frequency up to 120KHz and the RMS ripple current using long life (300,000 hour solid-state) capacitors is less than 5% as shown in figure 10.

The good news is that several groups are looking into the effects of flicker including the IEEE PAR1789 standards group which is recommending practices for modulating current in High Brightness LEDs for mitigating health risks to viewers. You can see more from this group on their website  Also Nema discusses flickering within NEMA LSD 49-2010  entitled “Solid State Lighting for Incandescent Replacement—Best Practices for Dimming” available free at

How to avoid flicker when specifying projects
•    Consult a trained, experienceD lighting designer to evaluate your lighting application sensitivity to flicker
•    Consider flicker when specifying Nproducts for sensitive applications
•    Ask manufacturers how their product modulates light output
    - LED driver
    - Integral PWM dimming (what frequency & bit resolution)
    - DC dimming (what switching frequency)
    - Behaviour during failure modes
•    Be aware of system components which can cause flicker
    - Mains dimmable switches and drivers
    - Step-down transformers connected to less than minimum load
•    Understand the issue further and stay up to date

Artificial Light in general does not pose too much of a hazard. However certain lighting technologies offer the opportunity to improve quality of lighting from a human visual perspective by increasing the flicker frequency and flicker percentage.
These aspects are really important but are generally never considered when installing lighting installation. For example, how many LED street light applications use LED drivers that have a flicker frequency of 100Hz – I don’t know the answer but I bet the specifiers of street lights have probably never specified they should be higher than 100Hz! Such a low frequency could cause serious implications when driving as stroboscopic effects could cause visual difficulties in estimating speed for example.  
Ideally, all light sources should try and maintain a flicker frequency > 30KHz and with a flicker percentage of less than 10%. This is definitely a major challenge for the LED lighting industry moving forward although the technologies are available to achieve this but not at the price the market currently wants.

Therefore, I suggest if you want a healthy LED lighting system for your clients invest in quality and procure a lighting system with high switching frequencies and preferably uses DC outputs with low ripple current rather than switched frequency outputs.
Geoff Archenhold has been seconded twice to the UK Government to support the Lighting, LED and Photonics industry and currently helps LED companies develop business plans to raise investment from the finance community. He is an active investor in LED driver and fixture manufacturers and a lighting energy consultancy. The views expressed in this article are entirely those of Geoff Archenhold and not necessarily those of mondo*arc.


Figure 1: The colour spectrum.

  • Figure 2: Various light source and eye response spectrums.

  • Table 1: Comparison of spectral light sources with the BLH and photopic functions.

  • Figure 3: Estimating the blue light hazard of lighting fixtures.

  • Figure 4: Comparison of light sources against the blue light hazard.

  • Figure 5: Calculated time to achieve maximum exposure at a distance of 30cm.

  • Figure 6: The parameters for defining flicker of a light source.

  • Figure 7: The ripple current from an LED driver connected to an LED.

  • Figure 8: The LRC results on human detection of stroboscopic effects.

  • Figure 9: The LRC results on human acceptability of stroboscopic effects.

  • Figure 10: The variation of ripple current in a DSP enabled LED driver. 

Related Articles


Follow us on…

Follow Mondo Arc Magazine on Twitter Follow Mondo Arc Magazine on Facebook Follow Mondo Arc Magazine on Linked In

mondo arc india

darc awards DWLF IALD PLDC LRO