So, what about LED flicker?

Washington, DC--How much flicker is tolerable in an LED bulb? This is a question that the LED industry is still pondering.

Mar 23rd, 2010

Washington, DC--How much flicker is tolerable in an LED bulb? This is a question that the LED industry is still pondering, and the U.S. Environmental Protection Agency's (EPA's) still-evolving specifications for integral LED lamps help to bring the question to the fore.

The source of flicker
Although integral LED lamps (those that take the form of a separate bulb that can be screwed into or otherwise attached to a lamp fixture) are already available, they are still not easy to find at your local hardware store. But LED manufacturers are actively developing them, improving their lifetimes and the purity of their white light, and at the same time reducing their cost.

But LEDs for illumination have another problem: that of flicker. Most home electricity is alternating current (AC), with a frequency of either 50 or 60 Hz depending on which country the power grid is in. Monochrome LEDs can be cycled on and off at greater than megahertz speeds; even white-light LEDs can be switched on and off faster than 60 Hz, because the phosphor decay time constant is on the order of 10 ms. And the easiest and least expensive way to build an LED to run on AC current results in the LED being modulated, sometimes even to a 100% modulation depth (cycling between completely on and completely off).

Minimum frequency: 120 Hz or 150 Hz?
Built into the EPA's ENERGY STAR specs for integral LEDs is a spec aimed at flicker. The ENERGY STAR Integral LED Lamps specification is a list of the criteria that an LED bulb must meet to qualify for use of the ENERGY STAR certification mark on product packaging (the EPA is still refining these specs).

As of December 3, 2009, the spec referring to flicker stipulated a minimum LED operating frequency of 150 Hz. This number was a change from the EPA's previous 120 Hz minimum requirement. As a result of the change, the EPA received comments from various LED and semiconductor manufacturers and researchers, with many of the commenters disapproving.

On January 22 of this year, the EPA proposed changing the operating-frequency spec back from 150 Hz to the original 120 Hz minimum requirement. The industry responded mostly positively: in fact, "several parties noted that there was limited research to justify the product redesign expense associated with the higher 150 Hz requirement," according to a letter from the EPA dated March 22, 2010 stating that the Integral LED Lamps specification is indeed being changed back to its original 120 Hz minimum.

Opinions both ways
The positive and negative responses to the EPA's January 22 proposal to switch back to 120 Hz are an indication as to the balance that the LED industry must strike. On the one hand (lower modulation frequencies), the cost of the LED product may be lower, leading to more-widespread purchase. On the other hand (higher modulation frequencies), the LED product may be more pleasant to use because it flickers less, and as a result could ultimately have more return customers.

The positive responses to the EPA letter included one from NIST: Yoshi Ohno, a well-known NIST researcher who develops LED measurement methods, said, "I support the action proposed in this letter. 120 Hz is generally no problem of flicker, as experienced with discharge lamps with magnetic ballasts, unless there are lower frequency components in it (e.g., 60 Hz). Some people are concerned, though, about deeper modulation of LEDs, e.g., AC LEDs, that might cause stroboscopic effects. I think some visual experimental study may be needed to clarify on these questions."

One negative response to the EPA letter was given by Victor Zwanenberg of NXP Semiconductors (Nijmegen, The Netherlands). He said:

"In line with your request for comments in your letter date 22 January 2010, some comments on LED operating frequency. I have some (6 years) expertise in the field of dimmable LED lamps and did several investigation in relation with visible flicker from LED lamps under both dimmed and undimmed conditions.
"In my opinion, visibility of modulation of LED light depends on the current through the device and the decay time of the phosphors in 'white' LEDs. LED's without phosphor can be modulated in light output with a bandwidth greater than 100 kHz. The decay time of current phosphors on white LEDs is much smaller than the time constant of human eye detection.
"For perception of flicker, two parameters are of importance: The modulation frequency/dutycycle and the modulation depth. There is a transitional field between these parameters where visibility of flicker is highly arbitrary. As example: Using a LED driven in pulse-width-modulation mode at a frequency of 150 Hz with full current modulation at a dutycycle of 10%, flicker will be visible. Using a LED driven in constant current mode with a current ripple of 20% at 100 Hz, flicker will not be visible. The frequency of variation for which the human observer is most susceptible is around 10 Hz. The variation causes severe stress and might induce epileptic attack.
"Visibility of flicker is, as previously stated, highly arbitrary and depends on surrounding conditions, like ambient light level, angle of view over which light variation is seen, color of the light, age of the observer, duration of the exposure, and application. In practice, using task lighting or office lighting using full modulation with a suggested frequency of 120 Hz, severe fatigue will occur with most people. This will in turn result in head-aches and rejection/replacement of the lamps. An early test at the local counsel of Hoogeveen (The Netherlands) shows such an effect. The motivation why 120 Hz is proposed by some manufacturers is, that energy storage to bridge the ripple caused by the mains input waveform is no longer required in full. This saves cost, saves space, and will make lifetime of electronics less critical since electrolytic capacitors can be avoided. However, when these products are mass-introduced into the market as replacement for existing lamp technologies, they will be rejected by the public because of perceived poor quality of light. This will hamper LED reputation and stalls further market introduction. It is fully feasible to design LED lamps using electrolytic capacitors and high lifetime, and the cost adding factor is in fact marginal in comparison to current end product costs. Deep modulation of light level will also cause motion artifacts on moving or dynamic objects, so unsuitable for area's with movement. As third factor, deep modulation will cause inter-modulation of light level when using a digital camera: recordings on these camera's will show slow but deep change in light level.
"I would propose a more balanced specification of light level variation than based upon frequency only:
-- At full current modulation ( 0-100% ) modulation frequency must be 200 Hz or larger.
-- Between 200 and 120 Hz modulation frequency, the ripple in light level must be within a bandwidth of +40%/-40% of the average value of light level.
-- At 120 Hz modulation frequency, the ripple in light level must be within a bandwidth of +30%/-30% of the average value of light level.
-- Between 120 and 95 Hz modulation frequency, the ripple in light level must be within a bandwidth of +20%/-20% of the average value of light level.
-- Modulation frequencies between 95 Hz and 1 Hz are not allowed.
Light level in above proposal is expressed in Lux."

We can work it out
Perhaps some additional research into the effects on humans of flicker at various rates is needed. In addition, don't discount the decades of experience we already have concerning light sources that modulate, including fluorescent tube lights (which have a limited modulation depth) and mercury-vapor and sodium lamps (with much greater modulation depths).

And perhaps LED bulb makers will someday have something to say to attract premium buyers: "Our bulbs produce only headache-free, 150 Hz light."

--John Wallace

Laser Focus World

More in Lasers & Sources