Yoshi Ohno, Ph. D.

1
TRB 16th Biennial Symposium on Visibility and
Simulation Iowa City June 2-4 2002
Analysis on Effective Intensity of Flashing
Lights and Modification of Allard Method
Yoshi Ohno Ph. D. National Institute of
Standards and Technology USA 100 Bureau Drive
Stop 8442 Gaithersburg MD 20899-8442 Email
ohno_at_nist.gov Dennis Couzin Avery Dennison
6565 Howard Niles IL 60714 Email
dcouzin_at_stimsonite.com
2
Effective intensity (of a flashing light) Unit
candela (cd) Symbol for quantity Ie
Luminous intensity of a fixed (steady) light of
the same relative spectral distribution as the
flashing light which would have the same
luminous range (or visual range in aviation
terminology) as the flashing light under
identical conditions of observation.
ILV (CIE 17.4 / IEC 50(845) 11-18
3
Formulae for Effective Intensity of Flashing
Lights
1. Allard (1876) 2. Blondel-Rey (1911) 3.
Blondel-Rey-Douglas (1957) 4. Schmidt-Clausen
(Form-Factor Method) (1967)

• Results vary significantly between different
  methods for various waveforms of pulses
  including a train of pulses. (For very short
  flashes all methods give the same results.)

4
Blondel-Rey (1911)
5
Blondel-Rey-Douglas (1957)
Blondel-Rey (1911)
(extension of B-R for a train of pulses)
t1 t2 are determined to satisfy
IeI(t1)I(t2)
where IeI(t1)I(t2)  a0.2 s
This is solved as
It is solved as
6
Practical Aspects of Blondel-Rey and
Blondel-Rey-Douglas

• Require the waveforms of the pulses. Waveforms
  must be measured accurately.
• Require iterative solution thus need a computer.
  It cannot be realized by simple analog circuits.

Thus difficult to produce portable (hand-held)
flash photometers.
7
Form Factor method Schmidt-Clausen (1967)
Practical Aspects

• Calculation is much simpler than
  Blondel-Rey(-Douglas).
• Requires only the integral and the peak of the
  pulse. (Waveform not needed)
• Can be realized with analog circuits consisting
  of a current integrator and a peak-hold circuit.

(a 0.2 s)
This is transformed to
8
Allard (1876) rarely used
Practical Aspects

• Requires only the peak of the convolution.
  (Waveform not needed)
• The convolution can be achieved with a simple R-C
  filter circuit.

Instantaneous effective intensity i(t) is solved
by the equation
Ie is the maximum of i(t).
This is solved as

• Works through DC. Calibration can be done by
  using a steady-light luminous intensity standard.

q(t) visual impulse response function.
convolution
9
Standards concerning Effective Intensity

• IES Guide for Calculating the Effective Intensity
  of Flashing Signal Lights (1964) - B-R-Douglas
• IALA standard Recommendations on the
  determination of the luminous intensity of a
  marine aid-to-navigation light (1977)
• - B-R FF or Allard
• ITE Purchase Specification for Flashing and
  steady-burn Warning Lights (under revision) - B-R
• SAE ARP5029 Measurement Procedures for Strobe
  Anticollision Lights (1999) - B-R
• ECE Regulation No. 65 Uniform Provisions
  Concerning the Approval of Special Warning Lights
  for Motor Vehicles - FF
• ASTM standard on photometry of flashing lights -
  being developed by E12.11.WG05 - to recommend one
  best method
• CIE recommendation on photometry of flashing
  light- being developed by TC2-49 - to make
  international standardization

10
A problem with Form Factor Method
Raised by D. Couzin at CIE TC2-49 Meeting in
April 2000 in London.
Form Factor method clearly fails for this pulse.

• Allard and B-R do not have this problem.
• Allard formula has a physiological basis
  (convolution with a simplified visual impulse
  response function).
• Decided to re-examine the existing three methods.

11
Analysis on the Conventional Three Methods
To evaluate the differences between the three
conventional methods computation analyses were
made by calculating the effective intensity of 10
different waveforms of pulses with duration from
0.001 s to 100 s.
Imax
DT
Duration DT
DT
DT
12
Results 1
13
Results 2
14
Results 4
15
Results 5
16
Results 7
17
Ratio should be 1
Blondel-Rey-Douglas failing.
18
Results 8
Form Factor failing.
19
Results 9
20
Ratio should be 1
BRD and FF are failing.
21
Summary of Comparison

• Form Factor method fails for a pulse with a sharp
  peak added on.
• Both Form Factor and Blondel-Rey-Douglas do not
  work well for a train of short pulses.
• Blondel-Rey-Douglas do not work well for
  modulated pulses.
• Allard method does not show problems above.
• Allard results deviate significantly from
  Blondel-Rey results (higher by 30 for
  rectangular pulses in the 0.1 to 1 s region).

22
Modification of Allard Method

• Observation
• Blondel-Rey equation is believed to be accurate
  for rectangular pulses.

Approach
Modify the q(t) function so that Allard results
match Blondel-Rey for rectangular pulses but can
be realized with simple analog circuitry.
23
Modification of Allard Method
Analytical calculation revealed a q(t) that
perfectly matches Blondel-Rey for rectangular
pulses

This function however cannot be realized with
simple analog circuits.
The function q(t) is approximated using two
exponential functions
where
a10.113 w10.5 a20.869 w20.5
Example
24
Circuit Example for Modified Allard Method
C1R1 a1 C2R2 a2
25
Modified Allard Method
Results 1
26
Modified Allard Method
Results 4
27
Modified Allard Method
Results 5
28
Modified Allard Method
Results 9
29
Ratio should be 1
30
(No Transcript)
31
Conclusions

• Both Blondel-Rey and Form Factor methods are
  shown to fail for certain forms of pulses
  including a train of short pulses.
• Modified Allard method has been developed with a
  modified visual impulse response function q(t)
  that produces equivalent results to Blondel-Rey
  for rectangular pulses.
• Modified Allard method gives reasonable results
  for all forms of pulses including a train of
  pulses at any duration.
• Modified Allard method can be realized with
  simple analog circuits for hand-held photometers.
• This method is considered for standardization by
  ASTM E12 and to be discussed at CIE TC2-49.
  Experimental verification is needed.

32
ASTM E12.11.05 (Flashing lights)
http//cie2.nist.gov/ASTM_E12_WG05/home.html
CIE Div.1 Div.2 Joint Symposium on Temporal
and Spatial Aspects of Light and Colour
Perception and Measurement Veszprem Hungary
Aug. 22-23 2002
CIE Div.1 Div.2 meetings Aug. 24 - 28 CIE
Division 2 website http//cie2.nist.gov
33
(No Transcript)
34
Results 3
35
Results 10
36
Circuit Example 2 for Modified Allard Method
37
Ratio should be 1
38
Modified Allard Method
Results 7
39
Results 6