COMMISSION
INTERNATIONALE DE L'ECLAIRAGE
INTERNATIONAL COMMISSION ON ILLUMINATION
INTERNATIONALE BELEUCHTUNGSKOMMISSION
DIVISION 2: PHYSICAL MEASUREMENT OF LIGHT AND RADIATION
Report on the CIE Workshop:
Photometry of Flashing
Lights
Warsaw, Poland, June 25, 1999
Chairman: Yoshi Ohno, USA
INTRODUCTION
This workshop was organized jointly by CIE Division 2 and
Division 1 to address issues of measurements of flashing lights
related to both the physical measurement aspects (Div. 2) and
human vision aspects (Div. 1).
Flashing lights are used in many applications, such as
signaling devices (aircraft anticollision lights, roadway
barricade lights, emergency vehicle warning lights, fire alarm
lights, light houses, etc.), imaging devices (strobe lights for
cameras, copy machines, image scanners, etc.), and measurement
instruments (colorimeters and spectro-reflectometers). Needs on
the uniform and accurate measurements of flashing lights have been
recently raised in these areas. While some references are
available (mostly published in the 1960s and 70s), there is no CIE
publication that defines and gives recommendations on the
photometry of flashing lights (note [1]).
A good example, in our daily life, of the effective signaling
by flashing lights is found in a little LED message lamp on the
telephone at the chairman's office. The red LED lamp used to blink
when a message was recorded, and it was immediately recognized.
Now the phone system has changed, and it no longer blinks. It
takes much longer time (sometimes a few hours) before it can be
recognized. When the LED is blinking, it takes less energy but
works much more effectively.
One specific area that has recently drawn much attention is
the measurement of aircraft anticollision lights. After a recent
aircraft accident and many near-misses reported in the United
States, the Federal Aviation Administration (FAA) issued a
regulation on mandatory maintenance of the anticollision lights in
commercial aircraft. The regulation created a need for written
procedures for measurement of the effective intensity of
anticollision lights. A task group was formed to draft ARP5029 by
the Society of Automotive Engineers (SAE) A20B Aircraft Exterior
Lighting Subcommittee. In the process of developing the document,
a number of questions were raised on the photometry of flashing
lights that have not been resolved. This issue has drawn
international attention, and led to the formation of a CIE
reportership R2-17 (Aviation Photometry) and a new technical
committee TC2-49 (Photometry of Flashing Light).
This workshop was planned to collect the latest knowledge on
the research and technologies available, identify the current
problems in this area, and to provide inputs not only to this
committee but also more widely to the future work of Divisions 1
and 2.
PRESENTATIONS
1. Historical Overview of Flashing Light
Photometry
H. -J. Schmidt-Clausen
Department of Lighting Technology
Darmstadt University of Technology, Germany
An overview was given on research into the visual response to
flashing lights, including work by Bunsen and Roscoe (1862), Bloch
(1885), Allard (1876), and Blondel and Rey (1911). The Blondel-Rey
equation [3], now widely used, was determined for
achromatic threshold illuminance using rectangular light pulses in
a dark background. It was shown that the duration of the
integration (t2-t1) is not the entire pulse duration (t3-t0) but
determined from the pulse shape by an integral equation. The
Blondel-Rey constant (0.2 s) is actually not a constant but a
variable as a function of field size, adaptation luminance, color,
and pulse shape [4]. To describe the effective intensity
for any pulse shape, the form factor method [2,4] has been
introduced as an extension of the Blondel-Rey equation. The form
factor is the ratio of the total area of the pulse waveform to the
area of the rectangle which the waveform fits in. With this
method, the effective intensity can be calculated from the
integrated intensity (over the entire duration) and the peak
intensity of the pulse. This method is adopted in the ECE
Regulation 65 "Flashing Lights". Recent results on the spectral
luminous efficiency of the human eyes for flashing lights were
also presented. The Blondel-Rey constant (variable) has been
analyzed as a function of wavelength, and shown to be largest for
the blue and smallest for the yellow region. The relative spectral
sensitivity for a light pulse of three different durations has
been obtained. The results show much higher sensitivity in the
blue region compared with the V(l) function [5].
2. Relevant Work in CIE Division 1
(Supplementary System of Photometry)
Ken Sagawa, Director of CIE Division 1
National Institute of Bioscience and Human-Technology,
Japan
Basic characteristics of the human visual system were
described. It was shown that the visual system response is very
complex, and must be characterized in three dimensions &endash;
spectral, spatial, and temporal. The photometry of flashing lights
deals with the temporal characteristics. The visual response
depends on the field size (foveal, peripheral), type of the task
(detection, recognition of pattern, etc.), and transient
(adaptation, etc). When a photometric system is defined, large
variations of response depending on these parameters must be
considered. Division 1 is establishing a photometric system based
on apparent brightness at any luminance level, which is called the
"supplementary system of photometry." Temporal aspects of light
are regarded as one of the critical factors affecting apparent
brightness. A need to introduce the efficiency of flashing light
into the supplementary system of photometry will be
considered.
3. Development of SAE ARP 5029 on the Measurement of
Aircraft Anticollision Lights
David F. King
Boeing Commercial Airplane Group, USA
Anticollision lights are one of many methods of increasing
pilot awareness of other aircraft so that collisions can be
avoided. In the early 1990s, the FAA issued new maintenance
requirements that aircraft anticollision lights be maintained by
the airlines at or above the required certification effective
intensity of 400 cd for both red and white anticollision lights
(based on Blondel-Rey equation) in the horizontal plane. This led
to industry concern about the accuracy and the uniformity of
measurements. A task group ARP5029 was established under SAE A20
Aircraft Lighting Committee to develop an SAE Aerospace
Recommended Practice (ARP) on the measurements of aircraft
anticollision lights [6]. The document defines test
constraints, develops detailed test procedures and equipment
recommendations, specifies a photometer calibration source and
traceability, and estimates the uncertainty for laboratory, shop
and field measurements. In conjunction with the development of SAE
ARP 5029, NIST developed a flash photometer calibration process
(later presented by Y. Ohno). The document specifies the effective
intensity based on the Blondel-Rey equation with an approximation
of Ętª0 since most of the anticollision lights used today are
xenon flash with very short duration (ª1ms), and the document
covers only this type of source. The document (45 pages) was
published in December 1998 after four year's efforts by more than
100 members. The document has limitations, and some questions
remain unresolved: 1) Is the Blondel-Rey relationship the best way
to determine the effective intensity? 2) Is 0.2 second the best
constant? 3) Is it correct to use the photopic response when
vision is mesopic? 4) Would a more orange color be better than
aviation red for the red anticollision lights? 5) Is 400 cd the
best compromise between glare and conspicuity? 6) How should
multiple flashes be measured?
4. Establishment of the NIST Flashing-Light Photometric
Scale
Yoshi Ohno
Optical Technology Division
National Institute of Standards and Technology
USA
Upon request from the FAA, NIST undertook the task to
establish the flashing-light photometric scale to provide
calibration services for anticollision light photometers. The work
was completed in 1997. A photometric unit for flashing light, lux
second [lx·s], has been realized based on the NIST
detector-based candela using four standard photometers equipped
with current integrators. Two different approaches were taken to
calibrate these standard photometers: one based on electrical
calibration of the current integrator, and the other based on
electronic pulsing of the signal from a steady-state illuminance
standard. The units realized using these two independent methods
agreed to within 0.2 %. The relative expanded uncertainty (k=2) of
the standard photometers, in the measurement of white xenon flash,
is estimated to be 0.6 %. The standard photometers are
characterized for temporal response, linearity, and spectral
responsivity, to be used for measurement of xenon flash sources of
various waveforms and colors. Calibration services have been
established at NIST for flashing-light photometers with white and
red anticollision lights. See Ref. [7] for further
details.
5. Frequency-Dependent Photocurrent Measurements of
Flashing Lights
George Eppeldauer
Optical Technology Division
National Institute of Standards and Technology
USA
Two circuits for flashing light measurements were discussed
and compared. The first circuit has two stages. The first stage is
a current-to-voltage converter that includes the photodiode
sensor. This stage can be used to measure the frequency components
of flashing light if the sensor has fast enough responsivity. The
second stage is a voltage integrator with two controlled switches.
One switch shorts the integrating capacitor before the start and
the other one controls the timing of the integration (measurement)
cycle. In the second circuit, the sensor and the integrating
capacitor are combined in one stage. The fundamental gain
equations (current-to-voltage gain, loop gain, and closed-loop
voltage gain) of the two different measuring circuits are
determined for the frequency interval of the frequency components
of the flashed pulse to be measured. Accuracy requirements for the
photocurrent-to-voltage conversion were discussed for the interval
of all frequency harmonics of the flashing light pulse. The effect
of pulse length and amplitude of the flashing signal and the
capacitance of the photodiode sensor for the measurement
nonlinearity were also discussed briefly.
6.Conspicuity of Point-Sequential Light Signals Used to
Mark Emergency Vehicles
Jan Berkhout, Terry Dell, and Frank Schieber
Heimstra Human Factors Laboratories
University of South Dakota, USA
A high intensity discharge (HID) short-arc lamp coupled with a
fiber-optic distribution system interrupted by rotating dichroic
filter assemblies can provide rapidly alternating sequences of
intense light pulses of different colors at the focal point of a
single lens. Such signals have unique properties that strongly
influence the way they function as emergency vehicle markers. In a
point-sequential light signal (PSL), pulses of different colors
alternate without apparent temporal gaps at a single point. There
are interactions of psychophysical properties across the color
boundaries. These include sequential color and brightness effects
that are quite striking, and not predicted by any current theory.
These sequential signal interactions in turn affect the function
of the light as an emergency vehicle marker. Flashing signal
lights embody trade-offs of conspicuity and trackability. A highly
conspicuous flash pattern may be hard to track in traffic where
there is considerable visual background noise. Using
point-sequential lights (PSLs), it is possible to design signals
that are optimum for particular background and traffic
environments. PSL cycle times and pulse duration ratios can be
modified over a wide range of values. This initial series of
investigations studied flash patterns incorporating two colors
presented within a basic cycle time of 800 ms. The shorter
duration color was present for 3%, 17%, 36%, 62% or 100% of the
time occupied by the longer duration color. Perceptions of image
size, color saturation and apparent point of origin varied with
these ratios. A signal consisting of long red pulses interrupted
by short white pulses seemed to provide a good combination of
conspicuity and trackability. Photometric documentation of these
signals is complex, since the changes in intensity and color at
the pulse boundaries include a brief dark phase, and the light
pulse onsets and offsets are asymmetrical and nonlinear.
7. Photography and Flash Exposures
Pierce Webb
Eastman Kodak Co., USA
From the perspective of photographic film, essentially all
exposures are "flash exposures", at least in the sense of being
short duration. While xenon flash lamps are important "flash
exposure" sources, steady state light (like daylight) sources plus
some shuttering mechanism are more common. High precision in
measurement and control of these short duration exposures is
extremely important in the "flash exposing devices", the
"sensitometers", used by those who design and build new films,
where subtle differences in photographic properties between
different coating formulations, etc. are easily masked by
unintended variability in exposure sources. From a measurement
perspective, there is much similarity in the metrology/photometry
of flashes from xenon lamps and from flashes from a sensitometer.
Some spectrally weighted integrated quantity of light, its
spectral characteristics, and reproducibility of "flashes", from
time to time and from one source to another, are similar concerns.
For historical reasons, film sensitometry involves red, green and
blue spectral sensitivity of films (itself derived from "spectral
sensitometry") and measurements in photometric units.
"Photographic flash lamp photometry" should include either
spectral data or spectral weightings related to R, G, B film
spectral sensitivities. Even flash duration may be similar;
concern with "reciprocity effects" (different photographic
sensitivities from the same quantity of exposing light, delivered
over different exposure times) leads to sensitometer exposure
times of 10-5 s at Kodak, derived from cw light-source
sensitometers. Shorter times still, representative of laser
printing, are used in "laser sensitometers." There are obvious
differences as well: sensitometers generally have a low flash
repetition rate, usually have (shuttered or scanned) continuous
light sources (greater stability), and invoke no psychophysics.
Given the need for high precision in sensitometry in creating new
films, and the need to define exposures accurately to determine
ISO speeds accurately, the photographic industry's greatest need
in flash light photometry will be improved sensitometer photometry
and spectral radiometry.
DISCUSSIONS
After all the presentations, an open discussion session was
held for about 30 minutes with all the speakers as panelists. Some
of the questions and discussions are reported below.
There were several questions on multiple flashes. D. King
answered that there are a number of people in the airline industry
who believe that multiple flashes provide a higher level of
conspicuity than equally spaced single flashes. Schmid-Clausen
mentioned that double flashes are effective in that the first
flash directs your attention and the second flash gives precise
information of the position. Multiple flashes can be treated as
one flash by the form factor method (as given in the EC
regulation) within a certain time delay (within ª0.1s).
The blinking LED on the telephone that the chairman mentioned
has smaller effective intensity than steady-state LEDs, yet
produces a more effective signal than steady-state LEDs. Should we
have a system of photometry that tells us this effect?
Schmid-Clausen answered that it is a conspicuity field; with the
same intensity, flashing LED is much more conspicuous than steady
burning light.
What is the definition of pulse? If the emission is not zero
between the flashes, how can we treat this as pulse?
Schmid-Clausen answered that, based on his investigation, such
effect (pulse on steady light) can be described by adaptation
luminance for a certain luminance level. The time constant a
changes as adaptation luminance changes. However, it was not
investigated for various levels.
The Blondel-Rey equation requires the pulse shape to determine
t1 and t2. Then what is the purpose of integrating the pulse?
Should the measurement of only the time function i(t) suffice?
Dave King answered that, in case of xenon flash, the shape does
not matter because t2-t1 is very small and the effect is
negligible. Schmidt-Clausen answered that this was discussed in
the late 60s with the US Coast Guard. Durations other than t1, t2
can be chosen as approximations similar to the form factor method.
We know there are some gray zones and we cannot presently describe
all cases. But why not start describing the effect and see what
comes out of it.
F. Hengsteberger, Director of Div.2, mentioned that a guidance
is now urgently needed from Div.1 because there are already
applications out there and different formulae are widely used. We
should find a compromise between what visual science requires and
what is needed by practitioners. We have to give a useful starting
point with some initial recommendation.
K. Sagawa, Director of Div. 1, answered that he basically
agrees with having a joint work by Divs. 1 and 2 on this subject.
However he sees a difficulty in finding good visual scientists in
this area who can undertake this job. Research in this subject was
active in 1960s, but not recently. He encourages young scientists
to work in this area to contribute to CIE on this subject.
There were also questions and discussions on the effect of
spatial properties, e.g., of LED barricade lights, effect of
modulation by discharge lamps, and the effect of background
illumination on the perception of flashing lights, the issues on
modulation transfer function, equivalent luminance, measurement
errors due to ambient light, etc.
CONCLUSIONS
There seems to be a consensus among the participants that a
CIE document is urgently needed to define the effective intensity
and give recommendations on the photometry of flashing lights.
While the issues on physical measurements will be addressed in
TC2-49, the chairman requested again that Div. 1 consider
initiating work in this area. A great amount of information and
new knowledge have been presented and discussed in this Workshop.
The chairman thanked the speakers and the participants.
REFERENCES
[1] There are two related publications from CIE: (1)
CIE Pub.105 Spectroradiometry of Pulsed Optical Radiation Sources
(1993), which does not describe photometry, (2) Ref.[2],
which is a scientific paper and not a CIE recommendation.
[2] H. J. Schmidt-Clausen, A Comparison of Different
Methods for the Determination of the Effective Luminous Intensity
of Signal Lights in the Form of Multiple Pulses, CIE Journal, Vol.
1, No.1, 18-22 (1982).
[3] A. Blondel and J. Rey, Sur la perception des
lumières brèves à la limite de leur
portée, Journal de Physique, Vol.1, p. 530 (1911).
[4] H. J. Schmidt-Clausen, The influence of the
angular size, adaptation luminance, pulse shape, and light colour
on the Blondel-Rey constant a, The Perception and Application of
Flashing Lights, Adam Hilger Ltd, London, pp.94-111 (1971).
[5] H. J. Schmidt-Clausen, Investigation of the
Spectral Luminous Efficiency and the Signal process in the Human
Eye, paper presented at the EPRI, Lighting Research Office Fourth
International Lighting Research Symposium: Vision at Low Light
Levels, Orlando, Florida, May 19-21 (1998).
[6] SAE ARP 5029 Measurement Procedures for Strobe
Anticollision Lights, Aerospace Recommended Practice (1998).
[7] Y. Ohno and Y. Zong, "Establishment of the NIST
Flashing-Light Photometric Unit," Proc., SPIE, Vol. 3140,
Photometric Engineering of Sources and Systems, 2-11 (1997).
