W REPORT NO 20S52-fcC02-RO-00
             ;NAt REF
   PORTABLE LASER VELOCIMETER
               FOR
 STACK VELOCITY  MEASUREMENTS
            NOVEMBER
L . 0 H E F c J N G E R 5

                 T T H £ W S  a n d H
             K r a p c r e ^ f c r
   E N V! R O N M 5 N T A L PROTtCTiON A G E N C Y
RESEARCH TRtANGL?  PARK NORTH  CAROLINA
             TRW
                                --. -

-------
                    TRW REPORT NO. 20852-6002-RO-OO
              FINAL REPORT

    PORTABLE LASER VELOCIMETER
                 FOR
  STACK VELOCITY MEASUREMENTS
              NOVEMBER 1972
               Prepared by
L.O. HEFLINGER, B.J. MATTHEWS, and H. SHELTON
               Prepared for
    ENVIRONMENTAL PROTECTION AGENCY
 RESEARCH TRIANGLE PARK, NORTH CAROLINA
               SYSTfMS GROUP
 ONE SPACE PARK • REDONDO B E*A CH • CALIFORNIA

-------
                               CONTENTS
                                                                    Page
1.   INTRODUCTION AND SUMMARY 	    1
2.   PORTABLE LASER VELOCIMETER 	    3
    2.1   Background 	    3
    2.2   Velocimeter Operating Principle 	    5
    2.3   Velocimeter Optical System 	    6
    2.4   Velocimeter Electronics 	    13
    2.5   Velocimeter Testing 	    22
3.   LONG RANGE OPTICAL VELOCITY METERS 	    28
    3.1   Introduction 	    28
    3.2   Smoke Optical Properties	    31
    3.3   Sampl e Vol ume Si ze 	    32
    3.4   Significance of the Smoke Optical
          Properties in Conjunction with the
          Sample Volume Size 	    35
    3.5   Shot Noise of Detection 	    36
    3.6   The Direct Doppler System 	    38
    3.7   The Fringe Velocimeter:  Granularity Noise 	    42
    3.8   Fringe Velocimeter:  Required Laser Power 	    48
    3.9   Reticle Velocimeter Noise Estimate 	    51
    3.10  Preliminary Experiments with a
          Reticle Velocimeter 	    53
4.   CONCLUSIONS 	    55
    4.1   Portable Laser Velocimeter  	    55
    4.2   Long Range Velocimeters 	    55
REFERENCES 	    57
APPENDICES
    A     SPECIFICATIONS FOR SPECTRA-PHYSICS MODEL
          120 GAS LASER WITH MODEL 256 EXCITER 	   58
    B     LASER VELOCIMETER DRAWING LIST 	   59

-------
                               ILLUSTRATIONS

                                                                       Page
 1.   Velocimeter electronics  cabinet  and  sensing head  	    1

 2.   Schematic of the interference  between  two
     coherent light beams  	    4

 3.   Schematic diagram of  laser velocimeter optical
     sy s tern 	    7

 4.   Laser velocimeter sensing  head assembly  	   10

 5.   Photograph of velocimeter  sensing  head showing
     component piacement 	   11

 6.   Photograph of velocimeter  sensing  head showing
     framework and controls  	   12

 7.   Three-quarter view of sensing  head 	   13

 8.   Oscilloscope traces of  PMT signals at  the  output
     of the bandpass amplifier  in the breadboard
     velocimeter 	   15

 9.   Oscilloscope traces of  spectrum  analyzer signals
     obtained with the breadboard velocimeter 	   17

10.   Electronic circuit diagram for the laser velocimeter  	   19

11.   Laser velocimeter sensing  head 	   23

12.   Velocimeter electronics  package  	   23

13.   Plot of measured versus  calculated velocity for
     laser velocimeter 	   24

14.   Air and predicted particle (droplet) velocity 	   27

15.   Direct doppler velocity  meter	   28

16.   Schematic of fringe velocimeter  	   29

17.   Reticle velocimeter 	   30

18.   Beam intersection geometry 	   33

19.   Effect of turbulence  on  direct doppler system 	   38

20.   Granularity noise in  the fringe  velocimeter  	   42

-------
                                FOREWORD
     This report summarizes work done under Environmental  Protection
Agency Contract No. 68-02-0308, dated 30 June 1971.   Activities  under
this contract were concerned with:   (1)  design and construction  of a
portable laser velocimeter for stack velocity measurements at close
range; and (2) analysis of optical  velocity measurement techniques
over ranges up to 1500 feet.

     Technical direction and administration of this  project by the
Environmental Protection Agency was provided by Dr.  Frederic C.  Jaye.
The TRW Project Manager was Mr. Birch J. Matthews.  Drs. Haywood She!ton
and Lee 0. Heflinger developed the laser velocimeter instrument and the
analysis for extended range use of optical velocity meters, respectively.

     The authors wish to thank Mr.  William F. Daley for his valuable
assistance in the design and fabrication of the laser velocimeter.
Mr. Robert F. Kemp also made significant contributions to the initial
design concepts for the optical and electronic systems.

-------
                      1.   INTRODUCTION  AND SUMMARY
     This report describes the design-development  of a  short  range
laser velocimeter together with an analysis  of potential  extended range
optical velocity measuring techniques.   The  work was done under contract
to the Environmental  Protection Agency.

     The laser velocimeter constructed during this program operates on an
interference-backscatter principle.  It is portable and designed to measure
velocities of flowing particle laden gases in power plant ducts or  stacks
at relatively close ranges.  The instrument is composed of a  sensing head
cabinet and an electronics cabinet.  These are pictured in Figure 1.  The
           Figure  1.   Velocimeter  electronics  cabinet  (1) and
                      sensing  head (r).
 sensing  head incorporates  a  laser illuminator,  beam  forming  optics  and  photo-
 detector.   The  electronics cabinet contains  a signal  processing  circuit,
 spectrum analyzer,  as  well as  power supplies for  the  laser and spectrum
 analyzer.   The  spectrum  analyzer is used  to  analyze  the  frequency signals
 generated  by the  sensing head.   Following are design  and operating  charac-
 teristics  for this  laser velocimeter:

-------
     •    Velocity sensing range
     t    Focal length range
     •    Measuring medium
     §    Sensing head weight
     •    Sensing head envelope
     •    Electronics weight
     •    Electronics envelope
     •    Readout   (Meter)

     •    Readout (meter)

     •    Power requirements
10 to 125 ft/sec
 2 to 15 ft
Fossil fuel stack gas
25 Ib
27-3/4 x 11-3/4 x 6-7/8 inches
25 Ib
12-1/2 x 16 x 7-1/4 inches
Mean velocity, ft/sec
20 sec.  time constant
Velocity deviation, ft/sec
5 sec time constant
llOv, 60 Hz,  <5  amp
     The design, construction and preliminary testing of this
velocimeter are discussed in Section 2.

     The second major program task was an analysis of optical  velocity
measuring devices for extended range operation.   Of interest here was
remote stack plume velocity measurements over distances  of 300 to
1500 feet.  Three types of optical velocimeters were considered;
namely, Doppler shift, interference-fringe,  and a reticle velocimeter.
This discussion is presented in Section  3.
  0-100 mv  analog signal  available.

-------
                    2.   PORTABLE LASER  VELOCIMETER
2.1  BACKGROUND
     When two coherent beams of light intersect in  space,  stationary
three-dimensional interference fringe patterns  are  established.   The
fringe spacing A is a function of the wavelength of light A  and the
angle Q  at which the two beams intersect.   This is shown  in Figure  2.
In this illustration, both beams are parallel beams of light,  as  though
from point sources at infinity.  Each is traveling  at the  velocity of
light C, each is of wavelsngth A, and both are  coherent to each
other.  The two beams of light pass through each other at  an angle  Q .
Since light is a transverse electromagnetic phenomenon, the direction
of vibration of the electric component is perpendicular to the plane
of the drawing.  For plane waves, the electric  intensity varies
sinusoidally in strength as one moves down the  wave train.

     The wavelength A is the physical separation between adjacent
maxima of the traveling electric field.  When  the maxima of one wave
are superimposed on the maxima of another, the  amplitudes  add,
constructively.  When the maxima of one wave are superimposed with the
minima of the other, the two destructively interfere.  If  the wavefronts
are of equal amplitude, the net amplitude is zero at these points.  The
loci of constructive interference trace out the straight lines shown
in Figure 2.  In the case of beams of finite cross  section, planes are
actually traced out.  These planes, or alternate lines of  darkness and
light, are parallel to the bisector of the angle 8  between the direction
of propagation of the two beams.  The regions  of destructive interference
form periodic planes of absolute darkness.  In  Figure 2, the loci of
constructive interference are indicated by the parallel lines  (regions
of light).  Destructive interferences are the blank regions (no light)

-------
EXP j
 EXP j
                                                                                            FIRST TRAVELING
                                                                                            WAVE OF WAVELENGTH \
STATIONARY LOCI OF
CONSTRUCTIVE
INTERFERENCE
                                                                                  SECOND TRAVELING WAVE
                                                                                  OF WAVELENGTH \
            Figure 2.   Schematic  of  the  interference between two traveling collimated  coherent
                       beams  of light of wavelength x passing through each other at an angle of e.

-------
 in between.  Neighboring dark or light regions are separated by a distance
                            A =
                                 2 sin  e/2
As an example, when the two beams  pass  through  one  another  at  an  angle
of 60 degrees, the loci of constructive interference are  separated  from
each other by the wavelength of light.   Deep red light has  a wavelength
of 0.7 micron or 27.5 microinches.

2.2  VELOCIMETER OPERATING PRINCIPLE
     The laser velocimeter described subsequently utilizes  the wave
interference properties of two coherent light beams described  in  the
previous section.  Two beams (<1 mm diameter) of coherent light derived
from a helium-neon gas laser (x=0.6328^) intersect in space establishing
an interference fringe pattern.  The beams intersect at an  angle  of
approximately 1 degree (1/60 radian).   Horizontal fringe  spacing  is
0.0015 inch in an intersection volume about 1/2 mm wide and 1  cm long.

     Particulate laden combustion gas passes through the  region of  inter-
ference.  A particle moving through adjacent fringes of constant  separa-
tion will scatter light periodically.   If this  scattered light is collected
and focused onto a photodetector,  it will produce a coincident pulsating
electric current whose frequency is proportional to the particle's  velocity.
In the present velocimeter system, the frequency of light modulation
ranges from 80 Kc to 1 Me corresponding to a velocity range of 10 to 125
ft/sec.

     The output of the photodetector is fed into a spectrum analyzer
which is a sensitive receiver that can be electronically swept from
near zero frequency to above 1 megacycle.  The input frequency is
indicated by the value of the sweep voltage at the instant an  output
signal is obtained.  The value of this voltage is stored and displayed
on a meter and read as the proportional velocity.  The velocimeter

-------
 electronic  circuit will  search for a lock on any frequency of
 sufficient  amplitude which  is present.

 2.3  VELOCIMETER OPTICAL SYSTEM
      The  laser  velocimeter  is a single station instrument requiring
 only  one  port or opening in the power plant stack.  The signals received
 by  the  optical  system are those generated by low angle particulate
 scattering  in the backward  direction.

      The  optical system  is  self-aligning which automatically focuses
 the two coherent light beams to a crossover point in space.  The same
 optics  collect  the backscatter and focus  it  onto  a photodetector.   A
 schematic of the velocimeter optical arrangement is shown in Figure 3.

      The  illumination source is a Spectra-Physics helium-neon gas laser
 which emits 5 mw power at x =0.6328/a(red light)*.   Specifications for
 this  laser  are  provided  in Appendix A.

      This laser was selected on the basis of reliability, plasma tube
 life, ruggedness and maximum power output consistent with size and weight
 limitations.

      From the schematic of Figure 3, it will be seen that the optical
 path  is folded to obtain a compact design.  Collimated light from the
 continuous wave helium-neon laser is incident on  an  adjustable-beam
*
  Spectra-Physics Model 120 Gas Laser with Model 256 Exciter.   Product
  of Spectra-Physics, Inc., 1250 West Middlefield Road, Mountain View,
  California.

-------
                                       SLOTTED FIRST
                                       SURFACE MIRROR
                                               CONVERGING LENS,
                                               f. I.  = 2.625 IN.
2 SIN (9/2)

 \  SOLID STATE
    PHOTODIODE
FIRST SURFACE
MIRROR
ADJUSTABLE 240 MM f.l.
ACHROMAT FOCUSING AND
COLLECTING LENS
                                        HELIUM - NEON
                                        GAS LASER
                                       ( X = .6328   )
                                              ADJUSTABLE
                                              BEAM SPLITTER
 Figure 3.  Schematic diagram of laser velocimeter optical system.

-------
 splitter.   The  beam  splitter  consists of back-to-back 90-degree prisms.
 The two-prism interface  transmits  50 percent of the beam onto a small
 first surface mirror mounted  at  45 degrees  to  the incident beam.  The
 coated prism interface reflects  the other 50 percent of the primary beam
 parallel  to the transmitted beam.   The prism assembly and small first
 surface mirror  move  in pure translation with respect to each other about
 a  common  centerline  thus  providing adjustable  and parallel beam
 separation.   Beam separation  ranges from a  minimum of 0.062 in. to 0.80 in.
 This  range  of beam separation allows a constant separation of the
 stationary  fringe pattern in  space of approximately 40 microns.

      After  division,  the  parallel  beams are turned 90 degrees by a second
 first surface mirror and made parallel to the  optical axis of the tele-
 scope.  The  beams are converged  to a focal  point by a 2.625 in focal
 length  lens.  After  crossover, the beams diverge and pass through a
 0.062 in.  split first surface mirror assembly. The split mirror passes
 the diverging beams  through to a focusing 240  mm focal length achromat
 lens.   This  4.332-inch diameter  lens causes the two parallel beams to
 converge in  space.   The achromat lens telescopes through a distance of
 5.50  inches  providing a beam  crossover location in space from 2 to 15 feet
 along the optical axis.   Movement  of the achromat lens holder is
 accomplished with a  rack and pinion gear assembly.  Fringe spacing is
 held  constant by manually adjusting the lens travel and beam splitter
 to previously calibrated positions.

      Backscattered light by particulate in the flowing gas passing through
 the stationary interference pattern is collected by the achromat lens and
 reflected by the split first surface mirror assembly.   This mirror, mounted
 at 45 degrees to the optical axis, allows the scattered light to be focused
at an aperture located before a photodetector.   With the exception of the
 traveling achromat lens,  all components of the optical system are precisely
and rigidly mounted to insure that the scattered light is brought to focus
before the photodetector.
                                    8

-------
     Figure 4 is a layout drawing of the laser velocimeter sensing  head.
This drawing shows the placement of the gas laser chassis  and critical
optical components.  A tabulation of the sensing head detail  drawings  is
presented in Appendix B.  The photographs in Figures 5 through 7 illustrate
the basic construction and optical component placement of  the sensing  head.
With the exception of the traveling objective lens, all  of the beam forming
optics are mounted on one rigid platform to preserve alignment.   The align-
ment of the objective lens is not as critical because the  system is self-
aligning with respect to its movement.

     In adapting the laser to the sensing head, the outer  case was
removed (to save weight) and mounting holes machined in the laser chassis.*
The front end of the laser is rigidly mounted to the optical  platform.
A single point pin mount parallel to the optical axis is used at the other
end to permit thermal expansion of the laser chassis.

     The laser and critical beam forming components on the optical
platform are mounted to a 1/2 inch aluminum angle frame at three points.
An aluminum sheet metal cover enclosing the optics is attached to the
frame.  The three point tie-down of the optics is designed to minimize
any optical path misalignment due to external forces applied to the
frame or cover.

     The basic design concept for the sensing head optical system was
predetermined by the requirement for an interference-backscatter instrument.
This being the case, sensing head design was limited to compact packaging
and maintenance of optical path considerations.  More attention, however,
was given to selection of a suitable photodetector and design of signal
processing electronics.  These subjects are discussed in the section
following.
* TRW Drawing No. SK 1038-12.

-------
                               . CABLE STOWAGE HOOKS .
             SLOTTED FIRST
             SURFACE MIRROR
                                                                         TRAVELING OBJECTIVE LENS
FIRST SURFACE
MIRROR	,
OBJECTIVE LENS
FOCUSING CONTROL
KNOB —v
                — BEAM SPLITTER
                  ADJUSTMENT
                 SCONTROL KNOB
                                   70cm
                                                                                                 29 ,
                                  Figure  4.   Laser  velocimeter sensing  head assembly.

-------
Beam Splitter
First Surface Mirror
                                                      Gas  Laser
                                                Traveling Objective Lens
                                        Slotted First Surface Mirror
                                  2.625 In Focal  Length Lens
 Figure 5.  Photograph of the laser velocimeter sensing  head  assembly
            showing the placement of the various components.
                                     11

-------
                         •Optical Platform
                                               Traveling  Lens Adjusting Knob
                                                      Traveling Lens Mount
                                1/2 In Angle Frame
                         Beam Splitter Adjusting
                         Knob
Figure 6.   Photograph of the other side of the sensing head assembly
           showing the location of the two control  knobs used to focus
           the adjustable beam splitter and traveling objective lens.
                                   12

-------
Figure 7.  Three-quarter view of the sensing head.   The focusing lens
           and cylindrical housing are shown in the foreground of the
           photograph.
2.4  VELOCIMETER ELECTRONICS
     In describing the optical  system, it was noted that an adjustable
beam splitter is used to maintain constant fringe spacing within the
limits of the objective lens travel.  A fixed beam splitter arrange-
ment was considered.  Such an arrangement (variable fringe spacing),
results in excessively high frequency signals at short range; extended
sampling (beam crossover) volumes at long range; and, more sophisticated
signal processing electronics.   For these reasons, the fixed beam
splitter design approach was not adopted in the present design.

     Initial work on the velocimeter electronics centered on selection
of a photodetector and signal processing system.  A survey of readily
available photomultiplier tubes (PMT) and solid state diodes was made.
A type 4473 PMT and a type MRD 500 silicon photo diode were selected
as candidates.  The type 4473 PMT is structurally similar to types
931, 1P28 and 1P21 and is, in effect, a 1P21 selected for maximum
"red" sensitivity.  The MRD 500 photo diode has a spectral range
                                   13

-------
 throughout  the  visible  and near-infrared.*  Peak sensitivity for this
 diode  is  about  0.8 micron.  Response time is typically <1 nanosecond.

     A  laboratory breadboard velocimeter system was set-up for the
 purpose of  generating typical electrical signals.  This set-up consisted
 of  a 3  mw helium-neon gas laser, a beam splitter, simple positive lens,
 a slotted first surface mirror, pinhole and PMT assembly, and a 10-inch
 focal  length, f/5 achromatic objective lens.  The optical arrangement
 was, in concept, the same as that shown in Figure 3.  A photomultiplier
 power  supply and bandpass amplifier were used with the PMT.

     A  moving target (smoke simulator) was devised by cutting a circular
 groove  in a plexiglas disk.  A mixture of flyash and coil dope was poured
 into the  groove and allowed to dry.  When rotated by a small motor, the
 disk made a useful target for the velocimeter and was capable of
 generating known constant or variable velocities within the design
 range of  the instrument.

     The  oscilloscope photographs of Figure 8 show traces of a 1P28 PMT
 signal  at the output of the bandpass amplifier in the breadboard
 velocimeter system.  These were obtained from the spinning disk.  At a
 sweep speed of  5 msec/div, Figure 8 shows a large number of spikes,
 resulting from  point-to-point differences in scattering properties or
 in  the  number of scattering particles in the laser beam at any one time.
 The density of  particles on the rotating track was intended to represent
 a relatively light loading (approximately 1  grain/ft ) in a gas stream.

     In Figure  8a, the spikes are nearly symmetrical about the zero
axis.   The bandpass  amplifier used  transmitted  frequencies  between
  Product of Motorola Semiconductor Products, Inc., Type
  MRD 500 P-I-N Silicon Photo Diode.
                                   14

-------
                                   (b)
Figure 8.  Photomultiplier tube signals at the output of the bandpass
           amplifier in the breadboard laser velocimeter.   The upper
           figure is at a sweep rate of 5 msec/division and shows
           scattering from clumps of particles.   The lower figure gives
           an indication of the autocorrelation of the signal  at the
           frequency of the velocity-generated signal (about 140 kHz).
about 30 kHz and 600 kHz.  At higher sweep speeds,  most of the spikes
are seen to contain strong component signals at the laser doppler
frequency corresponding to the velocity of the track.   Figure 8b is a
composite of many such traces taken with the oscilloscope trigger level
set to show synchronization with the velocity signal  at about 140 kHz.
After 9 or 10 cycles, the signal appears to merge into the statistical
noise in the phototube signal; however, close inspection of the original
photograph revealed a faint but persistent component of this signal
throughout the trace.  The task of detecting this signal, therefore,
becomes one of deriving an intermittent signal out of a comparable
amount of noise.
                                    15

-------
      Following these preliminary'breadboard tests, consideration was
next given to the means by which the photodetector signal would be
processed.  A wide variety  of signal processor systems have been used
with laser doppler velocimeters.  Of these, three types were used with
the breadboard optical system.  Attempts were made to measure the doppler
frequency directly with frequency counters and also with a pulse-averaging
wide-band discriminator.  An attempt was also made to use the frequency
counter in a period-averaging mode.  Finally, the signals were examined
with a spectrum analyzer. . It quickly became clear that for the first
of these methods to work, considerable pre-processing of the signals
would be required ahead of the counter or discriminator.

     A basic problem with the direct-measuring methods is the matter
of signal continuity.  If the velocity signal were completely continuous,
the appropriate readout .device would most nearly resemble a frequency-
modulation radio receiver having an indicating meter instead of the audio
circuits.  With discontinuous signals, simple burst counting methods will
work, provided bursts are of sufficient length to be processed after they
are recognized.  A fairly elaborate burst pre-processor for use with
                                                                  •\^
sparsely seeded gas streams has been described by Lennert, et. al.  The
spectrum analyzer, on the other hand, processes all signals, and leaves
the problem of recognition to be dealt with as a subsequent step.
                                  *
     A wide-band spectrum analyzer  shows that the signal from the
PMT (Figure 8) contains a nearly constant level of .noise within the
bandpass of the instrument together with a recognizable signal at the
laser doppler frequency.  Figure 9 is a photograph of a (selected) single
sweep of the spectrum analyzer through the range of 8 kHz to'about
300 kHz showing a recognizable signaTat about 140 kHz.  All sweeps do
*
  Tektronix Type 1L5, Plug-In Unit.
                                    16

-------
                                       (a)
                                      (b)
                                      (0
jure  9.     Spectrum analyzer signals obtained with  the breadboard velocimeter.  The
           upper  trace  is  a single  sweep showing an especially strong pulse.  The
           other  traces  are composites of about 250 sweeps each showing peaks at
           140  kHz  and  210 kHz,  respectively.

                                      17

-------
not have the same signal to noise ratio; a strong signal such as this
represents a coincidence between the spectrum analyzer local oscillator
and the passage of a particularly reflective particle through the laser
beam intersection volume.  Figures 9b and 9c are composites of about
250 such traces each with the rotational speed of the target wheel set
to generate signals of 140 kHz and 210 kHz, respectively. 'Nearly all
of these traces produce pulses which can be recognized by a level
detector set, for example, at the second major division from the bottom.

     These results suggested that a velocimeter instrument'having the
greatest versatility could best be made by using a spectrum analyzer as
the primary signal processor.  It would, in effect, continuously scan
the range of the instrument (or a selected portion of the total  range)
for recognizable signals.

     A silicon photo diode was next acquired and evaluated.  Use of a
solid state photodetector offered several potential advantages.   The
efficiency of the MRD 500 diode i.n converting incident photons (at the
laser wavelength =0.6328ju) to electrons is near unity.  Correspondingly,
the quantum efficiencies for PMT's is typically 2 to 3 percent.   Further,
the diode offered the desirable feature of being compact and rugged.
And finally, the signal-to-noise ratio for the photo diode and photo
diode circuit appeared superior in testing (at  X=0.6328/x) to
comparable PMT systems.

     These arguments led to design of a photo diode circuit and
incorporation of a Tektronix 1L5 spectrum analyzer for signal  processing.
The resulting electronics circuit is shown in the Figure lO^schematic.

     The Motorola photo diode has a back leakage current proportional
to the light falling on it.   Its current flows through a 47 kn  register
and a bipassed monitor resistance (so the current can be peaked  for
                                   18

-------
      TRIANGULAR SWEEP GENERATOR
                                   5W. OUT. AMP.
                                       *A/V
                                       82K
     SEARCH  '
+I5V IK > RECORDER
                                     50FT/S f 150 FT/5
I   I (WHITE)
I   ? 560 PHOTO
I   ?    DIODE      I
             4417   I
             FET
                                                             L5 TEKTRONIX SPECTRUM ANALYSER
                                            MODIFICATION: JUMPER Bl TO NO.. 7; PULL APART JUNCTIONS BJ, BD, BK, BG
                                                                                         LOOSE COUPLING
                                                                                         TO Its
                                                                                    '.! 7,3pV
                                                                                    •i'ZENERS
                                                                                                TYPICAL MC 14336
                                                                                                COMPENSATION
                                                            + 15V      -I5V .
                                          4 - HEP 244 - (ANY DIODES OF SUFFICIENT PIV)
                             Figure  10.   Electronic circuit diagram for  the laser velocimeter.

-------
alignment purposes).  This resistance and the shunt capacity (~4 pf)
give a 3 db high frequency attenuating at about 1 me.  The voltage
across this resistor is led to the gate of a 2N4417 Field-effect
transistor operating as a source-follower to drive the capacity of the
coaxial which leads to a 3:1 step up transformer.  This series capacity
and shunt inductance give a low frequency cut-off at about 60 kc.  High
frequency response is about 1 me with the input capacity of the spectrum
analyzer and cable.

     The 1L5 spectrum analyzer is modified as shown on the Figure 10
schematic so as to require least power, to functionally eliminate many
of the analyzer control knobs and assure stable calibration.  It is
powered by a light weight power supply (a kilohertz torroid core)
which also supplies plus and minus 15 volts for the photo diode, FET,
and operational amplifiers in the electronic sweep and sample circuits.
Cord line dropping resistors are used to reduce the heat load in the
box and increase the reliability of the transistors in the supply. A
Motorola HEP bridge and large filter capacities supply a voltage that
is regulated to about 75 volts.  This voltage drives a DC-DC converter
using high voltage HEP transistors (in parallel to provide current
rating).  The inverter is a high efficiency (special square loop tape-
wound core) saturating square-wave oscillator.  Multiple air-cooled
secondary windings supply the voltage to full-wave bridges with a
single output capacity.  The 225 volts is derived from the 350 volts
and regulated by Zener diodes.

     The read-out circuits all use the identical integrated circuit
chip (MC1433G)  for simplicity of stocking spare parts.  One chip forms
a triangular (not a sawtooth) sweep.   This is accomplished by feeding
the integral  of the snapping saturated square wave output into the
inverting input and into the sweep output amplifier which supplies
current, to charge the sampling capacitor (whose voltage is proportional
                                  20

-------
to the velocity and drives the output meter through  an  amplifier).  The
amplitude of the sweep (always the same dv/dt)  and  its  mean  position  is
controlled by the SPOT search-lock switch  which changes the
regenerative feedback by 5 and also feeds  the meter  output voltage  into
the circuit so that on "lock", the sweep circuit sweeps +_ 10%  about the
signal.  When a signal is received from the spectrum analyzer  it is
amplified and if large enough, triggers a  threshold  detector which
charges up the sampling capacity to the voltage value of the sweep  at
that instant.  A meter then reads the corresponding  velocity.   If the
signal is lost, the voltage on the capacity will drift towards zero
by the input current of the meter amplifier.  (This  capacitor  can be
recharged to full scale by pressing a reset button).

     The required velocity reading range is 10 to 125 ft/sec.   To
accommodate this range and provide reading accuracy  at lower velocity
conditions, two scales are provided on the one meter; namely,  0 to  50
ft/sec and 0 to 150 ft/sec.  The lower scale is provided by  feeding
only a third of the sweep voltage to the spectrum analyzer.  Precision
resistors provide the accurate negative offset to calibrate  the low
scale, and the constant impedance, 3:1 attenuator to match scales.

     A second readout requirement is a measure of the velocity
deviation (AV) about the mean value.  A peak-to-peak voltmeter allows
this information to be read on the velocity meter.   In reality, two
meters are provided on the velocimeter.  One is located on the sensing
head.  The other is on the electronics cabinet.  Values of AV  (as well
as mean velocity) are obtained from this remote meter depending upon
                           *
a selector switch position.   Because of the finite number of fringes,
a width to the frequency spectrum exists and so even a single velocity
will show some AV.
*
  This meter reads 10 XAV.
                                   21

-------
  2.5  VELOCIMETER TESTING
       The portable sensing head and  electronics  cabinet which  comprise
  the laser velocimeter are seen in  the  photographs  of  Figures  11 and 12,
  respectively.*  Final  testing was  conducted with this apparatus at the
  conclusion of the program to verify basic operating characteristics.
  This was done qualitatively with flyash  dust  and water sprays, and
  quantitatively using water droplets and  a spinning disc.

       Calibration of  the  velocimeter was  accomplished using a spinning
  disc in  the  following  manner.   An 18-inch  diameter smooth white paper
  disc backed  by 1/8-inch  thick  acrylic  sheet, was installed on the
  shaft of an  electric motor.  The nominal  motor  speed was 1800 rpm.

       The velocimeter was  set-up before the disc with the optical axis
  approximately normal  to  the plane of the paper.  The objective lens
  focus  was  adjusted until  the two incident laser beams merged into
  a  single spot.   The  distance between the target disc and entrance
  aperture on  the  instrument was  measured  and compared with the
  indicated  objective  lens  range  scale reading.   The beam splitter was
  next adjusted to an  equivalent  reading.   The distance (radius) from
  the  motor  shaft  center!ine to  the center of the laser beam spot was
  measured before  operating  the motor.
       Disc  rotational  velocity  (shaft speed) was next measured using a
            **
  Strobotac.   The  velocimeter was  allowed  to sweep  through the full
  frequency  range until a signal  "lock" was obtained at the lowest
  spectrum analyzer sensitivity  setting.  Velocity readings were made
  on  both the high  and  low  range  meter scales.  The  remote (electronics
  cabinet) meter reading was  compared with  the sensing head meter.  This
  procedure  was repeated at different target distances to verify the
  accuracy of the range scale on  the focusing objective lens.
   The operation and maintenance of this equipment is described in
   Reference 2.
**Product of General Radio Co., Cambridge, Massachusetts.  Strobotac
   Type 631-VL, S/N 21705.
                                    22

-------
Figure 11.   Laser velocimeter sensing head.
                              Figure 12.  Velocimeter
                                          electronics package.
                         23

-------
      After these preliminary tests, the velocimeter was set up a distance
 of  10 feet from the target.  A series of tests was made at different radii
 to  check  the velocimeter performance over the full velocity range.  These
 data  are  plotted in Figure 13.  An analysis of the data indicates a stan-
 dard  velocimeter measurement error of ±1.8 ft/sec over the range of 10 to
 125 ft/sec.   It is estimated that most of the velocity deviation can be
 attributed to small  errors in reading the observed measurements.  Some dif-
 ficulty with the test setup was encountered in making measurements at the
 extreme high and low end of the velocity range.  As a result, some apparent
 data  scatter is seen at both ends of the curve in Figure 13.
i—
u
O
                                          METER
                                          SCALE
                                          RANGE
          SPECTRUM ANALYZER
          SENSITIVITY SETTING
                                                       0.005 V/CM
                                                       0.002 V/CM
                                                       0.001 V CM
                                                       0.002 V/CM
                                                       0.001 V CM
LOW
HIGH
HIGH
LOW
HIGH
                                  ALL DATA RECORDED AT 10 FT RANGE
                        40    50   60   70   80   90   100

                           CALCULATED VELOCITY, VG IN FT  SEC

          Figure 13.   Plot of measured versus calculated  velocity
                      for laser velocimeter.  Target was  18-inch-
                      diameter spinning, white  paper disc.
                                     24

-------
     Additional  laboratory tests  were  conducted  using  first flyash dust
and later, atomized water sprays.   The flyash  tests were unsatisfactory
in that the dust generator did not  produce  a uniform discharge.  Rather,
the particulate tended to clump together resulting  in  very heavy loadings
resembling mostly dense clouds.
     A commercial atomizing nozzle  was substituted  for the dust generator,
with water droplets providing the scattering medium.   The  water spray
proved more satisfactory in that  the water  flow  rates  could  be varied by
changes in feed pressure.  Velocity measurements up to about  40 ft/sec
were recorded by the velocimeter  at ranges  of  2  to  5 feet.
     A velocimeter test was next  made  using a  TRW low  speed  wind tunnel.
The two-dimensional test section  has a 5- x 24-inch cross section which then
diverges linearly to an 8- x 24-inch cross  section.  Flow is  directed ver-
tically down through the 5-foot-long test section.  Four turbulence  damping
screens and an aluminum honeycomb section are  positioned just upstream of
the tunnel's stagnation chamber.   A tunnel  contraction ratio  of 8:1  further
reduces turbulence intensities.
     Both pitot tube and anemometer velocity measurements are used to
measure test section velocities.   The  constant temperature anemometer is a
TSI Model 1050 single tungsten wire probe.  A  TSI linearizer is used to
provide analog output signals proportional  to  the velocity.
     For the velocimeter test, the tunnel was  operated at a  constant speed
core velocity of 58.7 ft/sec at the test section entrance.   Air flow in  the
test section has a flat velocity distribution  with a  boundary layer  veloc-
ity equal to 90 percent of the core velocity  at a distance  of 1/2  inch  from
the wall.
     A water spray nozzle was installed at the entrance to  the tunnel  test
section.  Injection of a water spray into the  tunnel  air stream  provided
scattering centers for the laser velocimeter.   These  scatterers  were intend-
ed to simulate flyash particles in a stationary power  plant  stack  or duct.*
 Introduction of flyash or chemical aerosols into the closed-loop tunnel
 was rejected for reasons of tunnel contamination.
                                    25

-------
                                                               r)
     The nozzle was operated at a pressure drop (AP)  of  30 Ib/in   (limited
by the available supply pressure).   At this AP,  the  injection velocity V.
was calculated from the relationship V. = CD[2gAP/p] '  , where: CD is the
orifice coefficient (assumed to be  ~0.75); g  is  the  gravitational constant
                      2                                            3
equal to 32.174 ft/sec ; and p is the density  of  water at 62.4 Ib/ft  .  The
injection velocity V. was determined to be 50.06  ft/sec.
                    J
     Water droplet velocities as a  function of distance  from  the point of
injection into the air stream were  also estimated.  A mean droplet diameter
was calculated from the Ingebo correlation .   This  empirical  relationship
yields a volume-number-mean droplet diameter D™  expressed by the  following
equation:

                D_./u^^,  = s. nu  /  ii  u    -t  n  n~> n   i\i      ., \            i-,\
where the water jet (orifice) diameter D. and droplet diameter D30 have
units in inches, and the air velocity V  .  and water jet velocity V.  have
units of ft/sec.  In this instance, the injector orifice diameters were
0.040 inch and the wind tunnel air velocity was 58.7 ft/sec in the 5- x 24-
inch cross section.  From Equation (1) the mean droplet diameter was  cal-
                    _3
culated to be 9.8x10   inch or approximately 250 microns.
     Velocimeter droplet velocity measurements were made at axial  distances
of 2.3 and 4.6 feet downstream of the water spray injection point in  the
tunnel divergence section at a range of 6 feet.  Average droplet velocities
of 35 and 40 ft/sec were measured at the two locations.  These measurements
correspond to air velocities of 45.2 and 36.8 ft/sec, respectively.   Veloc-
imeter measurements at distances further downstream of the spray injection
and at greater range were prohibited due to physical limitations of the
wind tunnel test setup.
     The apparent discrepancy between air velocity measurements and veloci-
meter droplet velocity measurements was subsequently analyzed.  An expres-
sion was derived to estimate the average particle (droplet) velocity  V"
as a function of distance X from the point of injection into the tunnel  air
stream.   This derived relationship is
                                    26

-------
        pf
where:
              30
and subscript f
                     0.92x10
          3
                       '-V
                                                                    (2)
   volume-number-mean  droplet diameter

   initial  droplet  injection velocity

   drag  coefficient for  a  sphere based upon the
   local  droplet  and air velocities


   final  condition

   initial  condition
The results of this analysis  are  shown  in  Figure 14 where both air and es-

timated mean droplet diameter (DOQ) velocities are plotted as a function
of distance in the tunnel  divergence  section.  Velocimeter measurements
were within less than 5 percent of  the  predicted droplet velocities.  In
addition, droplet velocities  did  not  achieve  equilibrium with the test

section core velocity.
       Figure 14.
                         1234

                       DISTANCE FROM PARTICLE INJECTION ~ FT
Air and predicted particle (droplet)  velocity as
a function of distance from particle  injection
into the divergent wind tunnel  test section.
                                   27

-------
                  3.  LONG RANGE OPTICAL VELOCITY METERS

3.1  INTRODUCTION

     This discussion is concerned with the operation of optical  velocity
meters for use on large smokestacks, at ranges of 300 to 1500 feet (100
to 500 meters).  The smoke velocity to be measured is assumed to be in
the range of 10 to 125 feet per second (3 to 40 meters per second).  The
dust loading is in the range of 1 to 10 grains per standard cubic foot
(2 x 10   to 2 x 10   grams per cm ), and the mean particle diameters
are assumed to have the values of l^m, 10/am, or 50^m.  As representative
                                                    3                 3
specific gravities of the particles, 0.5 gram per cm  or 1 gram per cm
is used.
     Three types of optical velocity meters are considered here, all
of which are single station instruments which rely on backscatter.
The first type is called the direct doppler velocity meter.  Its con-
figuration is shown in Figure 15 where a single beam from a laser is
       Laser
               Figure 15.   Direct doppler velocity meter,
                                   28

-------
pointed at the smoke, and the backscattered  light  is  detected  by  a
heterodyne detector, using a fraction  of the laser's  outnut  as the  local
oscillator for detection.  From the  output frequency  of the  heterodyne
detector, the smoke velocity can be  obtained when  the elevation angle  e
is known.
     The second type of velocity meter uses  two beams from the same laser,
which intersect each other at the region to  be sampled (see Figure  16).
In the intersection volume common to the two beams,  a set of stationary
interference fringes is formed.  These fringes are effectively a  set of
"sheets of light."  When an object passes through  these fringes,  the
scattered light pulsates as the object passes from fringe to fringe.  By
measuring the pulsation frequency of the scattered light, one  obtains
the desired velocity information.  This type of velocity meter is
called the "fringe velocimeter."
                                          Fringes  or
                                       "Sheets  of  Liaht
              Figure 16.  Schematic of fringe velocimeter.
                                   29

-------
     This second instrument can also be explained purely in terms of
doppler shifts from the two beams, and hence is also known as a form of
laser doppler velocimeter.  However, the fringe explanation is used
because it makes many aspects of the operation more obvious and intui-
tive.  The fringes can be observed directly if desired by placing a
stationary card in the intersection region and examining the light on
the card with sufficient magnification to reveal  the fringes.  Analyti-
cally, the two alternate explanations are equivalent.
     The third type of velocity meter is called a "reticle velocimeter"
and is sketched in Figure 17.  This instrument is very simple and
can use sunlight as its illumination source.  It consists of a lens,
like a telescope objective lens, which forms an image of the smoke on
a reticle.  The reticle pattern consists of black stripes on a transparent
glass.  Behind the glass is a photodetector.  As a bright "blob" of  smoke
moves upward, its image passes over the reticle stripes and produces a
pulsating output from the detector.  The frequency of the pulsating
                     Figure  17.   Reticle  velocimeter.
                                   30

-------
detector output gives the velocity of the smoke.   This instrument thus
has considerable similarity to the fringe velocimeter.  The reticle at
the detector clays the same role that the fringes at the smoke play in
the fringe velocimeter.
     The remainder of this discussion will present information relevant
to the performance of these three types of velocity meters under the
smokestack conditions outlined at the beginning.   Since all three of
these meters present their velocity information as a somewhat periodic
signal on a noise background, the system  noise is  an  item  of  principal
concern.

3.2  SMOKE OPTICAL PROPERTIES

     For the performance evaluation, it is important to know the number
of particles in the sampled volume.  From the dust loading and particle
data of the introduction, the number of particles  per cubic centimeter
may be computed.  The  results are shown in the following table.
                   Table I.  Number of Particles per cm .
                        1 GRAIN/fr
10 GRAINS/ft'
Particle
Diameter
1 ym
10 ym
50 ym
Specific
Gravity
= 1
4.5 x 106
4.5 x 103
3.7 x 101
Specific
Gravity
= 1/2
9xl06
9xl03
7 x 101
Specific
Gravity
= 1
4.5 x 107
4.5 x 104
3.7 x 102
Specific
Gravity
= 1/2
9
9
7
x 107
x 10
2
x 10
      A second  characterization  of  the  particle density is  obtained by
 computing  the  mean  free  path  of a  photon in the smoke.  For present
 purposes,  it is  sufficient to use  the  geometric cross section rather
                                    31

-------
 than the more refined true optical  cross section.   Thus,  the
 mean free path is given by —, where n is the particle density  and  a
                            n w
 is the geometrical cross section area of the particle. The  mean  free
 path results are shown in Table II.
                 Table II.  Mean Free Path of Liaht in Smokes
1 GRAIN/ft3
Particle
Diameter
1 ym
10 pm
50 urn
Specific
Gravity
= 1
28 cm
280 cm
1400 cm
Specific
Gravity
= 1/2
14 cm
140 cm
700 cm
10 GRAINS/ft3
Specific
Gravity
= 1
2.8 cm
28 cm
140 cm
Specific
Gravity
= 1/2
1 .4 cm
14 cm
70 cm
3.3  SAMPLE VOLUME SIZE

     Before conclusions can be drawn from the foregoing tabular data,  it
is necessary tc know the volume and shape of the region sampled.   For
the first two instruments, this volume consists of the intersection  of
two beams (or of one beam and one observing path).  The configuration  is
shown in Figure 18.
     For a lens of diameter d, the minimum possible width of the beam at
range r is given approximately by
where A is the wavelength of light.  For a 4-inch (10 cm) lens at 100
meters range, this gives w = 1/2 mm, and at 500 meters w = 2.5 mm.  These
values are based on diffraction-limited performance, and in view of the
turbulent air encountered in the intended operation, it would appear
unwise to base an instrument on this high a performance standard.  This
standard of performance corresponds to a telescope of about 100 power or
greater.  The military seldom uses telescopes of powers greater than 20
because too often the turbulence renders higher powers useless.  Thus,
                                    32

-------
   Close-up of Intersection Volume
    Far  View  of  Intersecting  Beams
Figure 18.   Beam intersection  geometry.
                 33

-------
it seems prudent to expect at most a 2.5 mm beam width at 100 meters and
a 12.5 mm beam width at 500 meters, these values being 1/5 of the
diffraction-limited performance of a 4-inch objective, or equal  to the
diffraction-limited performance of a 0.8-inch objective.
     The length L of the intersection volume is given (to a close approx-
mation) by
where D is the separation cf the two beams at the observing station.
Since very high precision alignment between the two beams is required for
convenient operation, the beam optics must be mounted on a common rigid unit.
For a 2-meter sepe.raticn, about the maximum that seems practical, this
gives an L of 0.25 meter at 100 meters range, and 1.25 meters at 500 meters
range.  An instrument with such a large separation would require a special
truck.  Moreover, the large separation allows turbulence to affect the
beams independently, which introduces extra noise into the system.
     Referring to Table II, where the mean free path is given, it is
clear that for the small particle or heavy loading conditions, severe
attenuation of the beam both on its inward pass  and on its outward pass
will contribute to the difficulty of obtaining range resolution.  Moreover,
multiple scattering also imposes additional difficulty in obtaining range
information by degrading the signal-to-noise ratio when probing several
mean free paths deep.  Because of these difficulties, it appears appro-
priate to concentrate attention first upon a more convenient instrument
which makes no attempt to obtain range information, but simply probes the
plume with a long thin sample volume.  The returning information will
then be primarily from the region of the near side with a depth approxi-
mately one-half of the mean free path.  Unless otherwise stated, the
remainder of the discussion is limited to instruments with long thin probe
volumes having no range resolution.  At a later time, it may be appropriate
to consider more sophisticated instruments with range resolution.
                                     34

-------
3.4  SIGNIFICANCE OF THE SMOKE  OPTICAL  PROPERTIES  IN CONJUNCTION WITH
     THE SAMPLE VOLUME SIZE

     Three points of importance need emphasis.   The first  is  that  the
number of particles in the sample volume  is  large.  Even with the  smallest
beam width of 2.5 mm and the lightest loading  of coarse particles,  there
would still be about 600 particles in the sample volume across  a 3 meter
diameter stack.  For all other  conditions, the number  of particles in
the probe volume is much greater.

     The second point to note is that the mean free paths  are generally much
shorter than the length of the  probe volume  and of the stack  diameter.
Under these conditions, the amount of power backscattered  can most easily
be handled by a simple reflectivity coefficient rather than dealing with
the scattering functions from individual  particles.   Caution  must  be
exercised with white smokes where multiple scattering  may  cause complica-
tions, to be noted later.  Under these  short mean  free path conditions,
a change in smoke density does  not change the  amount  of reflected  light.
This absence of density dependence is most disadvantageous for the fringe
velocimeter.  In the fringe velocimeter,  it is desired that as a density
fluctuation, say a small local  increase in density,  passes by the  fringes,
a pulsating return signal is produced.   However, in  the case  where the
mean free path is short compared to the probe  volume,  no  pulsation of the
return is obtained because the  density  variations  simply  allow the light
to penetrate a little more or less so that the same  number of particles
are always involved, resulting  in a constant return.   This phenomenon
is referred to as "contrast washout."
     A third point to note is that the  mean free paths are longer than
the scale of some of the turbulent eddies.  This results  in a spread of
velocity components along the viewing direction.  As  will  be  explained
later, this spread creates a "granularity" noise which is significant in
the laser-illuminated cases.
                                    35

-------
 3.5  SHOT NOISE OF DETECTION

     One of the principal sources of noise in the instruments under con-
 sideration is the shot noise of the detection process.  This noise arises
 from random fluctuations in the emission times of the electrons at the
 photocathode of the detector.  In systems using photomultipliers, this
 source of noise is often dominant over other sources, such as thermal
 agitation noise.
     To develop a formula for the signal-to-noise ratio due to shot noise,
 the incoming light flux is represented in the form
                        S(t) = SQ + S^t) + S2(t)   .

Here, S(t) represents the light power (in watts) falling upon the detector
and as indicated in the above formula, this is represented as the sum of
three terms.  The first term S  is the average dc power of the light.  The
second term S-,(t) represents those fluctuation components within the
signal pass band of the instrument; that is, S-,(t) is the desired signal.
The third term, Sp(t) represents those fluctuation components outside the
signal pass band.  The average values of S-,(t) and Sg(t) are zero, since
S  is the dc average of S(t).
     The photocathode current is then given by

                     Kt) = ne {L_[so + S^t) + S2(t)]   ,

where n is the quantum efficiency of the photodetector (electrons per
photon), e is the electronic charge, h is Planck's constant, and v is
the optical frequency.  After this current is passed through the band
pass filter, there remains the signal,
When this current is passed through a resistor, R, the time-averaged
signal  power is then
                                     36

-------
                                       -*-^2   ,               (3)
                                       (hv)2  ]
where
                                     T
                        S,2 - J1m  1    SU)  dt
                         I     |-H*>  T
     The shot noise power in the pass  band is given by

                                  2eIRAf   ,

where e is again the electronic charge,  R is  the same load resistor, and
Af is the bandwidth of the pass band.*  Here  I is the average photocurrent
and hence is given by
Combining these, the shot noise power in the pass band is given by

                             2e2n J- R S  Af   .                   (4)
                                  n\>    o

     Taking the quotient of (3) and (4) , gives the signal-to-noise ratio,
                                               c" 2
                       Signal  Power   _ 1  1    1                 /5\
                     Shot Noise Power   2f hv SAf   *
     This formula can be written in a form so as to have an easily
visualized content.  To do this, note that
                                                                  (5'6)
is the average number of detected photons (i.e., photoelectrons) in the
time interval — F •   The time interval -TF corresponds to the time of one
full cycle at a frequency corresponding to the bandwidth.  Thus, one can
wri te
*  American Institute of Physics Handbook, Section 6m, "Radiation De-
   tection."

                                    37

-------
              Signal  Power
            Shot  Noise  Power
  /Average number of
T_J detected photons
2 \in time interval  T>
  x                 At
(6)
 The  fraction  S-/S  can  be thought of as the modulation index or modulation
 depth  of the  incoming light.   It is the ratio of the RMS signal component
 to the average light power.
     Note that it is the number of detected photons which enters the
 signal-to-noise ratio formula.  Thus high quantum efficiencies are
 desirable, as well as high powers of the light itself.  Also, narrow
 bandwidths favor the signal-to-noise ratio, for they provide longer time
 intervals  —F for the detection of photons in (6).
     Formula  (6) can be used to determine the required laser power output
 to achieve a given signal-to-noise ratio.  However, the pass band Af and the
 modulation depth for each system must first be determined.
3.6  THE DIRECT DOPPLER SYSTEM

     In order to maximize the signal-to-noise ratio, it is desirable to
keep the pass band as small as possible without losing signal.  The signal
width for the direct doppler instrument is determined by turbulence within
the smoke.   Referring to the following sketch (Figure 19), the observing line
         Figure 19.   Effect of turbulence on direct doppler system,

                                    38

-------
is at an elevation angle  e.   Within  the  sample  volume,  the  average vertical
component of velocity is  denoted  by  v.   This  is the  velocity which it  is
desired to measure.   Because  of the  turbulence, there is added to v  a
turbulent velocity,  VT-   As a rough  guide,  the  turbulent velocities  can
be taken to be about 10 percent of the main flow velocity v.  These  tur-
bulent velocities  can be  in any direction,  and  thus  some will be along
the sight line.

     The component of the main flow  velocity  along  the  sight line is
v sin e .  Thus, the average  doppler frequency  is
                                  _  V
When a turbulent velocity v-,- occurs  along  the  line  of sight,  the  frequency
will be

                           f = -(v sin  e  ± VT)
                               A             1

Thus, the spread in frequency due to turbulence is

                                Af = 2  ^   .                      (8)
                                       A

Combining (8) with (7), one obtains


                               -         1
                            f      v   sin e

     This equation exposes an important limitation of the direct dopoler
velocimeter as applied to the present turbulent smokestack case; namely, if
the elevation angle 9 is too small, then the turbulence causes a spread in
the signal frequency comparable to the basic doppler frequency itself.
Under these conditions, it is impractical to obtain the desired velocity
information.
     At steep elevation angles, some determination is possible.  For
example, at 45 degreees with the value of VT/V = 1/10  (turbulent
velocities 10 percent of main flow), then Af/f =0.28.  With such an
instrument, a 10% determination of average velocity may be possible.
                                    39

-------
     The preceding limitation may possibly be overcome with highly sophis-
ticated systems which  "track" the turbulent velocity fluctuations, thus
permitting an effective narrow pass band and accurate determination.  The
investigation  of such a sophisticated system is beyond the scope of the
present report.  In this connection, note that due to the fact that the
depth of penetration is larger than some turbulence cell diameters, the
returning signal is not a single frequency shifting about, but a complex
composite with different frequency components simultaneously present from
the different parts of the sample volume moving at different velocities.
     For the direct doppler instrument, the best signal-to-noise ratio is
obtained with a reference that is strong compared to the signal.  Under
this condition, formula (6) becomes

                                /Average number of detected\
                               fel P^cns In the      )  .    (10)
                               ytime interval —F           /

(This formula assumes that the reference source does not contain any excess
noise, a matter of practical  concern.)
     This formula can be applied to predict the required laser power
output for a direct doppler instrument.  Example conditions are  as follows:

          Elevation Angle e              =45 degrees
          Slant Range                    =  100 meters
          Reflectivity of Smoke          =  .1
                                                p
          Receiving Aperture             = .01 m  (4.5  in. diam. lens)
          Quantum Efficiency             =  .1
          Wavelength                     =  .5 ym
          Smoke Velocity                 =  40 meters/second
          Turbulent Velocity/Flow Velocity = .1
          Signal-to-Shot Noise Ratio     =  10
These conditions, together with the foregoing formulas, yield the
following:
                        Center Frequency       57 MHz
                        Bandwidth              16 MHz
                        Required Laser Power  .02 watt

                                    40

-------
     It thus appears  that a  direct  doppler  instrument may  be a
possibility inasmuch  as  argon  lasers with considerably greater power
than the above are  available.
     However,  there are  many further considerations beyond the
fundamental one of shot  noise.   There  are additional  sources of  noise,  for
example, thermal  agitation noise, noise  associated with  the photomultiplier
multiplication process,  and  noise from fluctuations of the laser intensity
and intermode beating.   Background  light which  is  inadequately filtered
out is an additional  source  of noise.  Moreover, operation of a  photo-
multiplier at a frequency of 57 MHz requires  proper design.  For the
direct doppler system to work  as described  above,  it  is  required that
the illuminated spot  on  the  smoke supplied  by the  laser  be at least as
small  as the theoretical resolution limit of  the receiving lens.  In  addi-
tion, turbulence  along  the path can spoil the resolution,  which  in  turn
can decrease the  signal-to-noise ratio.  Multiple  scattering in  the smoke
can also increase the size of  the  illuminated spot  from  which the light
returns, and this too degrades the  signal-to-noise  ratio.
     The coherent detection  process used in the direct doppler type instru-
ment only functions as  described above with light  from particles within the
theoretical limit focal  volume. This  is a  cylindrical volume approximately
  r                       2
A -r in diameter and A (r/D)   in length.   For  the above example,  this  is a
cylinder 0.5 mm in diameter  by 0.5  meter long.   Light returning  from
particles outside this  volume  does  not have a uniform phase relationship
with the reference light over  the  full entrance aperture.   The  result is
that light from particles outside  this cylindrical  volume  contributes
little or nothing to  the signal.  Such light  can  increase  the  noise level,
however.  This requirement of  a uniphase wavefront is a  fundamental limit
imposed by the coherent detection  process.  The uniphase condition can be
achieved more easily  with smaller  receiving apertures,  but they  collect
less light which  strains the system in another direction.
     Still another demand of the direct  doppler system is  that  either the
laser must have a coherence  length  equal to the round trip to  the sample
volume and back,  or that an  optical delay be  provided for the  reference
light equal to the round trip  distance.   This is necessary to insure that
the reference be coherent with the  returning  signal.
                                    41

-------
     In summary, the direct doppler type instrument appears to have the
following properties:  Nearly horizontal paths cannot be used due to the
turbulence contribution.  Elevation angles on the order of 45 degrees
appear necessary, and these eliminate long range applications.  A one-watt
or greater argon laser is required to give a comfortable signal-to-noise
ratio.   Careful attention must be paid to many noise sources and to the
high frequency detector design.  The velocity measurement is not likely to
be better than 10 percent, and will have no depth resolution.
3.7  THE FRINGE VELOCIMETER: GRANULARITY NOISE

     Before turning to an evaluation of the fringe velocimeter, it is
appropriate to discuss a source of noise relevant to the fringe velocimeter.
This source is called granularity noise because of its relation to
the granularity or speckle pattern produced when a laser illuminates a
diffuse surface.   The origin of this noise is shown in the following
sketch  (Figure 20):
                                                   Diffuse Reflecting
                                                        Surface
           Detector Aperture
          Figure  20.  Granularity  noise  in  the  fringe  velocimeter.
                                    42

-------
In this sketch,  a laser beam  is  shown  illuminating a  spot  of diameter w
on a diffuse reflecting surface.   The  light  reflected from this  surface
travels in all  directions,  and at  range  r  there  is placed  a large white
viewing screen.   It will  be found  that the light intensity on  the viewing
screen is not uniform,  but  has a speckled  or granular appearance.   This
stems from the  monochromatic  nature  of the laser.
     The intensity at a given point  of the viewing screen  is found  by
adding up the amplitudes  from all  the  different  points composing the
illuminated spot of diameter  w.  This  is a phase addition  with the  ohases
randomized by the diffuse nature of  the  reflecting surface. At some
points, the phases will largely  cancel each  other, resulting in a low
intensity or dark spot.  At other  points,  the phases  will  be largely  in
phase, resulting in a bright  spot.
     The "size", s, of these  granules, on  the viewing screen,  is approxi-
mately given by
                              c    1  r                             (1°'5)
                              S ~ A  W   '
Actually, the distribution  is random,  and  what is meant by the above  is
that two points closer together  than  s will  not  have  independent inten-
sities, but will be correlated.  To  the  eye, the characteristic apparent
"diameter" of the granules  is given  by s.
     [In the direct doppler instrument,  the  uniphase  wavefront requirement
discussed earlier corresponds to the condition that  the receiving  aperture
be no larger than the size  of one  granule, i.e., be  of diameter -  s.]
     In the present smokestack application,  the  motion of the  smoke causes
the granularity pattern to  change  in time, which creates intensity fluctu-
ations at the detector.  These constitute  noise, and  it is necessary to
evaluate the amount of this noise  which  is in the pass band of the
instrument.
     It can be shown that the RMS  fluctuation in light power entering
detector, AP, is given by
                                    43

-------
 where PQ is the average light power entering the  detector  and M  is  the
 number of granules present in the detector aperture.   (The foregoing assumes
 that the returning light is still oolarized after reflection.  If not, M
 should be replaced by 2M in Equation 11.)  If the  detector  aperture  diameter is
 D, then
                               M = (f)     .                        (12)

 Notice that if M = 1  corresponding to  D  =  s, then the  RMS  fluctuation is
 equal  to the average  light power.  By  making the  receiver  aperture  large so
 that M becomes large, or by making w large  so that s becomes small  and
 hence M large, these  granularity fluctuations are reduced, decreasing the
 system noise.
     To evaluate the significance of this granularity noise, we  need to
determine how  rapidly the pattern changes, and what fraction of  the  noise
is in the instrument's pass band.
     If the smoke were a rigid mass moving uniformly upward, then when
the smoke has moved upward a distance w, all new particles would be  in
the illuminated region, and a new granularity pattern would be  present.
Thus, the upper frequency limit of the  granularity noise is given by
                            rigid smoke
where v is the smoke velocity.
     However, rigid translation of the smoke upwards is not the only
source of change in the speckle pattern.  Recall that in the present
application, the probe volume length is long enough to span more than one
turbulent cell.  Thus, the reflector is effectively a set of two or more
partial reflectors which are in motion relative to each other.  When the
relative motion is of the order of one wavelength, then the phase addition
is uncorrelated with the previous addition, and a new speckle pattern is
formed.  Thus, letting VT denote the turbulent velocity components as
                                    44

-------
before, and recognizing that some  of  these  occur  along  the  line of

sight, the apparent frequency of speckle  pattern  change due to

turbulence is approximately given  by
Inasmuch as VT is approximately 1/10  of v,  the  frequency given  by

(14) is much higher than that given by  (13),  for realistic values of

w in the present application.  Thus,  formula  (14) is  the proper one

to use for estimating the effects  of  granularity noise in the  realistic
case.

    By comparison with other aspects  of speckle behavior, it is

inferred that the frequency (power) spectrum of this  noise is,  to
an  accuracy  sufficient  for present purposes, approximately  flat  from

zero up to f   , given by (14).  On this basis, the RMS fluctuation
            IT13X
of the arriving light due to granularity noise  in a pass band  of width
Af may be found by combining (11)  and (14).*  The result is


    (RMS Granularity Fluctuation of Arriving Light) =  Pa -L  f  —   ; (14.5)
                                                      a
hence,


                                                        T  .^^™
  (RMS Granularity Fluctuation of Arriving Light) = PQ —  A_ Af  .   (15)
 *  To  obtain  the portion  in the  pass band,  it  is  necessary to  use
    the fluctuation power  (the  square of  the  RMS).   The  fraction
    Af/f of  this total  fluctuation  power  is within  the passband.
    The square root of  this fraction gives the  RMS  of the  fluctuations
    in  the pass band.   An  alternate argument  leading to  the same
    result is  that the  basic fluctuations occur every l/fm3V seconds.
                                                        iNaX
    The number of basic fluctuations in the  time interval  1/Af  is
    therefore  fm,w/Af.   The RMS value of  the  sum of f   /Af samples
               max                                    max
    is   /Af/f       times  the  RMS  of  the  basic  fluctuation  RMS

    given  by (11), which yields  the  formula  (14.5).
                                   45

-------
     To obtain a signal-to-noise  ratio, it is necessary to consider the
 signal and  the pass  band Af over  which the signal is spread.  For the
 fringe velocimeter,  let A denote  the spacing of the fringes and N denote
 the  total number of  fringes across the probe volume's width W.  Then:

                                NA = w    .                        (16)

 The  output  frequency due to particles crossing the fringes with velocity
 v  is

                             fsignal = I   '                      (17)

 Since there are N fringes across  the probe volume, it can be shown that
 the  spectral width Af  .   , of the signal is given approximately by

                       Afsignal = fsignal ' 1   '                 ^

     Combining (15) with (16), (17), and  (18) yields the granularity noise
 in a pass band which just passes  the signal.  The result, after appropriate
 cancellation, is

       (RMS Granularity Fluctuations of Arriving
               Light in Signal Pass Band)        = P

     Attention is next turned to  the signal.  If one particle at a time
passes through the intersection volume, the returning light is 100
percent modulated by the fringes.  However, as more particles are simul-
taneously present in the probe volume, the returning light is less
modulated because some particles are "on" the fringes at the same time
that others are "off" the fringes.  It can be shown that if the particles
are randomly distributed, the following relation holds:

             (RMS Fluctuation of Arriving    i      i
               Light Power Due to Signal) = -=. Pa —   .          (19.5)
                                                   m
                                   46

-------
where P  is again the average light power returned  to  the  detector,  and
       a
m is the total  number of particles  returninq  light  from the  probe  volume.
This signal is spread over the frequency interval Af .  „ , given by  (18).
                                                    signai
     Dividing this by (19) yields  the following ratio,
                RMS Signal  Fluctuation    \  Jl  w  ^T M          /20>
              RMS Granularity Fluctuation/  i2  A  v  m

The ratio (20) is the ratio between the RMS fluctuation of the arriving
light flux due to the signal, and the RMS fluctuation of the light, due to
granularity, each within the signal passband.   After detection, these light
fluctuations become fluctuations in a current (or voltage).  Since the power
in the signal or noise is proportional  to the square of the current (or
voltage), the signal power to granularity noise power ratio is the square
of the ratio given in (20), namely,

                 AC Signal  Power         1_ w  ^T M_               /2i)
             Granularity Noise Power     2 x  v  m

This may be evaluated with the aid of the particle densities given earlier
in this report.
     The special case in which the mean free path is short compared to
the stack diameter yields a particularly informative relation.  For this
case, the total number, m, of particles returning light is approximately

since — is the mean free path, n being the particle density.  Moreover,
      an                    n
to first approximation a = 6  , where 6 is the particle diameter.
Substituting these in (21) yields


                                             =       ~ M   '
                                    47

-------
      Using  (10.5)  and  (12), this relation may be written,

                   AC  Signal  Power      -/iV  ^1/5.Y *L          (??}
                Granularity Noise Power  ~\xy   v  \r/  x          ^   '

The factors 6/x and Vj/v are properties of the smoke, and thus for
operation at a  given range, the designer has only the lens diameter D
and the probe volume width w with which to secure a favorable signal to
granularity noise  ratio.  The formula indicates that large values of both
D and w are desirable.  It will be seen later that for random particle
distribution, a small  value of w is desired to enhance the signal-to-shot
noise ratio.  Thus, an optimum w exists which minimizes the sum of
granularity and shot noise powers.

      It should  be  noted that  the uniformly random distribution yielding
formula  (19.5), on which the  above is based, may not give a true
description of  the behavior.   It seems  reasonable that local clumping
due to turbulent eddies may occur, and  if these clumps are comparable  in
size to the fringes, the clumps may serve as large "particles", giving
rise to much better performance than given by (21)  and (22).  On the
other hand, for the short mean free path case, there can be a contrast
washout because a  region of low particle density simply lets the beam
penetrate further  until the backscatter is the same as for a higher
density region.  In this case, the true situation can be worse than
predicted by (21)  and  (22).

     Notice that the ac signal power-to-granularity noise power is
independent of  the laser power.  Thus,  one must arrange the system so
that this ratio has a  favorable value,  and then use the formula for
signal-to-shot noise ratio given much earlier (formula 6) to determine
the required laser power.

3.8  FRINGE VELOCIMETER:  REQUIRED LASER POWER

     As an illustration, formula (6)  or (5)  is next applied to a
representative case to estimate the required laser power.  Again
assuming a random particle distribution in the sample volume, it. follows

                                   48

-------
that the modulation depth S,/S  of the received light has  the  value  -j=
                           •   U                                      > p1
where m is again the number of particles  in  the sample volume.   Thus, the
factor Sj2/S02 in (6) has the value 1/m .
     The time interval  —?- is  determined by the  signal  bandwidth, and is
         W
given by - , where v is the smoke  velocity to be measured, and w is  the
probe volume's width.  This follows from  formulas (16), (17),  and (18).
     Substituting these values in  formula (6),  and incorporating appro-
priate geometric factors for the fraction of laser light collected,  the
following is obtained:
                 Signal Power   _ 1_ /   p   _A_ 1_ w\ 1           (9*\
               Shot Noise Power     2 lnp  n    ° u- •• ' -   '       *•    ;

where
            n   =  quantum efficiency of  the detector
            p   *  reflectivity  of the smoke
            P.   =  laser power output
            A   =  area of the detector collection aperture
            r   =  range
            hv  =  energy of a photon (4  x  10    joules for 0.5/urn light)
The quantity in parentheses is the average  number of photons detected in
time interval —F  •
              At
     Denoting the probe volume's length by  L:

                              m  =  n w2L   ,                        (24)

where n is the particle density.
     Incorporating this into formula (23):

                 Signal Power      1_    p  A	1  1    1           (25)
               Shot  Noise Power    2 np  L    2 hv v  nwL   '

     If it is further  assumed that the mean free path, — , is  short
compared to  the stack  diameter,  and that the particle  cross section a is
approximately 62, where  6 is the particle diameter, then:
                                    49

-------
      (Short mean
      free path
      case)
  Signal  Power
Shot Noise Power
np
p   n  '   i- °
 L   2 hv v w
     Notice that both formulas (25) and (26) show that small  values of w
are desirable in order to improve the signal-to-noise ratio.   This stems
basically from the fact that small values of w correspond to  fewer
particles in the probe volume, which makes the fluctuations proportion-
ately greater.  As noted earlier, this is based upon a completely random
particle distribution, and may not apply if turbulence causes local
clumping.
     If the particle distribution is truly random, then the small values
of w required by (25) and (26) prevent using large values of  w
to combat the granularity noise.   In this event, a large receiving
aperture is the only method for providing a large M to combat granularity
noise.
     An example of the power required by formula  (26) is given below.
The example conditions are similar to those of the direct doppler example;
however, the receiving aperture is increased to keep the granularity noise
down.
     Conditions:
          Range r                          =
          Reflectivity of Smoke p
          Receiving Aperture  A             =
          Quantum Efficiency  n             =
          Wavelength
          Smoke Velocity v                 =
          Signal-to-Shot Noise Ratio       =
          Particle Diameter 6
          Probe Volume Width  w
          Turbulent Velocity/Flow  Velocity =
     These values yield the following:
          Reauired Laser Power             = -05 watt
                         100  meters
                         .1
                             Q
                        .04 m  (9 in. diam. lens)
                         .1
                         .5 ym
                         40 meters/second
                         10
                         10 ym
                         25 mm
                         .1
                                    50

-------
     The foregoing  does  not  depend on the number N of frinpes across the
probe volume because  it  is assumed that the optimum bandwidth is used.
However, to give a  feeling of  the frequencies  involved, it is further
assumed that N = 10,  in  which  case:
                   Signal Frequency        =   16  kHz
                   Signal Frequency Width  =   1.6 kHz
     In conjunction with formula (20), the example conditions yield
                         Signal Power
                     Granularity  Noise  Power
                                              =  8
     The example  conditions  are  for the  near  range  conditions  of
100 meters.   For  the  far range conditions of  500 meters,  the situation
                                                                      o
is more demanding.   In formulas  (22),  (25), and  (26)  one  observes  an  r
dependence.   In addition, w  may  depend on r,  depending  on the  optical
configuration.
     It is to be emphasized  that the  example  conditions are  given  as  a
guide to scale  from,  not as  a design.   In particular, the smoke reflec-
tivity may be quite different, and a  signal-to-noise  ratio of  10 is not
a comfortable design  point in view of other secondary noise  sources.

3.9  RETICLE VELOCIMETER NOISE ESTIMATE

     A performance estimate  for  the reticle velocimeter is not easy to
make because of unknown smoke properties.
     It appears desirable to sample the  flow immediately after it emerges
from the stack, for otherwise large vortices of entrained air will  compli-
cate the situation and reduce the accuracy.
     The contrast of the issuing smoke is  not known.   When sunlight is
used as the light source, its direction  is  automatically different from
the viewing direction, and hence the contrast washout mentioned earlier
does not apply.  It seems likely that performance can be optimized by
choosing a viewing direction appropriately oriented relative to the sun.

                                    51

-------
     As with the fringe velocimeter, the relative signal bandwidth |^
                                   1                "               T
is given to first approximation as JT, where N in this case is the number
of reticle lines across the field of view.  This first approximation will
be correct if the smoke "blobs" remain intact long enough to traverse the
field of view.  If not, the spectrum will be broadened.  Since the tur-
bulent velocities are of the order of 10 percent of the flow velocity,
one might estimate that a system with 10 reticle lines across the field
of view would satisfy the condition that the "blobs" remain intact across
the field of view.
     Formula (6) for the signal-to-shot noise power can also be applied
for the reticle velocimeter.  The illumination power in this case is
supplied by the sun.  The following example is given for comparison, but
it is emphasized that the smoke property assumptions are rather arbitrary
and may not be realistic.
    Conditions:
         Sunlight at 45°
         Area Sampled on Smoke
         Number of Reticle Lines =
         Range                   =
         Smoke Reflectivity
         Quantum Efficiency      =
         Smoke Velocity          =
         Receiver Aperture
                      Power density 300 W/m
                     .01 m2  (.1 m x  .1 m)
                      10
                   =  100 meters
                   =  .1
                   =  .1
                   =  40 m/sec
                   = .01 m2
         Contrast Ratio of Smoke,
              I
               max
     - I
        mm
                 I
                  ave
         Number of "Blobs"
              Simultaneously in
              Field of View
                                 =  .1
                   =  100
     The conditions preceding the last two give the following results:
         Signal  Frequency                  =  4 KHz
         Signal  Bandwidth  Af              =  400 Hz
         (Average Number of Detected
aye MUIIIUCI  ui ucocv-ucu   -i
Photons in Time Interval —p) =   6x
                                                  10'
                                    52

-------
The last two conditions give the modulation  depth,  5./S ,  as .01,  since
there are approximately /TW "blobs" above or below the average "on" the
fringes.  Together with formula (6), these results  yield

                 Signal Power     1    fi v ,n7   /  mA2   ,nnn
               Shot Noise Power" 2    6 x 10  '('01)   = 300°  '
This result suggests that the shot noise of  detection  may not pose much
of a problem.  This basically stems  from the abundance of power provided
by the  sun  and the narrow bandwidth Af.
     Whether artificial light (e.g., sootlights) could be used to  permit
night operation is not clear because of the  large  uncertainties in the
above conditions.
     Another source of noise, and quite possibly the primary noise source,
is noise generated by the turbulence modulating the reflectivity of the
smoke.  Experiment appears to be the only approach to  determine the
magnitude of this noise source.

3.10 PRELIMINARY EXPERIMENTS WITH A RETICLE  VELOCIMETER

     A  very rough set of experiments was done to explore the reticle
velocimeter concept.  A 2-inch diameter telescope objective was equipped
with a  photomultiplier and a reticle.  The output of the photomultiplier
went to an audio amplifier and a loudspeaker.
     In a first experiment, a lambsv/ool polishing wheel mounted in  a
hand electric drill served as simulated smoke.  The wheel was placed
in the  sun, and visually would be described as a low contrast target.
At 50 feet, the maximum space available, the speaker output had a
clearly recognizable tone which moved up and down appropriately as  the
instrument was slowly  scanned along a diameter of the wheel.
     Indoor night experiments at a  few feet with a dc tungsten lamp
as illumination also performed similarly.
                                    53

-------
     The performance of this type of instrument is independent of range
 under  the condition that the angular field size does not change.  This
 condition means that the object area sampled increases as the range is
 increased.
     The instrument also performed well as a velocimeter for automobiles.
 At about 75 feet, the passing of each car yielded a distinct tone pulse,
 and differences in velocity were readily apparent to the ear.  Car head-
 lights at a distance of 2 miles at night gave a just detectable tone.
 The reticle spacing was such that the tone produced was actually too low
 for the pass band of the audio amplifier at this long range.
     No smoke plumes were readily available in the Los Angeles area, but an
 attempt was made to use the velocimeter on a refinery burn-off flame.
The attempt, done at night, was unsuccessful.   The main reason was
that at the nearest available distance to the  flame (about 2000 feet),
the field of view of the instrument was larger than the entire flame.   It
should have been small  enough to sample only a small area near the mouth
of the pipe.  In consequence, the principal output consisted of noise
from intensity fluctuations of the flame itself.  Another possible reason
for the failure with the flame may be that the flame propagation velo-
cities are higher than  the average translation velocity.  Thus, the
luminous structure is changing faster than the "blobs" can be swept
across the reticle lines by the average translation.  There is reason
to expect smoke to be more coherent in this respect than a flame, due  to
the high flame propagation velocities.  At times a crackling or frying
noise was apparent,  and visually corresponded to a "sparkle" in the
f1ame.
     In spite of the failure of the flame experiment, it is still felt
that a properly designed experiment that examines a small area near the
pipe's rim has some chance of success with flame and a good chance for
success with smoke.
                                    54

-------
                              4.   CONCLUSIONS
4.1  PORTABLE LASER VELOCIMETER

    A portable laser velocimeter was  designed  and built to  measure  particle
laden stack gas velocities based upon backscatter signals generated by the
particulate as it passed through the  region generated by two intersecting
laser beams.  Only limited testing of the instrument in a laboratory environ-
ment was accomplished due to cost and schedule constraints  of the program.
Sufficient work was done to satisfactorily demonstrate its  operating range
characteristics.  Of significance in  these tests was the sensitivity of the
laser velocimeter to signals generated by backscattering of the incident
laser beam on a smooth white paper disc.   Application of the velocimeter to
actual stack measurements, however, remains to be demonstrated.
     Work on the present velocimeter has  led to the conclusion that future
models could be made even more compact and lighter weight.   Utilization of
a  velocimeter in a fixed installation as  a continuous stack gas velocity
monitor would greatly simplify both the electronics and optical system.
Unit costs  in even limited production would be considerably reduced.
4.2  LONG RANGE VELOCIMETERS
      This study examined the expected performance of three types of
long-range  smokestack velocimeters.   The  discussion attempted  to expose
the  problems  associated with  each  type of  instrument,  and  to indicate what
parameters  are  effective  in optimizing the performance.
      Definite conclusions  are  not  possible because  of  unknown  smoke
properties.   The  reticle  velocimeter  appears  attractive because  of its
extreme  simplicity  and  resulting  low  cost.  Its  performance  with smoke
is still  untested,  but  it  has  performed with  a  rotating wheel  simulating
"rigid"  smoke.   It  should  be  explored  further.
      The two laser instruments each  appear to require  a one-watt argon
laser in order  to operate well at long distances.
      The fringe velocimeter type  appears  to be the simpler to  construct
and can advantageously  use a  large collection aperture.   Its actual  per-
formance will  depend on the smoke properties;  in particular, non-random
clumping may assist its operation greatly.  If the clumping does not

                                     55

-------
exist and the instrument is forced to operate on pure random particle
distributions, the principal problem of the instrument then stems from
the small magnitude of the intensity fluctuations, which in turn arise
from the large number of particles simultaneously present in the probs
volume.
     The direct doppler instrument will only work at steep elevation
angles due to turbulence in the smoke.  This eliminates the really long
range possibilities.  The collection aperture is limited in size by
coherence considerations.  Multiple scattering may adversely affect
its operation.  Very high frequency electronic design is required for
the direct doppler instrument, adding to the difficulty of constructing
the direct doppler instrument.
     More sophisticated instruments which give range resolution, and
which function with very light particulate loading have not been con-
sidered in detail.  It is felt that some actual experience should be
acquired with a simple long-range system before embarking on the design
of a more sophisticated system.
     In  interpreting the significance of the noise discussed in this
report,  it should be noted that the signal-to-noise ratios given are
power ratios, as is customary practice.  Note, however, that most display
devices,  such as meters and the oscilloscope deflection of a spectrum
analyzer, are proportional  to voltage.  Thus, the ratio of deflections
for signal  and noise is given by the square root of the signal-power-to-
noise-power ratio.
                                   56

-------
                              REFERENCES
1.   A.  E.  Lennert,  D.  B.  Brayton,  F. L.  Crosswy, W. H. Goethert, and
     H.  T.  Kalb,  "Laser Technology  in Aerodynamic Measurements,"
     von Karman Institute  for Fluid Dynamics Lecture Series 39, June 1971


2.   B.  J,  Matthews  and H.  Shelton,  "Operation Manual  - Portable Laser
     Velocimeter for Stack Velocity Measurements," TRW Report No.
     20852-6001-TQ-QO,  June 1972.
3.   R.D. Ingebo, "Drop-Size Distributions  for  Impinging-Jet  Breakup  in
     Airstreams Simulating the Velocity Conditions  in  Rocket  Combustors,
     NACA Technical  Note 4222, March  1958.
                                   57

-------
                         APPENDIX  A






SPECIFICATIONS FOR SPECTRA-PHYSICS MODEL  120 GAS  LASER WITH



                     MODEL 256 EXCITER

-------
                                APPENDIX A


        SPECIFICATIONS FOR SPECTRA-PHYSICS MODEL 120 GAS  LASER WITH
                           MODEL 256 EXCITER
Output Power:  5.0 mw s 632.8 nm
Transverse Mode:  TEMoo
Warm-up Time:  > 3 mw 2 minutes after turn-on
               > 5 mw 30 minutes after turn-on
Operating Temperature:  10 to 40°C
Long-Term Pov.-er Drift :  < 5$
Altitude:  Sea  level to 10,000 feet
Beam Amplitude Noise (1-100 KHz):  < 3^
Beam Amplitude Ripple (l20 Hz):  < 0. 5\>
Beam Polarization:  Linear to better than 1 part per thousand
Plane of Polarization:  Vertical, adjustable ±20°
Beam Diameter:  O.G5 mm at 1/e2 points.'
Beam Divergence:  1.7 milliradians at 1/e2 points
Resonator Configuration:  Long Radius
Axial Mode Spacing:  385 MHz

Plasma Excitation:  Direct current self-starting
Cable Length  (Exciter to Laser):  8 feet (Extension sections  available)
Dimensions:  Model 120 Laser Head:  3.26" w  x  3.66"  h   x  18.4S1' 1
             Model 256 Exciter:     7.25" w  x  3.72"  h   x    9.88" 1
             Weight:  Laser - T/2 Ibs.
                      Exciter - 1% Ibs.
Input Power:  115/230v, 50/60Hz, 50va
                                    59

-------
         APPENDIX B





LASER VELOCIMETER DRAWING LIST

-------
                         APPENDIX B

               Laser  Ve1oc1meter D r a w i ng L1 st

Drawing No.                                Title
 SK 1038-1                Velocimeter Probe
 SK 1038-2                Support, Rack & Pinion Lense Assembly
 SK 1038-3                Collar, Rack Pinion Assembly
 SK 1038-5                Tube, Lense Holder Velocimeter Probe
 SK 1038-6                Split Mirror, First Surface Support
 SK 1038-7                Holder, Lense 30 MM dia.
 SK 1038-9                Support, First Surface Mirror
 SK 1038-10              Base, Sub Velocimeter Probe
 SK 1038-11               Body, Beam Splitter
 SK 1038-12              Laser Modification
 SK 1038-13              Cover, Velocimeter
 SK 1038-14              Panel, Velocimeter
 SK 1038-15              Bracket, Grip Holding Tripod Support
 SK 1038-17              Dial Face, Beam Splitter
 SK 1038-18              Dial Face, Focus
 SK 1038-21               Support, Laser
 SK 1038-22              Support, Laser
 SK 1038-31               Collar,  Front Rack & Pinion
 SK 1038-32              Collar,  Rear Rack & Pinion
 SK 1038-33              Support, Collar
 SK 1038-34              Gear,  Pinion & Shaft, Collar
 SK 1038-35              Bearing, Tube, Lense Holder
 SK 1038-51              F'lut,  Lense  Holder Velocimeter  Probe
 SK 1038-52              Washer,  Lense Holder Velocimeter Probe
 SK 1038-53               Rack,  Tube  Lense  Holder
 SK 1038-61               Strap,  Split Mirror
 SK 1038-62               Split Mirror  First Surface
 SK 1038-71               Nut,  Lense  Holder 30 MM  Dia.
 SK 1038-72               Washer,  Lense  Holder  30  MM  Dia.
                              61

-------
Drawing No.                                Title

SK 1038-91                Strap, Support First Surface Mirror
SK 1038-92                Mirror, First Surface
SK 1038-101               Spacer, Sub Base
SK 1038-102               Rubber Mount Assembly
SK 1038-111               Pawl,  Beam Splitter Velocimeter
SK 1038-112               Link,  Beam Splitter Velocimeter
SK 1038-113               Shaft, Beam Splitter Velocimeter
SK 1038-114               Slide, Prism Beam Splitter
SK 1038-115               Slide, Mirror First Surface Beam Splitter
SK 1038-116               Holder, First Surface Mirror Beam Splitter
SK 1038-117               Plate, Body, Beam Splitter
SK 1038-118               Gib,  Vee Slide Beam Splitter
SK 1038-119               Gib Spring, Vee Slide,  Beam Splitter
SK 1038-131               Window, Lense Focus
SK 1038-141               Focus  Dial Window
SK 1038-142               Beam Dial  Window
SK 1038-151               Support, Tripod
SK 1038-171               Post,  Dial Beam Splitter
SK 1038-172               Pointer, Dial  Beam Splitter
SK 1038-181               Pointer, Holder
SK 1038-182               Pointer, Focus
SK 1038-1110              Spring, Vee Slide, Beam Splitter
SK 1038-1111              Spring Drag, Beam Splitter Adjustment
                                62

-------