Flicker in CFLs, linear fluorescent lamps and LEDs.

It is generally known that fluorescent lamps operated on
line frequency EM ballasts produce a substantial amount of
flicker at twice the power line frequency: 100 Hz or 120 Hz
depending upon country. This flicker is known to cause
create headaches in some people.

The conventional wisdom is that fluorescent lamps operated
on high frequency fluorescent ballasts will not create
headaches. The phosphor cannot respond to variations in UV
intensity at frequencies of 20 kHz and above, and the small
amount of light emitted directly by the mercury will be
modulated at frequencies far too high to be detected by the
eye/brain system.

However, after receiving a number of questions about
electronically ballasted CFL's and headaches from Ira
Krepchin of E Source, I ran a short series of experiments
and have concluded that the conventional wisdom may be

Electronic ballasts for both linear and compact fluorescent
lamps, along with drivers for LEDs, almost always operate
from 50 Hz or 60 Hz single phase AC power sources. The AC
power is rectified to make DC power using a full-wave bridge
or perhaps a voltage doubler. The DC power is then used to
operate a high frequency inverter that, in turn, drives the
fluorescent lamp. (For an LED the DC power is used to with
a current regulator to drive the LED.)

The DC power produced by such a full-wave bridge will be
100% modulated at twice the power line frequency. If the
DC power source is modulated at 100 Hz or 120 Hz, the high
frequency power produced by the inverter will likewise be
amplitude modulated at 100 or 120 Hz. The extent of the
modulation of the high frequency power will be dependent on
the design of the inverter, but many inverter designs will
produce approximately the same percentage of 100 Hz or 120
Hz modulation on the high frequency power as exists on the
DC power supply. No practical inverter can produce a
non-modulated high frequency signal from a DC power source
that has 100% 100 Hz or 120 Hz modulation.

To reduce the modulation of the DC power supply in the
ballast an energy storage device, typically an electrolytic
capacitor, is connected to the output of the rectifier.
However, due to cost and size constraints, plus the fact
that the addition of this energy storage capacitor creates
low input power factor, the value of the capacitor used is
large enough to eliminate much of but not all of the 100 Hz
or 120 Hz modulation of the ballast's DC power source.

The result of this DC power modulation is that the high
frequency power generated by a typical high frequency
fluorescent lamp ballast can have a substantial amount of
100 Hz or 120 Hz amplitude modulation superimposed on the
high frequency power used to drive the lamp. This
modulation, in turn, creates 100 Hz or 120 Hz modulation of
the light output from fluorescent lamps operated with high
frequency electronic ballasts.

I have run a series of measurements on various
electronically ballasted CFLs and found substantial amounts
of 120 Hz flicker; typically 1/3 to 1/4 as much as found in
linear lamps operated at 60 Hz on EM ballasts. I have also
located an obscure paper (Study on Fluorescent Lamp
Illumination and Flicker; Xiaoming, Ke, Ying & Wenziang;
Fifth International Conference on Power Electronics and
Drive Systems, 2003) that reports flicker index measurements
of 1.1% to 5.4% for 36-watt linear fluorescent lamps
operating on electronic ballasts. (This same lamp had a
flicker index of 8.8% when operated on an EM ballast.)

If you attended Light fair 2007 and received one of the
"Flicker Checker" tops provided by Sylvania you can run your
own experiments. Line frequency operated fluorescent lamps
will show strong strobe lines as expected. However, you
will be able to observe weak strobe lines when the spinning
disk is illuminated by CFLs or even line powered
incandescent lamps. The only light sources that I have
found that do not produce strobe lines are flashlights and

While I strongly suspect that the lower percentage flicker
from fluorescent lamps operated on high frequency electronic
ballasts will reduce the likelihood of people developing
headaches, the data shows that we can no longer rule out the
possibility that some people may develop headaches even when
using electronically ballasts linear and compact fluorescent

Note that the same flicker issues will apply to LED lighting
systems. LED flicker may be worse than Fluorescent lamp
flicker because, unlike a fluorescent lamp ballast, there is
no technical reason to filter the DC power link at all. In
addition, LEDs that do not use phosphor have faster response
time than the phosphors used in fluorescent lamps. On the
other hand, many LED systems employ some sort of current
regulator, which may serve to reduce the percentage ripple
in the LED drive current relative to that on the internal DC
power supply.

I don't know of any studies that show the prevalence of
lamp flicker induced headaches as a function of flicker
index. If no such studies have been done they need to be
undertaken. In addition, based on a pervious discussion in
this group initiated by Terry McGowan, I believe this data
shows that Energy Star needs to add a flicker limit to the
Energy Star Requirements for both CFLs, and LED-based

Ira and I will be submitting a letter on this subject to
LD+A and a longer paper with data will probably be submitted
to the IEEE Transactions on Industrial Applications or the
Journal of the Illuminating Engineering Society. The data
will also be posted at www.cflfacts.com.

Vic Roberts
To reply via e-mail:
replace xxx with vdr in the Reply to: address
or use e-mail address listed at the Web site.

This information is provided for educational purposes only.
It may not be used in any publication or posted on any Web
site without written permission.