Flicker in LED lighting
25 March 2017

In 2009 and 2011, I performed some tests of both CFL's and LED lamps, part of which was to quantify the light modulation spectra of each, comparing them to incandescent lamps. The articles touched on flicker and at that time I had identified that the worst types of lighting for flicker were the long tubular fluorescent lights used with magnetic ballasts. Incandescent lamps had less flicker, then CFL's generally had lower levels still. We only touched on a couple of LED examples at that time. As it turns out, LED lamps are quite likely to produce flicker, although it is not an inherent problem of LED's themselves, but rather to do with the methods of powering them.

A standard emerges

In 2015, the IEEE (Institute of Electrical and Electronic Engineers) published a new standard on the subject, based on the findings of a working party with many contributors. This standard, referred to as IEEE 1789-2015 is entitled:

IEEE Recommended Practices for Modulating Current in High-Brightness LEDs for Mitigating Health Risks to Viewers
Even abbreviating that to IEEERPFMCIHBLEDFMHRTV was a little cumbersome so they just called it IEEE1789.

What is it about?

This standard is formed from studies into the effects of light flicker on people and is not restricted to those who suffer epilepsy, but a wider group who might become nauseous or have headaches or other effects; some of which are not readily apparent. The standard is copyrighted but they usually don't mind being paraphrased and summarised.

Purpose of the standard

"Presently, there are no standards on safe modulating frequencies for high-brightness LEDs. Vendors suggest various driving frequencies; some at low frequencies and others at high frequencies. In the late 1980s and early 1990s, studies showed that office fluorescent lighting with magnetic ballasts modulating at twice the ac line frequency increased the incidence of health-related problems, such as headaches, eyestrain, and, when the lamps were in failure, epileptic seizures. The detrimental effects depend on factors such as brightness, angle of viewing, wavelength, and depth of modulation, among others. The purpose of this document is to describe some possible health risks associated with low-frequency modulation of high-brightness LEDs and provide recommended practices to aid the design of LED driving systems to modulate at benign frequencies in order to help protect against the described health risks."

"For most people, flicker that occurs with a frequency of less than 60 Hz is visible. The frequency at which a flickering light source fuses into an apparently constant source varies for individuals and depends on the modulation amplitude, adaptation luminance, and visual field size of the source. However, this critical flicker fusion frequency (CFF) occurs generally in the range of 60 Hz to 100 Hz (Kelly [B65]). Invisible flicker, occurring at a rate greater than the CFF, may nonetheless have physiological effects even though the individual normally cannot report the conscious perception of flicker (see Clause 6). Flicker was an issue when magnetically ballasted fluorescent lamps were common, before the mid-1990s. Research at that time identified flicker of the light source to be related to migraines, headaches, reduced visual performance and comfort, and other possible neurological health issues (see Clause 7). When high- frequency electronic ballasts were introduced for energy efficiency, the negative effects of flicker were reported less frequently and largely disappeared from public discourse. In the meantime, magnetically ballasted high-intensity discharge (HID) lamps have been continuously used for outdoor light with relatively few complaints despite their high modulation depth."

"With the introduction of SSL [solid-state-lighting] products to the marketplace, flicker has re-emerged as a consideration, partly because the modulation of light-emitting diode (DOE [B106]) light output has been frequently observed to be greater than the modulation seen with fluorescent or HID sources (Poplawski et al. [B84]). For LED sources, flicker is primarily determined by the driver. Some driver designs produce little to no detectable flicker at full or dimmed outputs; others flicker noticeably at both full and dimmed output; still others produce little to no flicker at full output, but flicker objectionably when dimmed. Some LED products produce flutter or light level instabilities while the dimming level changes from one level to another. The flutter or light level instabilities disappear when the dimming level remains constant."

The recommendation

Maximum tolerable percent flicker is related to the flicker rate. The higher the flicker rate, then the greater percent flicker which can be tolerated. Limits determined over many studies are:

Percent flicker should be less than 0.08×f for Low-Risk Level and less than 0.0333×f for No Observable Effect. This latter limit is more strict and would be applied if you wanted to guarantee there would be no impact on the most sensitive of individuals. For the most common flicker frequency (which is 2 x AC line frequency) the low-risk limit in 50Hz countries is 8% flicker and is 10% for 60Hz countries. Similarly, the No observable effect limit (NOEL) is 3.3% in 50Hz countries and 4% in 60Hz countries.

What situations are important

Not everyone is sensitive to flicker and not every lighting situation matters. However, it must be remembered that there are often subtle and delayed effects on people. That headache when you come home from work might not be your posture - it might be the lighting. In general though, flicker matters where there is bright area lighting, like in supermarkets, warehouses, workplaces, clinics, classrooms, task lighting or where you want to shoot video indoors. It is not so important where there is lower intensity lighting, such as roadways, parking lots, accent lights, Christmas tree style lighting and LED indicators, although the known epilepsy flicker rates of between 3Hz and 70Hz must always be avoided. Time of exposure has been shown to be a factor, but there is little data on just how long exposure is required to cause effects on people and so best practice means reducing percent flicker to below the limits given by the standard.

Powering LEDs and pulse-width modulation

There are many ways to power an LED lamp. Remember LEDs operate from dc, so operating them from pure dc is best. There is zero flicker from LEDs fed with pure dc. Powering from the AC line means that some method of generating dc from ac has to be employed. In the most simplistic form, an bridge rectifier and resistor are used to power a chain of individual LEDs. One gets 100Hz (or 120Hz) pulses of light and if nothing else is done, this will result in a high percentage of flicker and vastly exceed the 8% (10%) recommendation. One mitigation strategy is to power two identical strings of LEDs which are connected in the opposite sense so that one chain of LEDs works from the positive cycles and the other from the negative cycles. If the LEDs are spatially integrated, then much less flicker is the result, although this alone is not enough to reduce flicker below the standard. Sometimes a filter capacitor will be used, but they take up space, so only low values are generally used and even these measures aren't enough.

Modern LED luminaires and lamps have an efficient switching converter built into their base. These convert the AC line frequency into a much higher frequency (usually 24kHz to 48kHz ). Flicker at those rates is not a problem, however most of these do not eliminate enough of the 100Hz (120Hz) twice-line frequency, which is still passed to the LEDs.

Pulse-width modulation [PWM] is quite common for smaller lamps such as cycle lamps, torches, and some task and counterpoise lamps. This method generates pulses at a rapid rate and then varies the width of the pulses fed to the LEDs. If the width is narrow, the LEDs are dim and if the pulse is wide, the LEDs are at maximum. Very efficient and relatively easy to build; at least for low power lamps. However, PWM is 100% flicker. Nonetheless, it can be used safely if the frequency is high enough. Solving the percent flicker equation for frequency using 100% flicker, the minimum PWM frequency becomes 1.25kHz, or 3kHz to encompass the most sensitive of individuals. A PWM system running at or above these frequencies will not cause physiological problems.

Dimming

Retrofitting LED lights in place of incandescent lights which have a standard domestic dimmer plate is going to be problematical. In fact I would say remove the dimmer. I personally consider that the need for dimming is over-rated. Actually I would prefer to see manufacturers make lamps which have say four banks of LEDs, individually wired out, so that you could power them all for full brightness or just one chain for low brightness. There is no need for infinitely variable brightness. Anyway, I digress; this article is not about the design of dimmers. I note that studies have found that, for LED lights with dimmers, even assuming they work normally at full brightness, can create additional flicker when dimmed. Frequencies unrelated to AC line frequency can be produced and under fault modes some quite significant low frequency pulses are noted. Even during the time the dimming knob is rotated, flicker and flashing can be caused. The standard 2-wire leading edge, trailing edge and universal dimmers ought to be avoided when running LEDs, even if the nice man from Bunnings says they work fine.

Brief flicker tests

It is relatively easy to measure percent flicker as defined by IEEE1789. Any electronics lab equipped with a photodiode and an oscilloscope can do it ... I suspect more flicker meters will be available before long.

A collection of LED and other lamps found around the lab have been measured and here are the results.

Incandescent Globe:
flicker =12.5%

Elite CFL:
flicker = 7.1%

Osram CFL:
flicker = 7.7%

Martec Downlight:
flicker = 23.1%

LEDLUX Ceiling Button:
flicker = 25.5%

Philips 18W LED:
flicker = 7.4%

Philips 10.5W LED:
flicker = 13.7%

Orbit 7W LED:
flicker = 0%

desk magnifier/LED:
flicker 84%

Results of my flicker tests

A once common, but now hard to find frosted incandescent globe was used as a point of comparison. Few people complain of flicker from these, however it measured above the 8% limit recommendation. A clear glass incandescent was also tried to see if the frosting changed anything, but the flicker increased by only 2%.

Next, a couple of mid-aged CFL's were tried, both inside the spec. Two Philips LED globes were tested and the large 18 watt was below 8% although the 10.5 watt version produced 13.5% flicker. However a relatively low power 'Orbit' brand LED globe produced no measurable flicker.

Both a Martec 'Genesis' LED downlight and a LEDLUX ceiling button measured well over 20% flicker. Finally, a desk magnifier with built-in LED lighting that was bought from Jaycar measured an astonishing 84% flicker. I had already noticed that photographing in the light from this lamp produced strobe effects. Since this lamp is used for long periods in the workshop, I will attempt to make this better.

All the lamps in this test had flicker rates of 100Hz and although the HF switching noise could also be seen on some of the solid state lamps and CFL's, the flicker levels measured here are all from the 100Hz component and so the standard 8% limit applies.

Going forward

It has been estimated that by 2020, half of all lighting will be LED. Manufacturers should be giving attention to flicker from their lamps already. Improving flicker from LED lamps is not hard and the incremental cost is small. Everyone involved in the industry from consumers to consultants and manufacturers should be elevating the importance of flicker, ensuring that it is part of a lamp specification, and making it as important as lumens per watt and colour-rendering index and all the other new terms that we have had to learn in recent times.

Further Reading

For further explanations of lighting flicker, have a look at the site of Peter Erwin. Peter is an electrical engineer in Germany. He has made extensive studies of LED flicker and also created a new measurement index called Compact Flicker Degree (CFD). The CFD takes into account all aspects of the modulation of light from lamps and expresses the result of a lamp measurement as a single percentage value, and also produces a traffic-light style indicator of merit. Peter presented his technique at the 6th LED Professional symposium in Bregenz, Austria during September 2016, where it was well received. A number of LED manufacturers agreed that it is time for such a measurement method and are undertaking to use it for their LED products.

Comments

From: Axino May 2017

The Jaycar desk magnifier/LED lamp mentioned above has been investigated. It uses a simple full-wave rectifier from mains. For the technically-minded, the schematic and an improvement to the flicker percentage is described on the Axino Test Bench page.

Make a comment via the contact page