IEEE Power Electronics Magazine Compendium - March 2018 - 59

depend upon a variety of lighting applications and exposure
has a modulation that is usually less than this, versus the
factors. For example, outdoor lighting with high-pressure
intolerance of gas-discharge lighting, which can often have
sodium lamps on magnetic ballasts and low-pressure
a modulation greater than 9%. Their observations are consodium lamps sometimes has nearly 100% modulation. They
sistent with creating a low-effect level line that multiplies
are often used without health incident, perhaps because the
the no-observable-effect line by a factor of 2.5 to obtain
)
light is not being used for prolonged reading, studying, or
Mod% < 0.08 fFlicker .
other more difficult visual tasks.
The conservativeness of the low)
According to DOE/PNNL researchers
risk level line Mod% # 0.08 fFlicker can
[16], a lighting situation may deterbe examined by comparing it with
Although
the
eyes
are
mine the case-by-case tolerance of
data from a study by the U.S. DOE/
flicker. They conjecture that flicker
PNNL, which measured the detection
open and the retinal
matters most in 1) general lighting, 2)
of flicker in LED troffers in real-world
cells
are
functional,
we
spaces where children or susceptible
lighting situations [41]. The LED lamps
populations spend time (e.g., hospitested by the U.S. government lab had
are usually unaware
tals), 3) task lighting, and 4) industrial
a flicker frequency ranging from 120
of what we see during
spaces with moving machinery. Even
to 480 Hz. At full output, none of the
in other lighting applications, pruobservers was able to reliably detect
a saccade.
dence should be used when designing
flicker, even when the percent moduthe amount of light flicker. This article
lation reached ~24%. However, when
creates generalized low-risk and nodimmed, some of the lamps had flicker
effect flicker regions that would be both task and lighting
that was noticeable either by the stroboscopic effect proapplication independent.
duced by hand motion or pencil movement in the troffer
A low-risk criterion of Mod% # 0.08 ) fFlicker for flicker frearea, or, in some cases, several observers detected flicker
with normal eye motion only. The study introduces a numerquencies above 90 Hz may be a reasonable target for flicker
ical perceived flicker rating on a scale from 1 to 5. Flicker
design for LED bulb manufacturers. Some bulb manufactur"scores" below ~2.6 might lead to observer distractions.
ers have already met the criterion, while others are close to
Several items should be noted from the DOE/PNNL
meeting it. In fact, two of the seven tested LED bulbs actually
study [41].
satisfy the no-effect level criteria, Mod% # 0.0333 ) fFlicker . It,
therefore, seems possible that commercial LED luminaires
■■Each of the seven luminaires with a flicker score below
may be able to reduce their Mod% with small changes in
2.6 would violate the condition that Mod% < 0.08 ) fFlicker .
their controllers or software, or even by changing a few fil■■All 21 luminaires tested by PNNL that satisfied the conditer component values (perhaps without altering the packtion Mod% # 0.08 ) fFlicker had satisfactory flicker ratings,
age size) if they desire to meet the low-risk level criterion
whether dimmed or not.
for flicker.
■■Two LED luminaires at full output and one at dimmed
level did not satisfy the condition that Mod%
# 0.08 ) fFlicker, but they had flicker ratings above 2.6,
About the Authors
which suggested that, in the conditions tested, the flicker
Brad Lehman (lehman@ece.neu.edu) is a professor in the
was not regularly detected.
Department of Electrical and Computer Engineering, NorthIt can be inferred, therefore, that for flicker frequeneastern University, Boston, Massachusetts. He is also the
cies above 90 Hz, Mod% < 0.08 ) fFlicker should contain the
editor-in-chief of IEEE Transactions on Power Electronics.
He is a Senior Member of the IEEE.
low-risk level, but with proper design of specific lighting
Arnold J. Wilkins (arnold@essex.ac.uk) is a professor
circumstances (glare, brightness, spectral content, etc.),
in the Department of Psychology, University of Essex, Colit may be possible to operate outside the region and still
chester, United Kingdom.
be acceptable. In any case, the DOE/PNNL experiments
suggest that the low-risk region %Mod < 0.08 ) fFlicker is not
overly conservative.
References
[1] M. S. Rea, Ed. The IESNA Lighting Handbook: Reference and Application,

Conclusions
This article presents criteria that can mitigate the biological
effects of light flicker. The results were derived for LED
lighting, but they can be applied to all general illumination
technologies. LED lighting, however, has been the focus of
more attention recently because its driving methods may
introduce more flicker than other lamp technologies. The
flicker depends on the drivers selected for the LED bulb, as
described in this article. In addition, flicker health risks

9th ed. NY: Illuminating Eng. Soc. of North Amer., 2000.
[2] A. J. Wilkins, J. Veitch, and B. Lehman, "LED lighting flicker and potential
health concerns: IEEE standard PAR1789 update," in Proc. IEEE Energy
Conversion Congr. Expo., 2010, pp. 171-178.
[3] A. J. Wilkins, I. Nimmo-Smith, A. I. Slater, and L. Bedocs, "Fluorescent lighting, headaches and eyestrain," Light. Res. Technol., vol. 21, no. 1, p. 11, 1989.
[4] R. S. Fisher, G. Harding, G. Erba, G. L. Barkley, and A. Wilkins, "Photic- and pattern-induced seizures: A review for the Epilepsy Foundation of
America working group," Epilepsia, vol. 46, no. 9, pp. 1426-1441, Sept. 2005.

September 2014

z IEEE PowEr ElEctronIcs MagazInE

Table of Contents for the Digital Edition of IEEE Power Electronics Magazine Compendium - March 2018