DESIGN CONSTRUCTION AND EVALUATION OF A
MOBILE LIDAR SYSTEM FOR THE REMOTE MEASUREMENT
OF SMOKE PLUME OPACITY
FINAL REPORT
EEI RESEARCH PROJECT RP39 AND
EPA CONTRACT NO. 68^
C.S. COOK
G.W. BETHKE
SPACE SCIENCES
LABORATORY
SION
GENERALO ELECTRIC
-------
DESIGN, CONSTRUCTION AND EV ALUA TION OF A MOBILE LIDAR
SYSTEM FOR THE REMOTE MEASUREMENT OF SMOKE PLUME OPACITY
Final Report
Period Covered: December 24,1969 to December 21,1971
EEl Research Proj. RP39 - GE-SD Requisition No. 2l4-A 77
EPA Contract No. 68-02-0093 - GE-SD Req. No. 028-N52
Prep ared by:
C. S. Cook and G. W. Bethke
General Electric Company
Space Division
Space Sciences Laboratory
P. O. Box 8555
Philadelphia. Pa.
19101
Prepared for: Environmental Protection Agency
Durham, North Carolina 27701
and
Edison Electric Institute
90 Park Avenue
New York. New York 10016
-------
ABSTRACT
A mobile (truck mounted) ruby laser lidar system has been designed, con-
structed and evaluated for the remote measurement of smoke plume opacity (or
transmittance ).
The system has been tested at ranges of 211 and 319 meters using synthetic
targets of known laboratory measured transmittance.
The targets used were
made of bright and black anodized aluminum screen,glass, plexiglass, white
painted plywood and black felt.
These tests indicated an error which increased
as target reflectance increased,the reflectance increase being due either to a
decrease in transmittance or to an increase in reflectance at a given transmittance
This error was found to be less than + 2.5 percent for target transmittances
greater than 0.80 and was less than + 12 percent for target transmittances great-
er than 0.50.
The error was always positive making the lidar measurements
always an upper bound measure of transmittance.
The transmittance measurement error was found to be largely due to photo-
multiplier tube afterpulsing, which, after extensive photomultiplier tube gating
investigations, could not be avoided with the then available photomultiplier tubes
and lack of detailed understanding of the afterpulsing mechanism.
However, two
new techniques, one recently available, seem likely to eliminate the afterpulsing
problem particularly since the mechanism is now more evident as a result of the
investigations during this program.
The lidar system was evaluated at two local and three non-local field test
sites consisting of Philadelphia Electric Company and Allegheny Power and Light
-------
Company power stations.
Both oil fired and coal fired plumes were observed.
In addition to the lidar measurements, at some sites telephotometer measure-
ments of plume transmittance and plume-to- sky contrast were made by EPA
personnel for comparison purposes.
In general, the lidar and telephotometer
determined transmittance values agreed within the accuracy expected for a given
plume transmittance as indicated by the synthetic target test results.
Plume-to-
sky contrast (plume visibility) was found to have no correlation with plume trans-
mittance because of the variability of ambient
illumination of the plume.
-------
ABSTRACT
1.
TABLE OF CONTENTS
INTRODUCTION
. . . .
......
.....
............
1. 1
1. 2
The Lidar Concept
...........
.......
.....
Measurement of Plume Transmittance by Lidar . . .
. . . .
1. 2. 1
.....
1. 2. 2
Historical Development of the Technique. .
The Single Beam, Single Shot Technique. .
.....
2.
MOBILE LIDAR SYSTEM DESIGN. .
. . . .
2. 1
2. 2
............
Vehicle and Mount Design. . .
. . . .
.......
.....
Lidar Design. . . . .
.....
.........0..
2. 2. 1
2. 2. 2
. . . .
Lidar Optics.
. . . . .
........
........
Lidar Detector and Data Handling Components. . . .
3.
LIDAR SYSTEM TESTS USING SYNTHETIC TARGETS OF
KNOWN TRANSMITTANCE. . . . . . . . . . . . . . . . . . . . . .
3. 1
3. 2
The Synthetic Targets and Their Laboratory Determined
Transmittance. . . . . . . . . . . . . . . . . . . . . . . . .
Lidar Target Te sts. .
3. 2. 1
3. 2. 2
........
.....
. . . .
. . . .
The Effect of Afterpu1sing. .
. . . .
.........
Lidar Transmittance Measurements
.........
4.
FIELD TESTS ON REAL SMOKE PLUMES. .
...........
4. 1
4. 2
Local Field Tests. . . . . . . . . . . .
...........
Field Tests in Western Pennsylvania and West Virginia. . .
4. 2. 1
......
4. 2. 2
4. 2. 3
Tests at Albright, West Virginia. . . . .
Tests at Point Marion, Pennsylvania.
........
Te sts at Springda1e, Pennsylvania. . . . . . . . . .
Page
1
1
2
2
3
7
7
9
11
14
22
22
26
26
28
44
44
46
48
52
57
-------
5.
CONCLUSIONS.
6,
. . . .
.................
......
RECOMMENDATIONS.
...............
........
7.
ACKNOWLEDGEMENTS
...0..
......
..........
8.
REFERENCES.
APPENDIX A:
APPENDIX B:
.......
.......
.............
THE LIDAR RANGE EQUA TION . . .
....08
PHOTOMULTIPLIER TUBE GATING
INVESTIGA TIONS. . . . . . . . . . . . .
. . . .
EXPERIMENTAL TEST METHODS
......
OFF-GATING METHODS AND RESULTS. . . .
Afterpulsing and Focus Electrode Gating
of the Amperex 56 TVP. . . . . . . . . . . .
Afterpulsing and Focus Electrode Gating
of the R CA 7265. . . . . . . . . . . . . . . .
External Grid Gating of the Amperex
56 TV P . . . . . . . . . . . . . . . .
. . . .
External Grid Gating of the RCA 7265
. . . .
Dynode Gating of the Amperex 56 TVP. . . .
Dynode Gating of the RCA 7265. . . . . . . .
DISCUSSION AND CONCLUSIONS. . . . . . . .
Page
61
64
68
69
A-I
B-1
B-3
B-4
B-6
B-9
B-ll
B-ll
B-13
B-17
B-17
-------
1.
INTRODUCTION
A mobile lidar (light detection ~nd Eanging) system has been constructed and
evaluated under the joint sponsorship of the Edison Electric Institute (EEl Research
Project RP39) and the Environmental Protection Agency (EPA Contract no. 68-02-
0093).
This system was built for the purpose of remotely monitoring the opacity
or transmittance (T) of stationary source smoke plumes.
It employs a pulsed ruby
laser source and was designed to operate at slant ranges of from 200 to 400 meters.
Data has been obtained to a range of 487 meters using a real smoke plume.
1. 1 The Lidar Concept
The conventional single ended lidar configuration involves a laser transmitter
sending out a short pulse of light (""' 10-8 sec. ) from a laser source in conjunction
with a receiving system mounted on a parallel or nearly parallel optical axis which
collects backscattered light from the outgoing laser pulse as it propagates away
from the system.
This collected light is focus sed on a field-limiting aperture and
is then collimated before passing through a narrow band interference filter to a
photomultiplier
detector.
Thus, backscattered light intensity at the receiving
aperture is linearly transformed to voltage acros s the photomultiplier load re sis-
tor as a function of time, and thus range.
When the lidar optical path is through the clear atmosphere the backscattered
signal at the laser wavelength is made up of two components.
These are the com-
ponent due- to molecular (Rayleigh) scattering and that due to aerosol (Mie) scatter-
ing.
A complete expression for system output voltage as a function of range is
shown as equation 15 of appendix A.
However, thi s expre s sion can be simplified
1
-------
somewhat when considering the relatively short slant ranges at which the plume
opacity lidar obtains data.
For these ranges attenuation of the beam and atmos-
pheric density changes can be neglected, with the correct expression for system
output voltage being
v
=
cASE
o
2 r2
ER
[ n (r) . (]R (1800) + m (r) . (]M (1800)J .
(1. 1)
Here c is the velocity of light, A is the area of the receiving aperture, S is the
photomultiplier sensitivity (amps /watt), E is the overall system optical efficiency,
o
E is the laser output energy, R is the load resistance,
r i s th e rang e, n (r) is
the number density of molecular scatterers, (]R (1800) is the Rayleigh backscatter-
3
ing cross section for air, m (r) is the mass concentration of aerosols (JJ gm/m )
o
and (] M (180 ) is the mas s normalized aerosol (Mie) backscattering cros s section.
If n (r) and m (r) are constant with range then the bracketed term of equation
(1. 1) is a constant atmospheric backscattering coefficient, k , and equation (1.1)
s
becomes
v
cASE
o
ER
2 r2
k,
s
(1. 2)
=
indicating a basic lIr2 dependence of the system signal.
1. 2 Measurement of Plume Transmittance by Lidar
1. 2.1 Historical Development of the Technique
Prior to the construction and evaluation of the lidar system des-
cribed in this report consideration had been given to various techniques for
2
-------
remotely obtaining plume transmittance using both active lidar techniques and
pas sive telephotometer techniques.
Basic work by W. D. Conner and J. R.
Hodkinson relative to the optical properties of plumes supported jointly by the
Public Health Service and the Edison Electric Institute is described in reference
1.
They discuss the use of lidar for plume transmittance measurements employ-
ing the technique of observing atmospheric backscattering from a range just
beyond the plume on consecutive lidar shots; first through the plume and then
adjacent to the plume.
In addition, early studies of lidar techniques for measur-
i~ plume transmittance were conducted at Stanford Research Institute 2 funded by
the Edison Electric Institute.
These studies involved the use of three basic tech-
niques for determining plume transmittance; consecutive lidar shots through and
adjacent to the plume, a dual beam lidar with simultaneous beams through and
adjacent to the plume and a single beam, single shot technique involving the obser-
vation of light scattering from just in front of and just beyond the plume.
This
latter technique was examined experimentally using an existing lidar system and
synthetic targets of known transmittance.
Following this, a contract was awarded
to the General Electric Company's Space Sciences Laboratory by the Edison Elec-
t ric Institute for the development and construction of the present mobile lidar
system employing the single beam, single shot technique.
Further, following
completion of the system the present contract was awarded by the Environmental
Protection Agency for an evaluation of the system I s performance using synthetic
targets of known transmittance as well as real smoke plumes.
1. 2. 2 The Single Beam, Single Shot Technique
Although the SRI experimental work2 had pointed out a "receiver
3
-------
paralysis" problem associated with this technique, it was felt that the system
simplifications as sociated with the technique as well as the po s sibility of getting
around the paralysis problem with receiver gating techniques dictated the single
beam, single shot approach.
The basic technique for obtaining the lidar deter-
mined one-way transmittance T is illustrated in Figure 1. 1.
L
Figure 1. I shows a drawing of a lidar return with photomultiplier
tube signal voltage plotted against time and thus range.
Although the signal volt-
age is inherently negative it is shown inverted for clarity in this drawing.
The
laser pulse is initiated in the transmitter and defines t = 0 for the system.
The
initiation of this pulse is monitored by a photodiode looking at the laser cavity pro-
viding a triggering signal for the system electronics.
Since the transmitting and
receiving system axes are parallel but not coincident and since the receiving syst-em
has a narrow field of view, the backscattered light from the outgoing laser pulse
cannot be seen by the receiving system for the first 50 meters or so of range.
Following this, the angular field of view of the receiving system combined with
the collimated divergence of the laser pulse allows gradual overlap of the two
beams with full overlap occurring at a range of approximately 130 meters (0.8 IJ
From this point onward the signal return shows a general l/r2 dependence
sec. ).
as shown by equation (1. 2) as long as the atmospheric scattering coefficient, k ,
s
is a constant.
Of course a smoke plume means a discontinuous change in k
s
which
shows up as a sharp signal spike at the range of the plume as indicated in Figure
1.1.
The amplitude of this spike can be up to 40 db above the ambient light scatter-
4
-------
U1
>
I
...J
ct
Z
(!)
-
U)
~ LASER
a.. PULSE
/
o
r-
LIDAR RETURN FROM
SMOKE PLUME
0.5
2.5
PLUME
RETURN
AMBIEN
AIR RETURN
r=
aT
A
TL=JA/B'
tp
r= et /2
2.0
t -JL see
Figure 1. 1 A sample lidar return through a smoke plume showing the single beam, single shot technique
for obtaining the one way lidar determined transmittance, T L'
-------
ing for low transmittance highly reflecting plumes.
Light scattered to the re-
ceiver from ranges greater than the plume will have been attentuated by the plume
twice, once going out and once coming back.
Therefore the received signal shows
a discontinuity at the range of the plume.
If the ambient air scattering signal from
each side of the plume is extrapolated to a common point in a 1/r2 fashion, as
shown by the dotted lines in Figure 1. 1, the ratio of the two signal levels can be
computed as A /B as shown.
The one way lidar determined plume transmittance
is then given by
T =
L
(A/B)1/2
(1. 3)
All of this yields an absolute measurement of plume transmittance
independent of laser pulse energy provided that k
s
is homogeneous in front of and
beyond the plume.
The measurement is easy in principle but in practice a phenom-
enon called photomultiplier tube afterpulsing limits the accuracy with which the
signal on the far side of the plume can be read.
This will be dealt with in detail
in Section 3 when te st
results are discussed.
6
-------
2. MOBILE LIDAR SYSTEM DESIGN
2.1 Vehicle and Mount Design
As indicated in the proposal, a Ford E-100 Econoline Super- Van was chosen
as the vehicle in which to mount the lidar system.
Previous experience with
anothe r lidar system indicated that the best way to avoid road shock damage to
mounted equipment was to rely on the vehicle's suspension, using the softe st
springing available for the anticipated load to achieve a passenger-car-like ride.
The anticipated equipment weight of approximately 1100 lbs. coincided nicely with
the 1025 lb. rated payload of the truck selected and the resulting ride is quite
smooth.
This allowed all of the system components to be hard mounted to the
truck body or floor.
In a further effort to produce smooth operation an automatic
transmis sion was specified.
Since the truck would always be operated fully loaded
an extra capacity radiator was specified and since extended storage periods were
anticipated a 70 ampere-hour battery was ordered.
The color specified was white
both for thermal reasons and because it resulted in a white interior for the truck,
improving visibility.
Figure 2.1 shows the truck in an operating configuration at the Barbadoes
Island Station of the Philadelphia Electric Company.
The lidar transmitter and
receiver can be seen protruding through the roof opening.
The cable s going to the
portable motor generators can be seen at the rear of the truck.
One operator is
at the lidar mount aiming the system using the 4-12 power rifle scope while the
other operator is seated in front of the oscilloscope controlling the data collection
and laser firing.
The two roof doors, shown in their open position, are made of
7
-------
Figure 2. 1
Mobile lidar system shown in operating configuration at the
Barbadoe s Island Station of the Philadelphia Electric Company.
8
-------
aluminum and when open expose a 6 1/2 by 4 1/2 foot opening in the roof.
This
modification was performed by the Allegheny Body Company of Philadelphia, a
fabricator of custom truck bodies.
The lidar transmitting and receiving assembly is attached to a Quick Set
Corporation Model 6222 Geared Head combined with a Model 6800 Stationary
Elevating Pede stal.
This mounting configuration is shown in the storage/travel
position in Figure 2.2 and in an operating position in Figure 2. 3.
The mount
allows 3600 rotation in azimuth with _50 to 420 elevation allowed for most azimuth
angles.
In going from the storage /travel to the operating position the whole
mount is translated vertically between 15 and 20 inches using the stationary elevat-
ing pede stal.
The stationary pedestal is bolted to a large aluminum plate which
in turn is bolted through the ribbed sheet metal truck floor at points near where
the floor is spot welded to the frame cross members.
This mounting method has
proved quite satisfactory during field use since the pointing accuracy, which need
not be high for this application, is limited by overall truck motion on its suspen-
sion rather than mount to truck motion.
The dexion brace shown supporting the
lidar in Figure 2. 2 is a removable storage /travel brace and stiffens the lidar
mount considerably for travel.
This has proved successful in combination with
the soft vehicle springing, in that the lidar has retained its optical alignment
quite well during extensive road trips including rough secondary roads.
2. 2 Lidar Design
The lidar de sign employed is conventional although refracting instead of
the reflecting optics employed in a previous system3 were used for compactness
9
-------
Figure Z. Z
Lidar in storage/travel position
~~---
\- --
-\ ' -
-\ -
~:
!1,
--
Figure Z. 3
Lidar in operating position showing lidar mount, laser water
cooler and system electronics.
10
-------
and ease of fabrication.
A block diagram of the complete system is shown as
Figure 2.4. For purposes of discussion it is convenient to divide the lidar system
into its optical and non-optical components.
2. 2.1 Lidar Optics
The transmitted laser light pulse originates in a 1. 0 cm diameter
7. 6 cm long ruby crystal installed in a Hadron/TRG Model 200B laser head.
The
laser characteristics as well as the remaining system optical characteristics are
shown in Table 2.1.
The laser is Q-switched by means of a rotating prism rear
reflector yielding 30 nanosecond wide pulses (FWHH).
The front cavity reflector
is a single sapphire resonant reflector (etalon) of about 27% reflectivity, a more
durable design than the more common dielectric coated reflectors.
Th
cavity
gain is further increased by aligning the flat ends of the laser rod with the output
reflector.
Single pulse output of up to 1. 0 joules is obtained without the use of a
dye cell accessory.
The laser pulse is then collimated using a Galilean telescope reduc-
ing its divergence from the 5 milliradians or less characteristic of the laser cav-
ity to approximately O. 5 milliradians full angle at the half power points.
The
negative lens is pIano-concave with the plane side facing the laser so that no
focussed laser reflection will cause a thermal distortion of the optical path.
The
positive lens is a 5 inch air spaced, coated achromat of 25 inch focal length.
The backscattered light is collected by a 6 inch diameter £/5 air
spaced, coated achromat and focus sed on a field stop with 4 milliradian field of
vIew.
Light emerging from this field stop is collimated by a 2.14 inch focal length
11
-------
......
N
PMT-PS
PULSE GEN.
(FOR PMT
GATE)
DOUBLE BEAM
SCOPE AND
CAMERA
LASER PS
WATER
COOLING
SYSTEM
MOTOR
GEN.
~
....-
4-12 X AIMING SCOPE I 6 IN. DIAM., f/5)
C FIELD STOP f
IF
PM TUBE I
DETECTOR I
FROM LASER MONITOR
RUBY LASER
~
5 IN. DIAM., f/5
IF. THERMALLY CONTROLLED
INTERFERENCE FILTER
Figure 2.4 Block diagram of mobile lidar system
-------
TABLE 2.1
Lidar Optical Characteristics
Laser
Manufacturer and Model:
Type:
Output Wavelength:
Maximum Output Energy:
Pulse Width (FWHH):
Ruby Rod:
Beam Divergence:
Repetition Rate:
Cooling:
Laser Collimation
Collimated Beam Diameter:
Collimated Beam Divergence:
Receiving Optics
Type:
Objective Lens:
Collimating (eye)
Field of View:
Lens:
Interference Filter
Manufacturer:
Type:
Diameter:
Transmission Peak Wavelength:
Transmission Center Wavelength:
Thermal Shift:
Bandwidth (FWHH)
Peak Transmittance:
Residual Transmittance:
Filter Rejection:
Filter Tilt in Lidar:
Hadron/TRG Model 200B /104A
Rotating prism Q- switched ruby
6943A
1. 0 joule Q- switched
< 30 nanoseconds
1. 0 cm x 7. 6 cm
< 5 mrad. (full angle - 1/2 power)
o - 3 ppm
Deionized water
lOcm
'" 0.5 mrad. (full angle - 1/2 power)
Refracting
6 inch diameter,f/S
2. 14 inch focal length
4. 0 mrad. full angle
Infrared Industries
3 Cavity Interference
2 inche s
6946.0A (250 C)
6943.4A (250 C)
+ O. 2 A/o C
12.0 A
0.66
< 0.01%
Far uv to ~ 1. 0 micron
00 (perpendicular to radiation)
13
-------
achromat before it passes to a 3 cavity narrow band pass interference filter
placed normal to the optical axis.
The filter is temperature controlled by the
laser cooling water (in parallel with the laser head) and is generally held at from
o 0
20 C to 25 C.
The detailed filter characteristics are listed in Table 2.1.
Ther-
mal separation of the filter from the laser coolant temperature can be achieved by
plugging in a temperature controlled electrical heater tape wrapped around the
filter holder.
o
This heater has been preset to control the filter at 30 C.
It will
do this even with laser coolant flowing at 200 C.
After pas sing through the filter
the backscattered light encounters the photocathode of the photomultiplier tube
where primary signal detection occurs.
A 4 to 12 power riflescope for aiming the lidar system is attached to
the receiving lens tube and completes the lidar optics.
This riflescope can be
seen in Figure 2.5 showing the complete transmitting and receiving assembly.
All of the optical components are mounted on a common 3/4 inch thick aluminum
base plate.
The receiving system has built-in lockable adjustments for azimuth
and elevation angle with respect to the transmitting system.
These are used for
aligning the two optical axe s.
Figure 2. 6 shows the transmitting side of the sys-
tem with the laser cavity, the output reflector (etalon) and the negative collimating
lens exposed.
2. 2. 2 Lidar Detector and Data Handling Components
The major effort in designing, constructing and evaluating this lidar
system was centered around the photomultiplier tube detector.
As has been men-
honed in Section 1. 2, in order to make the lidar transmittance measurement, it is
14
-------
Figure 2.5 Lidar transmitter and receiver in storage/travel position.
.. ~ .~ - .....iIL.. ..-::
~ (' , ~~ c,~.;,.,..
~ -- ..'"
c... ., .- --) - " ':
, ,
, " .
. \ ,--!:-:=:--\- t~ I '.., I
~ "0 '~..- .:-
11
Figure 2. 6
Lidar transmitter and receiver in elevated position with laser
cavity cover and output window - negative lens cover removed.
15
-------
necessary to look at a low level signal from the ambient air light scattering on
the far side of a plume with good accuracy immediately following the intense
return from the plume itself.
In order to do this either the detector and the am-
plifying electronics would have to recover from a signal range of up to 40 db
within 100 to 200 nanoseconds or the detector and amplifying electronics would
have to be prevented from responding to this intense signal by gating for the dura-
tion of the signal.
It is difficult to obtain amplifying electronic s that will accept
such a wide dynamic signal range with the required recovery time even though
the stated equipment recovery time per decade when projected over four decades
would indicate this.
Thus, it was decided that the amplifying electronics should
not see the main plume pulse.
Further, as has been mentioned, earlier work at
Stanford Research Institute 2 had shown the existence of photomultiplier tube after-
pulsing related to receiving an intense signal from a plume or an opaque target.
They had eliminated these afterpulsing effects by applying a 50 nanosecond wide
gating pulse to the focus electrode of their Amperex 56-TVP photomultiplier tube.
Accordingly, it seemed best to use this gating technique (for which 30 db plume
signal suppression was claimed) since it would get rid of afterpulsing and ampli-
fier saturation effects.
An Amperex 56-TVP tube was purchased and a proper photomulti-
plier tube tapered voltage divider chain was designed and implemented with con-
sideration given to linear response to high peak currents (near range scattering),
space charge limiting at low voltage per stage, fast response, and coupling a 100
volt gating pulse to the focus electrode.
The tube's performance was then evalu-
16
-------
ated in the laboratory using a pulsed light source in experiments described in
Appendix B.
Although the ungated tube response to high intensity light pulses was
linear to the level expected, the gating response was unacceptable.
These results
are described in detail in Appendix B but the general results were that 30 db off-
gating could not be attained and that increasingly slower tube on- recovery,follow-
ing removal of the gating pulse,occurred as the off-gating pulse duration was
lengthened beyond 100 nanoseconds.
It had already been concluded that a minimum
practical off-gate width for real plumes, considering plume thickness and laser
pulse shape, would be about 200 nanoseconds, so this focus electrode gating tech-
nique had to be discarded.
An alternative off-gating technique, mentioned by
Stanford Research Institute2, of using a wire mesh grid externally covering the
photocathode was also tried in the laboratory even though it was suspected from
the focus electrode results that any technique applying strong electric fields near
the photocathode would damage the on-response of the tube for realistic off-gating
pulse durations.
This was confirmed and it suggested that the semi-conductor
nature of the photocathode prevented the necessary rapid rate of charge flow re-
quired for fast on-response following the integrated result of a charge redistri-
buting electric field.
Since the detailed mechanism of afterpulsing was not understood and
since,
in any case, it was desirable to prevent the amplifying electronic s from
receiving the effect of the strong plume pulse it was decided to off-gate the tube
by applying the gating pulse to one or more of the early dynodes. The method
settled on was to apply a 100 volt pulse to dynodes 2 and 6 which produced off-
17
-------
gating of greater than 45 db at the projected operating tube voltage of 1500 volts.
This then was the configuration first tested on the lidar system and resulted in
being the final configuration as well.
It was hoped that this off-gating procedure would eliminate after-
pulsing but it did not.
Lidar returns from a white building wall used to simulate
of plume of zero transmittance and high reflectance showed adequate suppression
of the main plume return (shown in detail in Section 3) but also showed a large
afterpulsing signal occurring at a time later than the wall signal where the signal
return should have been zero.
It was then clear that afterpulsing originated
somewhere between the photocathode and the beginning of the dynode structure
leading to the present supposition that the large number of photoelectrons liber-
ated from the photocathode during the main plume return pulse cause a sufficient
amount of ionization of the residual gas or surface ionization in this region to
account for the afterpulsing.
These ions drift back to the photocathode with a
time delay from their formation dictated by their mass and the electric field,
and cause secondary electron emis sion there.
The groups of secondary elec-
trons formed then make up the afterpulsing which can probably be prevented only
by never allowing the main plume pulse electrons to leave the photocathode. Again,
the only methods for doing this available with the Amperex tube had been tried
with unacceptable degradation of the on-recovery response of the tube.
It was felt that afterpulsing might depend critically on the detailed
design of the photocathode, focus electrode, and first dynode region.
Thus, an
RCA 7265 photomultiplier tube was substituted and all of the three gating tech-
18
-------
niques were tried.
Although the results (shown in Appendix B) differed in detail
from those obtained with the Amperex tube the essential conclusions were the
same.
Not only was afterpulsing unavoidable but it was more continuous follow-
ing the main plume pulse allowing no minimal afterpulsing region for measure-
ment as was the case with the Amperex tube.
Thus, the RCA tube was removed
and the Amperex tube with dynode gating was chosen as the final system configu-
ration.
The remaining detector and system data handling characteristics
are shown in Table 2. 2.
Data recording is done using Polaroid photographs of
oscilloscope traces.
A double beam oscilloscope is used so that the far side sig-
nal from plumes of low transmittance can be displayed at a higher amplification
improving reading accuracy.
For those cases where single beam presentation
is sufficient the remaining beam can be used to display the output of the laser
energy monitor if desired.
This is handy for keeping track of changes in laser
output shot to shot as well as providing the most accurate zero time reference
for range measurements.
The output energy is monitored by means of a silicon
photodiode collimated on the internal volume scattering from the rotating prism
and filtered to reject flashlamp light.
The photo diode output is available either
in integrated form as a calibrated energy monitor or non-integrated as a trigger
signal for the system electronics.
When both oscilloscope beams are used for lidar signal presentation
the oscilloscope is triggered externally by the non-integrated output of the energy
monitor.
When the integrated output of the monitor is displayed triggering
19
-------
TABLE 2.2
Lidar Detector and Data Handling Characteristics
Laser Monitor
Detector Type:
Receiving System Detector
Manufacturer and Type:
Photocathode Type:
Receiving System Detector Gating
Gating Method:
Maximum On/Off Ratio:
Optical Gating Response:
Gating Pulse Generator:
Minimum Pulse Delay:
Pulse Width:
Data Presentation and Recording
Oscilloscope:
Oscilloscope Preamplifers:
System Risetime:
System Maximum Sensitivity:
HP 4220 PIN Silicon Photodiode
Amperex 56 TVP Photomultiplier Tube
S-20
+ 100 volt Pulse Applied to Dynode s 2 and 6
of Photomultiplier Tube
45.4 db at 1500 volts on PMT
< 50 nanoseconds
Hewlett-Packard Model 214A
500 nanoseconds from Laser Trigger
Continuously Variable
Tektronix #551 (double beam)
Tektronix Type H (two)
25 nanoseconds
0.005 volts/em
20
-------
is internal.
In either case, the gating pulse generator is then triggered by the
gate out of the oscilloscope.
This pulse generator is capable of putting out a
single 15 nanosecond risetime, .2:. 100 volt pulse into a 50 ohm load with continu-
ously variable pulse delay and continuously variable pulse duration.
The mini-
mum delay from receipt of trigger is 500 nsec, which, when coupled with an
approximately 100 nanosecond oscilloscope delay from receipt of laser monitor
trigger signal, yields a minimum range for gating of approximately 90 meters.
In practice, the proper gating delay and pulse width is set by first observing the
plume return ungated at low sensitivity and then setting the pulse generator for
the correct delay.
Finally, the lidar system is powered by two 2.5 kw portable motor
generators which are carried in the truck but are removed for operation.
The
system electronics (oscilloscope, photomultiplier tube power supply and gating
pulse generator) are operated on one motor generator while the laser and laser
water cooler are operated from the other.
This was done to allow the system
electronics to operate on surge free power.
21
-------
3. LIDAR SYSTEM TESTS USING SYNTHETIC TARGETS OF KNOWN TRANS-
MITTANCE
The mobile lidar system was evaluated using synthetic targets of known trans-
mittance (laboratory measured) to simulate smoke plumes.
Although the original
intent was to perform these measurements at 200, 300 and 400 meters range, the
400 meter range was not used because of the difficulty of obtaining a suitable 400
meter path length.
Instead data was taken at 211 and 319 meters with a greater
variety of targets than had been originally planned.
3.1 The Synthetic Targets and Their Laboratory Determined Transmittance
Since it was anticipated that testing would be done at 400 meters range, a
target aperture of 42 inches was designed which subtended 2. 5 milliradians (full
angle) at that range.
It was estimated from the laser specification of less than 5
milliradian uncollimated output divergence (full angle, half power) coupled with
an experimentally determined divergence curve from an optically similar laser
cavity, that the collimated output beam would have greater than 98 per cent of its
energy contained within 2. 5 milliradians.
This would have meant careful aiming
at the 400 meter range but allowed a wide variety of target materials because of
the common availability of up to 48 inch widths.
Although it was originally planned that targets of 0.50, O. 70 and 0.90
transmittance would be tested, it was easy to expand this to a wider range using
aluminum screening, glass and plexiglass in both single and double target config-
urations.
Data was also taken using white and black opaque targets (T = 0), and
using the clear target holder aperture (T = 1. 00).
The transmittance of the various targets was determined at the laser wave-
22
-------
length in the laboratory by illuminating the targets from one side with a collimated
white light source while viewing the source from the other side of the target with
the lidar receiving system.
The targets were alternately placed in and out of the
beam while the photomultiplier output was read on a Keithly electrometer.
This
procedure was repeated with the source turned off to see if the front lighting of
the targets introduced a significant error in the measurement.
It did not, the
maximum contribution being less than 1 percent of the reading with the source on.
The targets used and their laboratory-measured transmittance values are shown
in Table 3. 1.
In all cases except the glass, the laboratory measured values agreed well
with the expected re sults.
The Plexiglass-G is listed by Rohm and Haas, the
manufacturer, to be about O. 92 at this wavelength.
The transmittance of the wire
screening could be calculated from the wire size and spacing and was in good
agreement with the measured values.
When questioned, Libby-Owens-Ford
stated a value of 0.819 for the transmittance of their Parallel-O-Float glass in 1/4
inch thickness, a value significantly lower than that measured in the laboratory.
However, nothing could be found in error in the measurement technique and a
repeat measurement confirmed the previous result.
In addition, lidar measure-
ments made later on indicated that the laboratory measurement was correct so it
was decided to accept the value of o. 861.
Although the lidar measurements could be directly compared with the labo-
ratory transmittance value s for the diffusely reflecting screen targets, a subtle
difference existed in the case of the specularly reflecting glass and plexiglass
23
-------
TABLE 3.1
Synthetic Targets Used for Lidar Evaluation
Target Material
Plexiglass-G (1/4 inch thick)
Plexiglass-G (1/4 inch thick)
L-O-F Parallel-O-Float Glass (1/4 inch thick)
L-O-F Parallel-O-Float Glass (1/4 inch thick)
Aluminum Insect Screen
Aluminum Insect Screen
50% Open Area Aluminum Screen
(Black Anodized)
50% Open Area Aluminum Screen
30% Open Area Aluminum Screen
24
Laboratory Transmittance (6943 ~
0.915
0.915
0.861
0.861
0.640
0.645
0.525
0.515
0.288
-------
targets.
The specular surface reflections from these materials projected a
known fraction of the laser pulse into the reflected direction (deliberately placed
off-axis) on the near side of the target.
A small fraction of the light scattering
return from this reflected path was reflected towards the receiving system and
added to that from along the optical axis on the far side of the target, thus influ-
encing the transmittance measurement and giving too high a value.
An analysis
of this situation was made for the case of a first surface reflectivity,
ex, and an
absorption loss factor per 1/4 inch thickness, {3, assuming nearly normal mCl-
dence.
The result of this analysis is an expression for the backscattered lidar
signal ratio, A/B, of the scattered signal from the far side of the target (includ-
ing near side reflections) to the scattered signal from the near side of the target
which is
A/B =
2 2 2 22 24
(1 +8 -4ex8 +60' B +3ex R
(3. 1)
where higher order terms have been dropped when numerically negligible.
Using
this relationship and the correct values of
ex (= 0.0423) and f3 (= O. 9387) for the
glass used, one can compute A/B = 0.7466 with T = (A/B)1/2 = 0.8641.
Here 8
has been calculated using the laboratory value of T (- O. 861) and the published
value of 0' (= 0.0423).
Thus the lidar system should measure a transmittance
value O. 36 percent higher than the true one way transmittance of O. 861, a negli-
gibly small difference.
An even smaller difference exists for the case of plexi-
glass where ex~ 0.04 and 8~1. O.
Therefore the lidar determined transmittance
values were not corrected for this effect.
25
-------
3. 2 Lidar Target Tests
3. 2.1 The Effect of Afterpulsing
A phenomenological description of afterpulsing has been given in
Section 2. 2. 2.
Here, the effect of afterpulsing on the lidar signal return is dis-
cus sed.
This is best illustrated by looking at Figure 3. 1 which shows a typical
lidar return at two amplifications from an opaque. highly reflecting target.
The
photomultiplier tube signal is inherently negative so the lidar returns are negative
going on the oscillogram rather than as drawn in Figure 1. 1.
The lower trace, photographed on automatic sweep either before or
after the lidar shot, is the differentiated gating pulse, and shows the position and
duration of the photomultiplier gating pulse.
The uppermost trace shows the lidar
signal increasing from zero as beam overlap begins, reaching an off scale maxi-
mum and then beginning to fall off in a 1/r2 manner until the tube is gated off at
which time the signal goes to zero.
The white wall "plume" signal then appears
even though the response of the tube is down by about 45 db.
From this point on
the lidar return should be zero, and any further signals must be spurious. Immed-
iately following the removal of the off-gating pulse, early afterpulsing appears
and begins falling off to a lower level until the main afterpulsing spike occurs at
about 600 nanoseconds.
The early afterpulsing and the intermediate minimum can
best be seen on the middle trace which has a factor of 10 gain increase over the
upper trace.
It is clear, that, although the signal shows a 250 to 300 nanosecond
period of nearly constant minimum value, that value is not zero as it should be.
The form of this afterpulsing is quite repeatable, with the main afterpulse always
26
-------
LIDAR RETURN FROM WHITE OPAQUE TARGET
N
-..J
GATED PLUME SIGNAL) (MAIN AFTER PULSE
LIDAR { ............ 0 5 VIcm
RETURN. --.....
.......lIiiii 0.05 V/em
..........
..........
GATE POS. ========.0.05V/em
0.5 II- see/em
Figure 3. 1 A lidar return from a white building wall at a range of about 200 meters. This simulates a
plume of zero transmittance and high reflectance.
-------
significantly larger in amplitude than the early afterpulsing.
Thus, the
transmittance value for targets where afterpulsing is obviously present (as in-
dicated by the main afterpulse) is obtained by using data just praceeding the main
afterpulse by 200 to 300 nano seconds.
It will be seen in Section 3. 2. 2 that as the
target reflectance decreases, the afterpulsing-induced transmittance error de-
creases to the point of becoming negligible at T ~ O. 80 or so.
A comparison between lidar returns from white and black opaque
targets is shown in Figure 3. 2.
The plot is double logarithmic so that the basic
l/r2 lidar signal return will appear (for a homogeneous scattering atmosphere)
as a straight line of slope -2.
The sloped lines through the data points are con-
structed to have a -2 slope and it can be seen that up to the target range the lidar
performance is l/r2.
The targets used were white painted plywood and a double
layer of black felt over plywood.
It is obvious that there is a dramatic decrease
in afterpulsing going from the white to the black targets, yielding only a factor of
2 reduction in lidar measured transmittance.
Closer inspection reveals that
while the main afterpulse peak dropped bya factor of about 20, the afterpulsing
level near the minimum decreased by only a factor of 4.
However, the general
shape of the afterpulsing remained the same.
3. 2. 2
Lidar Transmittance Measurements
Two target holders were constructed with a 42 inch diameter circu-
lar aperture and were placed on the roof of a small building at the Valley Forge
Space Center and spaced 23 feet apart to simulate plume depth.
These are shown
on the building roof in Figure 3. 3 as used for the 211 meter range data.
A corner
28
-------
1.0
f/)
~
o
>
I
..J 0.1
c
z
(!)
-
f/)
a::
c
c
-
..J
C
1&.1
N
:J 0.01
c
:.
a::
o
z
LIDAR RETURN FROM OPAQUE
TARGETS - 211 METER RANGE
,
i
~
j
o WHITE PAINTED
PLYWOOD
xSLACK FELT
Tw=Tb==O
TLW./AW/S".0.247
.TLb =/Ab/S'.0.121
0.001100
Figure 3. 2
1000
RANGE-METERS
10.000
A comparison of normalized lidar returns from white and black
opaque targets illustrating the relative afterpulsing and resulting
values of lidar determined transmittance, T L'
29
-------
Figure 3. 3
--- - -----1
Test target holders in position on a building roof at the Valley Forge
Space Center. This location was used for the 211 meter range data.
The white plywood target is shown in place and was used for align-
ment purposes.
30
-------
of the building was used, allowing both targets to be at the roof edge with only the
23 foot path length over the building itself.
Although the wisdom of this arrange-
ment was not foreseen, later tests at 319 meters range using a single target
located back from the roof edge indicated that atmospheric homogeniety along
the optical path was difficult to obtain when a large portion of the path was over
and cIo s e to building roofs.
For a major portion of the data taking the lidar system was placed
211 meters from the target configuration shown in Figure 3. 3.
The target holders
were arranged so that any specular target reflections were off axis and the tar-
gets were not parallel to each other.
The lidar system was aligned, using the
white gridded target shown, so that the transmitting and receiving axes crossed
at 211 meters.
This was not necessary in principle, since full overlap of the
fields and consequent l/r2 performance begins at about 130 meters range with
the axes parallel, but it facilitated aiming the system and was easy to do.
A first check on system performance was made by firing through
the clear aperture of both target holders, effectively measuring a target with
T = 1. 00.
The results of this shot are shown in Figure 3. 4 with the actual oscil-
log ram shown above the double logarithmic plot of the data.
The line through the
2
data points was constructed with a slope of -2 and shows the l/r performance
on both sides of the target location which is shown by the vertical line.
The
transmittance measured is unity indicating that the laser pulse was not inter-
cepted by the target holder, the off-gating did not disturb the measurement, and
the atmospheric scattering was locally homogeneous on both sides of the target.
31
-------
en
t- 0.6
...J
o
>
I
...J 0.4
~
z
(!)
-
en
0:
~
c 0.2
-
...J
1.0
TL =J A/S = 1.00
TARGET
0.1100
Figure 3.4
200 400 600
RANGE-METERS
1000
Lidar shot through the clear target a t (
per ure T = 1 00) t 211
meters range. . a
32
-------
Figure 3.5 shows a typical oscillogram of a lidar shot through two
glass targets.
Note the absence of any afterpulsing.
Again the double log data
2
plot shows good l/r performance on both sides of the target and the measured
lidar-determined target transmittance (T L) of O. 757 agrees well with with the
laboratory value (T) of 0.741.
A lidar return from a target with high diffuse reflectance is shown
in Figure 3. 6.
The target in this case was two bright aluminum screens with
individual transmittances of 0.640 and 0.645 yielding an overall calculated trans-
mittance of 0.412.
The reflectance of this target configuration was large enough
that the laser pulse could be seen hitting the target with the aid of the sighting
riflescope at 10 power even though the target was sunlit.
Even though suppressed
45 db by the off-gating of the photomultiplier tube, the main target return pulse
can be seen approximately 3.1 oscillogram centimeters from the start of the
sweep followed by the start of the main afterpulse at 4. 3 oscillogram centimeters
from the sweep origin.
Far
side of the target data was taken in the interval
between removal of the off-gating pulse and the start of the main afterpulse and,
as can be seen from the plot, it does not exhibit 1/r2 behavior.
This is due to
the presence of the early afterpulsing described in Section 3. 2.1.
2 .
The l/r lme
was drawn through the far side data points in the manner shown in an effort to
weight most heavily those points near the target, a policy used throughout the
data reduction process even though in this case it gives a worst case answer. It
will be seen later on that the problem of thermal lensing with real smoke plumes
makes this choice of data weighting advisable.
The afterpulsing, then, introduces
33
-------
2.0
en
!J
o
>
I
..J
c:r: 1.0
z
C)
-
en 07
a:: .
c:r:
Q
:::i 0.5
0.3100
Figure 3.5
T L = IA/B' = 0.757
TARGET
200 400 600
RANGE-METERS
1000
Lidar shot through two glass targets (T = 0 741) t 211
. a meters range.
34
-------
1.0
0.6
U)
~
o 0.4
>
I
..J
c(
Z
(!)
-
U) 0.2
a:
c(
Q
-
..J
0.1
0.06100
T= 0.412
TARGET
TL = JA/B = 0.491
1000
GATE POSe
Figure 3.6 Lidar shot through two screen targets (T = 0.412) at 211 meters range.
35
-------
a substantial error giving a lidar determined transmitt ance, T L' of o. 491 as
12 If a Parallel 1/ r 2 line is
compared with the laboratory measured value of 0.4 .
constructed through the minimum amplitude data points, a value of T L = 0.462
is obtained, still yielding a substantial error.
A summary of the averaged results for all of the synthetic target
testing is shown in Table 3. 2.
With the exception of the clear aperture and opaque
target shots, these data represent the averaging of from two to four lidar shots
per data point.
As can be seen, results are shown both with and without off-gating.",
The results for the 211 meter range are shown in Figure 3. 7, the
45 degree line through the origin indicating perfect agreement.
The most obvious
feature of the results is that off-gating the photomultiplier tube significantly im-
proves the system accuracy even though, as mentioned previously, it does not
prevent afterpu1sing.
This improvement must be due to not overdriving the am-
plifying electronics when off-gating is used.
Although the stated and measured
risetime per decade of 25 nanoseconds yields only a 75 nanosecond response time
when extrapolated to 3 decades, it is clear that this extrapolation cannot be made
over such a wide dynamic range and that the real recovery time over this signal
range 1S longer.
The detailed mechanism of this slow recovery was not identified.
Next, it can be seen by considering the open circled points connect-
ed by the curve, that as the target transmittance decreased there was a continu-
ous increase in measurement error.
This is really due to the attendant increase
in target reflectance and thus, afterpu1sing.
The solid circle points illustrate
the reduction in measurement error obtained at a given target transmittance by
36
-------
TABLE 3. 2
Averaged Results of Lidar Measurements of Target Transmittance
Target and
Configuration
Target
Rang e
Lab.
Transm.
Clear aperture
1 Plexiglas s sheet
" II "
211 m
1. 00
0.915
II
"
"
1 glass sheet
"" fI
"
0.861
"
II
1 - insect screen
II
0.640
"
II
II
II
"
1 - 50% black screen
"
0.525
"
"
II
If
II
"
1 - 50% screen
0.515
II
"
II
"
"
'II
1 - 30% screen
0.288
II
II
II
II
"
II
2 P1exig1ass sheets
"" H
"
0.837
II
"
2 glass sheets
"" "
"
0.741
"
"
2 - insect screens
"
0.412
II
11
"
II
II
White plywood
Black felt
"
o
If
II
Clear aperture
1 Plexiglas s sheet
II II "
319 m
1. 00
O. 915
"
II
"
1 - insect screen
II
0.640
II
'II
"
II
II
37
Gated
Off ?
Avg. lidar
T ransm.
Yes
1. 00
0.956
0.925
No
Yes
No
Yes
0.891
0.861
0.758
0.672
No
Yes
No
Yes
0.609
0.561
0.685
0.573
No
Yes
No
Yes
0.521
0.409
0.914
0.880
No
Yes
No
Yes
0.762
0.742
0.659
0.506
No
Yes
Yes
0.247
O. 121
Yes
No
0.961
0.959
0.908
No
Yes
No
Yes
0.708
0.648
-------
-I
..
1&1
o
Z
~ 1.0
I-
-
2
en
z
C 0.8
a:
I-
a:
C
Q 0.6
-
..J
Q
I&J
C!)
C O~
a:
1.&.1
~
0.2
Figure 3. 7
SYNTHETIC TARGET TEST RESULTS
211 METER RANGE
o NON-BLACK TARGET} GATING
. BLACK TARGET USED
o NO GATI NG
00
o
[]
[]
Q2 OA- 0.6 08 1.0
LABORATORY TRANSMITTANCE-T
[]
~ AVG. DEVIATION
Averaged lidar determined target transmittance T , vs. laboratory
determined transmittance, T, with and without of{.:gating at 211
meters range.
38
-------
decreasing target reflectance to a minimum.
Since the unblackened targets
were either white or bright aluminum, the two curves shown probably represent
upper and lower bounds on the lidar system's measurement error for the range
o :0:;; T :0:;; 0.6.
The remaining open circle data points lie generally along the line
of perfect agreement and are for the specularly reflecting glass and plexiglass tar-
gets where afterpulsing should not have occurred. An exception is the double plexi-
glass target result (T:: O. 837) which clearly falls above the line.
In the case of both
plexiglass target configurations (T = 0.837 and T = 0.915), off-gating yielded a
definite decrease in system error.
Afterpulsing is clearly visible, although
small, on the osci11ograms corresponding to the double plexiglass target measure-
ments although it cannot be seen on the single target osci11ograms.
The only ex-
planation here is that the plexiglass was cleaned repeatedly with Windex and a
soft cloth, and may have become sufficiently scratched to have a diffuse compo-
nent of reflectivity giving rise to the error.
Before discus sing the similar plot of test results for the 319 meter
range it is instructive to look at Figure 3.8 which shows a lidar shot through the
clear single target holder aperture (T = 1. 00) similar to that shown in Figure 3. 4
for 211 meters range. It is clear from the plot that the system exhibited 1/r2
performance for extended ranges on both sides of the target holder.
However,
the measured transmittance was O. 961 as opposed to unity.
Further, it can be
seen that a deviation from 1/r2 performance occurred just before the target range.
This all points out what, in retrospect, was a poor choice of target location.
39
-------
The target holder was placed approximately 10 meters from the edge of a building
roof on a large flat portion of the roof that extended for more than 100 meters
beyond the target; that is, the target range occurred just beyond a sharp discon-
tinuity in the terrain.
Prior to the roof edge the optical path was up to 60 feet
above a parking lot while once beyond the roof edge it was only a few feet from
the roof top.
Although the effect is in the right direction and although it was a
bright sunny day, an unreasonably large temperature increase over the roof is
required (...... 450F) to explain the drop in scattering signal.
Close examination of the original oscillogram reveals the presence
of a sharp signal spike of about 1. 4 cm amplitude relative to the ambient signal
precisely at the target range (4.6 cm from the start of the sweep). Clearly some
of the beam was intercepted by the black target holder.
However, interception
of the entire beam by the black felt target yielded a signal more than 200 times
the ambient atmospheric scattering level (it produced saturated photomultiplier
tube output of approximately 20 volts, thus the light scattering may have been
much more than 200 times greater) while this interception provided an additional
signal level only O. 7 time s the ambient level.
Thus the resulting beam energy
los s through the aperture must have been less than 0.5 percent yielding a mini-
mum transmittance of T = o. 995.
A more likely explanation for T L < 1. 0 is that
the building roof acted as a poor source or a better sink for particulate aerosols
than the parking lot, causing a drop in light scattering over the roof.
Figure 3. 9 shows the results of the synthetic target tests at 319
meters.
Had it not been for the clear aperture result discussed above, the data
would look quite good and consistent with those shown in Figure 3. 7.
Again the
40
-------
0.7
(I)
~ 0.5
o
>
~
cr
z 0.3
C!)
-
(I)
~ 0.2
Q
-
.J
1.0
0.1
100
Figure 3.8
ENERGY
MONITOR
LIDAR
RETURN
T L =J A/Be =0.961
A B
200 400
RANGE-METERS
600 1000
Lidar shot through the clear target aperture (T = 1. 00) at 319
meters range.
41
-------
...J
....
I
1&.1
(.)
Z
~
I- 1.0
-
::E
tn
Z
ct
~ 0.8
....
~
ct
3 0.6
Q
1&.1
(!)
<[ 0.4
~
LLI
~
0.2
Figure 3. 9
SYNTHETIC TARGET TEST RESULTS
319 METER RANGE
o GATING USED
[J NO GATING
~ AVE. DEVIATION
00
02 OA 0.6 0.8 1.0
LABORATORY TRANSMITTANCE - T
Averaged lidar determined target transmittance, T , vs. laboratory
determined transmittance, T, with and without off-~ating at 319
meters range.
42
-------
plexiglass target data (T = 0.915) shows an improvement through off-gating al-
though no visible afterpulsing appears on the oscillograms, while the single in-
sect screen target (T = 0.645) result shows an expected improvement through
off-gating with visible afterpulsing on the oscillograms.
It is hard to say to
what extent the "roof effect" influenced the target results.
All of these data
were taken on one afternoon within a period of about 1 1/2 hours and it is tempting
to simply "correct" the results by dividing through by 0.961.
However, since
only one clear aperture oscillogram was taken, it is difficult to be sure that this
effect remained constant throughout the test period.
It seems best to simply
accept the additional error and rely on the 211 meter range results to give the
be st indication of system capability and accuracy.
43
-------
4. FIELD TESTS ON REAL SMOKE PLUMES
4. 1 Local Field Tests
Two separate local field tests were made at Philadelphia Electric Com-
pany power stations near the General Electric Valley Forge Space Center.
The
first tests were made shortly after completion of the equipment in Aprill971 for
the purpose of initial checkout of the system using a real plume.
This was done
with the cooperation of the Philadelphia Electric Company at their Barbadoes
Island power station near Norristown, Pa.
The site is shown in Figure 1.1.
The equipment was operated, without difficulty, for several hours while shots
were taken through both an oil fired and a coal fired plume.
Much of the time
was spent assessing system operation and alignment as well as determining the
effect of varying the off-gating pulse width in search of an optimum value for real
plumes
and as a result not much real transmittance data was obtained.
Only a
few oscillograms were reduced indicating a transmittance of approximately 0.65
for the coal fired plume and nearly unity for the essentially invisible oil fired
plume.
A second field test, in which Mr. W. D. Conner and Mr. N. White of EPA
participated, was made on 8/26/71 at the Cromby Station of PECO near Phoenix-
ville, Pa.
Two stacks were available, again one coal fired and the other oil
fired.
The coal fired plume was clearly visible against a blue sky and visual
judgement by those present brought an estimate of less than 0.50 transmittance
The oil fired plume was visible only through the refractive effects of the hot gases. ,
The lidar system was fired through both plumes both with and without photomulti-
44
-------
plier tube off-gating.
The results of these measurements are shown in Table
4. 1 below.
Table 4. 1 - Local Field Test Results Cromby Station - Philadelphia Electric Co.
Type of Oscillogram Trans-
Stack No. mittance Gating Range
Oil Fired 8/26/71-3 0.930 No 225 m
Oil Fired 8/26/71-4 0.956 Yes 225 m
Oil Fired 8/26/71-5 1. 000 Yes 225 m
Oil Fired 8/26/71-6 0.915 No 225 m
Coal Fired 8/26/71-8 0.369 No 210 m
Coal Fired 8/26/71-12 0.148 Yes 210 m
------------ --_.---
No afterpulsing was observed for the lidar shots through the oil fired
plume, and probably the gated and non-gated shots are equally valid measures of
the plume I s transmittance.
The variation in transmittance was probably real
and related to the periodic rapping of precipitators to dislodge collected particu-
late matter.
Some of this dislodged material goes up the stack and, on denser
plumes, can be seen as dark puffs of smoke.
On the other hand, strong after-
pulsing was observed on the lidar returns from the coal fired plume.
Compari-
son of the gated and non-gated results shows the usual large improvement in sys-
tem performance when gating was used.
Because of the reduced accuracy of the
measurement at these low transmittance values (as shown in Figure 3.7), the
value of T = 0.148 can only be regarded as an upper bound for the real value.
L
It was clear from this measurement that visual plume appearance was not strongly
correllated with plume transmittance or opacity.
45
-------
4.2 Field Tests in Western Pennsylvania and West Virginia
Following the lidar evaluation using synthetic targets, the system was
taken on a 3 day field trip to West Virginia and western Pennsylvania.
Specific
test sites had been previously selected by Mr. W. D. Conner of EPA with the
cooperation of the Allegheny Power and Light Co.
The three sites selected were
all coal burning power stations and were specifically the Albright Station at
Albright, West Virginia, the Fort Martin Station near Point Marion, Pa., and
the Western Penn Station in Springda1e, Pa.
Two of the sites selected, Albright and Western Penn, had stacks situ-
ated in valleys in such a way that te1ephotometer plume transmittance measure-
ments could be made simultaneously with the lidar measurements.
The te1e-
photometer technique for plume transmittance measurements is described in
detail in reference 1 and is best understood by examining Figure 4.1
Plume-to-sky contrast can be obtained for any plume regardless of p1ume-
hill-sky geometry simply by a comparison of the luminance of the sky through
and directly adjacent to the plume and is given by
C =
(I - I ) /1
s sp s
(4.1)
where Is is the sky luminance and I is the apparent sky luminance seen through
sp
the p1ume.I will be the sky luminance attenuated by the transmittance of the
sp
plume plus the luminance of the plume itself due to the scattering of ambient light
by the plume.
It is this latter component that disallows plume-to- sky contrast from
being a good measure of plume transmittance or opacity.
In order to remove this variable plume luminance factor, it is neces sary
46
-------
P LU ME
@ <9 SKY
@
C. I. - I.p
I,
Tt = I.p-Ihp STACK
1.- Ih
POWER PLAN T
Figure 4. 1
The telephotometer technique of measuring plume-to-sky contrast,
C, and plume transmittance, T , by comparison of the luminance
of the circled areas shown. t
47
-------
to arrange the geometrical situation shown in Figure 4. 1 where the change in
luminance of two separate backgrounds (sky and hill) caused by the plume can be
compared and the effect of plum~- scattered ambient light can be factored out.
In this situation the plume transmittance is given by
T
t
=
(1 - L )/(1
sp np s
- ~)
(4. 2)
where I and I are as previously defined, L is the luminance of the hillside
sp s n
viewed adjacent to the plume and ~p is the apparent hillside luminance seen
through the plume.
In actual practice the measurements were made with a portable, battery
operated, narrow field telephotometer using a red filter with a mean response of
6510 A in order to correspond with the 6943 A lidar measurement.
It was not
possible or desirable to locate the telephotometer and the lidar in the same exact
spot since the telephotometer had to have its optical path terminating on a hillside
while this was to be avoided with the lidar from a safety point of view.
The rela-
tive viewing angle between the two instruments never exceeded 10 degrees, however
restricting the variation in plume path length for the two measurements to less
than 1. 5 percent.
4. 2. 1
Tests at Albright, West Virginia
The first site visited was the Albright Station on 9/18/71.
This site
had been selected because of geometry allowing the telephotometer transmittance
measurement to be made in addition to the high velocity of its plumes which yield-
ed a large distance of vertically rising, well defined plume structure.
It wa s
known to be an older plant than some and to have fairly low transmittance plumes.
48
-------
On arriving at the plant, a discussion was held with the local management at
which point they agreed to reduce the load on the plant in an attempt to increase the
transmittance of the plume.
The test site is shown in Figure 4. 2.
These photographs were
taken at different times during the day under quite different lighting conditions
although the telephotometer measurements indicated a relatively constant trans-
mittance of about 10 percent.
This illustrates the lack of correlation between
visual plume appearance (plume-to- sky contrast) and plume transmittance or
opacity.
It is interesting to note that although the plumes are not dark in Figure
4.2a, the left hand plume totally obscures the far hillside.
A series of lidar and telephotometer measurements were made on
both the number 2 (center) and the number 3 (right hand) plumes throughout the
afternoon.
The lidar measurements were made at a range of 300 meters.
The
results of these measurements are shown in Figure 4.3 as plume transmittance,
both lidar-determined (T L) and telephotometer-determined (Tt)' as well as tele-
photometer-determined plume-to- sky contrast (C) vs. observation time and power
plant stack ope rating load.
The amount of useful lidar data was limited by occa-
sional rain and some experimenting with operating without gating in addition to
system alignment checks.
The telephotometer transmittance data shows an aver-
age transmittance of about 8 percent on both the number 2 and 3 stacks prior to
the load reduction on the number 3 stack. The oscillations shown in this data are
real and usually represent the effect of rapping the precipitators.
The time required
for the series of four measurements needed to determine transmittance by telepho-
tometry was approximately 5 seconds, and short enough to measure the variations.
49
-------
\J1
a
\
(a)
(b)
Figure 4. 2
Two photographs (a and b) of the Albright Station of the Allegheny Power and Light Co. .
Albright. West Virginia. The right hand photograph (b) was taken with much more overcast
lighting conditions than (a) although telephotometer measurements indicated a relatively con-
stant transmittance of about 10 percent. The telephotometer and lidar truck are shown in the
foreground in (a). This illustrates the lack of correlation between visual plume appearance
(plume-to-sky contrast) and transmittance or opacity. Note that the hillside is totally obscured
by the left hand plume.
-------
1.00 0.50
I- x 0 LIDAR - TL
I- A TELEPHOTOMETER- TT u
.. I
.J 0.80 xTEL£PHOTOMETER-C 0.40 l-
F C/)
, «
~ a:
o ./\ x I-
z 0.60 z
~ \I 0.30 0
u
t: >-
2 \."k~ ~
C/) l~ 0.20 ~
z 0.40
«
a: l-
I- f LLI
\.11 L1J 2
...... 2
0.20 0.10 3
:3 ~ 00 00
o Cl.
a.. ;1
~
0 0
#=2 ~ #3 #3
140 MW 100 MW
- 0.10
1300 1400 1500
TIME - HRS.
Figure 4. 3
Lidar and telephotometer determined plume transmittance plus plume-to- sky contrast vs.
observation time and power plant stack operating load. Data taken at the Albright Station,
Allegheny Power and Light Co., Albright, West Virginia, 9/18/71.
-------
Lidar data during this period shows a transmittance of 15 to 17 percent and must
be considered an upper bound measurement since severe afterpulsing was present
and the laser pulse could be seen visually on the plume through the 10 power rifle
scope.
This is not surprising when the synthetic target re suIts shown in Figure
3. 7 are compared with these results.
The target tests would predict a lidar-
determined transmittance of from 16 to 27 percent for an 8 percent transmitting
target depending on its reflectivity.
This implies a low reflectivity for the plume
although it certainly was not as low as the black felt target on which the lidar pulse
was invisible.
At about 1500 hours the load was changed on the number 3 stack from
140 MW to 100 MW.
The telephotometer data show an average increase in trans-
mittance to greater than 20 percent, while the lidar data, taken only at the very
end of the period because of intervening rain, show only a slight increase in trans-
mittance.
There is too little data and temporal correlation to conclude anything
from these results.
Finally, it can be seen that plume-to-sky contrast (C) varies widely
during periods of relatively constant plume transmittance as shown pictorially in
Figure 4. 2.
This wide variation in contrast was due to wide variation in ambi«ent
illumination of the plumes resulting fro'm uneven overcast sky conditions.
4.2.2 Tests at Point Marion, Pennsylvania
The lidar system was driven to the Fort Martin Station, Allegheny
Power and Light Co. near Point Marion, Pennsylvania for the second day of test-
ing on 9/19/71.
Again the weather was overcast and rain threatened.
It was not
52
-------
possible to arrange to make telephotometer transmittance measurements at this
site although plume-to- sky contrast measurements were made with the telephotom.
eter sitting next to the lidar system.
This site was chosen as representing one of
the most modern coal burning power stations available with a relatively clean
plume from each of its two stacks.
Again the local management was consulted
and as had been prearranged, they agreed to vary the load on each of the two
stacks in an attempt to vary the transmittance.
A sample of the lidar data and results is shown in Figure 4.4.
Here
the shot is through the number 2 plume operating at 535 MW and at a range of 487
meters.
The oscillogram and the reduced data plot look similar to those obtained
with the synthetic targets shown in Section 2.
2
Good l/r performance is evident
on both sides of the plume and no afterpulsing is visible.
The computed trans-
mittance is unambiguous and is supported by the near invisibility of the plume.
A majority of the data taken at this site looked this good (Figure 4.4)
2
but, in a limited number of cases, lack of l/r performance was observed on the
far side of the plume.
2
These cases showed steeper than llr dependence and
occurred in the absence of any visible afterpulsing.
It has been concluded that
this must be the result of thermal lensing due to the hot plume gases.
Limited
attempts at a rough analysis of the problem have not yielded a simple optical con-
figuration from which numerical results can be obtained.
This is particularly
true when the plume is turbulent and its optical properties vary widely from place
to place and with time.
Thus, all that can be said at this point is that the problem
occasionally exists, it can be readily recognized and by weighting most heavily
53
-------
U)
~
o
>
I
..J
«
z
C)
-
U)
0::
«
c
-
..J
0.20
0.10
0.07
0.05
MON.
GATE
POSe
TL= IA/B'= 0.78
0.03 0
10
Figure 4. 4
200 400 600
RANGE-METERS
1000
Lidar shot through a power station plume; Fort Martin Station,
Allegheny Power and Light Co., Point Marion, Pennsylvania,
9/19/71. This is a coal fired plume operating at 535 MW with a
lidar range of 487 meters.
54
-------
those points nearest the plume, its effect can be minimized.
This effect had not
been noticed on previous real smoke plume results, but experience with high
transmittance plumes had been limited and early afterpulsing on low transmittance
plumes masked the effect if and when it occurred.
The results of the lidar and telephotometer measurements can be
seen in Figure 4.5 which is similar to Figure 4. 3 in format.
Here there are no
telephotometer transmittance measurements to compare the lidar results with,
but some interesting observations can still be made.
The first set of measure-
ments were made on the number 2 stack plume operating at 535 MW.
This plume
was nearly invisible as indicated by the low plume-to-sky contrast shown.
The
lidar-determined transmittance oscillated around O. 80, a transmittance level
where the measurement error is vanishingly small according to the synthetic
target test results.
The second measurement period was devoted to the number 1
stack plume operating at 480 MW.
This plume was clearly visible, as the higher
values of plume-to-sky contrast indicate.
The lidar-determined transmittance
of about 0.40 is an upper bound since afterpulsing was present on the oscillograms.
The synthetic target test results of Figure 3. 7 indicate the real transmittance
should have been in the range of O. 30 to 0.40.
Following a lunch break the load was alternately changed on both
stacks.
First the load on the number 2 stack was dropped to 380 MW.
Intermittant
rain showers and the existence of uncooperative cooling tower plumes allowed
only two good lidar shots but they indicate an increase in transmittance to about
0.92.
No telephotometer data was taken during this period.
The final data series
was taken using the number 1 plume at the reduced load of 400 MW.
This series
55
-------
1.00
U1
0'
.J
.... 0.80
I
1&.1
U
Z
~ 0.60
....
-
2
U)
~ 0.40
a:
....
~ 0.20
::)
.J
a.
Figure 4. 5
~
)(""x
o
1100
o LIDAR-TL
00 x TELEPHOTOMETER-C
0.40
'\
I
x
x
\
~
#2
380 MW
x
x
J
#1
x 400MW
1500
U
I
....
0.3 0 U)
C
II::
....
z
0.20 8
>-
~
U)
0.1 0 0
...
1&.1
2
::)
.J
0..
Lidar-determined plume transmittance and telephotometer determined plume-to- sky contrast
vs. observation time and power plant stack operating load. Data taken at Fort Martin Station,
Allegheny Power and Light Co., Point Marion, Pennsylvania, 9/19/71.
x
1# 2 ------t!lt
535MW 480MW
1200 1400
TIME" HOURS
o
-0.10
-------
is the best example of the lack of correlation between plume-to-sky contrast and
plume transmittance.
Here the transmittance rose to about 0.60 from 0.40 due
to the load reduction and was relatively constant, while the plume-to- sky contrast
varied from -7 percent to nearly + 30 percent due to wide lighting changes.
4. 2. 3 Tests at Springdale, Pennsylvania
On the final day of the field trip, 9/20/71, the lidar system was taken
to the Western Penn Station of the Allegheny Power and Light Co. at Springdale,
Fa.
This site was chosen because of suitable hill-plume geometry for the tele-
photometer measurement and because at least one of the plumes was relatively
clean (in fact nearly invisible except for refractive effects).
This one plume was
used for all of the data and the load was not varied during the day.
Again the
weather was quite overcast with occasional short rain showers until about 1500
hours when a steady rain halted the operation.
The results of this testing are shown in Figure 4.6, again with lidar
and telephotometer-determined plume transmittance as well as telephotometer-
determined plume-to-sky contrast plotted vs. observation time.
This data was
taken at a range of 390 meters.
One feature, in addition to the weather, limited
the amount of data taking.
A small smelting plant with a broad, diffuse smoke
plume was located directly up wind of the power plant.
This plume occasionally
enveloped the top or near side of the stack being observed, interfering with the
measurement.
This was easiest to spot from the lidar traces which showed large
discontinuous returns at ranges other than the stack range and these traces were
discarded.
57
-------
1.00 0.50
.... 0
.... ~ (,)
.. 0 0 \ 0 I
~ 0.80 ~ 0.40 ....
V U)
I C
I&J a:
(,) ....
z 0.60 0.30 z
~ 0
o LIDAR - TL (,)
t:: 0 TELEPHOTOMETER-TT >-
~ ~
U) U)
z 0.40 x TELEPHOTOMETER - C 0.20 ~
\J1 C
00 a:
.... I&J
I&J r :E
:E 0.20 0.10 :)
..J
:) .t a.
..J
a. J
,
0 0
1300 1330 1400
TIME (HRS)
Figure 4. 6
Lidar and telephotometer-determined plume transmittance and telephotometer-determined
plume-to-sky contrast vs. observation time. Data taken at the Western Penn Station,
Allegheny Power and Light Co., Springdale, Pennsylvania, 9/20/71.
-------
The data show
good general agreement between the two transmit-
tance measuring techniques, especially in the last series of measurements where
there is reasonable temporal overlap.
Again, the plume transmittance was high
enough that the lidar error should have been negligible.
Since the ambient light-
ing conditions did not vary 'much, the day remaining generally overcast, plume-
to- sky contrast did not show much variation and was low verifying the low plume
visibility.
Finally, it should be mentioned that lack of l/r2 performance on the
far side of the plume occurred for the last three of the five lidar data points shown.
As has been discussed in the preceding section, this is believed to be due to ther-
mal lensing by the hot plume gases causing some of the lidar pulse to diverge
beyond the field of view of the receiving system.
(It is implicit in this argument
that the field of view of the receiver is much larger than the laser divergence
such that the two see different mean portions of the plume's thermal lens and thus
differential focussing is possible).
Again, the closest in far side data points were
used for obtaining the transmittance, T , in an attempt to minimize this effect.
L
This generally meant a point 40 meters down range of the plume.
If, on the other
hand, data points 100 meters down range had been weighted most heavily the com-
puted transmittance values would have been lower by less than 10 percent.
It is,
of course, not proven whether the thermal lensing hypothesis is correct or
whether random inhomogenieties in the atmospheric scattering caused this lack
2
of l/r performance.
The evidence is strong, however, for thermal lensing since
in all cases the signal falls off monotonically and faster than l/r2, while with
59
-------
atmospheric variations this should be random.
2
Further, the lack of l/r per-
formance is always observed on the far side of the plume.
Finally, it is clear
that it is not an inherent characteristic of the lidar system since the majority of
such plume shots and all of the synthetic target test shots where afterpulsing did
not occur show good far side l/r2 performance.
60
-------
5. CONCLUSIONS
During the construction and evaluation of the lidar system it was used exten-
sively under field operating conditions.
It was driven over 1000 miles in its
present configuration on a variety of road surfaces and operated in a wide variety
of temperature and relative humidity conditions.
The system performed satis-
factorily through all of this requiring little or no maintenance.
Optical alignment
was stable and electronic s performance was reliable.
The physical arrangement
in the truck proved to be quite convenient and the pointing capability and azimuth
and elevation limits were more than sufficient.
It is estimated that once famil-
iarity has been gained with the equipment, it is possible to begin data taking with-
in 15-20 minutes after having arrived at a test site.
This time could probably be
shortened to 5 minutes or less by employing all solid state electronics (eliminat-
ing warm-up) and by providing motor generators that did not have to be removed
from the vehicle.
Some observations can be made relative to the measurement of real smoke
plumes.
It is generally necessary to use a minimum off-gating pulse width of
200 nanoseconds with 250 and even 300 nanoseconds being commonly used. Plume
ranges varied from just over 200 meters to just under 500 meters and were dic-
tated by accessibility of the site to the vehicle, wind direction and strength,
the existence of cooling tower plumes,
multiple
power plant plumes,
and the desire to reduce the elevation angle as much as possible so that the lidar
optical axis intersected the plume close to perpendicularly.
These ranges men-
tioned do not represent system limits.
The minimum range limit with the lidar
61
-------
axes aligned at infinity is about 150 meters (closer if lidar alignment is changed)
while the maximum range depends on the transmittance of the plume being meas-
ured and the atmosphere backscattering.
However, taking the data shown in
Figure 4.4 as an example, it probably would have been possible to have made
this measurement of a fairly high transmittance plume (T L = 0.78) at more than
twice the range or nearly 1000 meters based on the signal levels and system noise.
The accuracy with which the lidar system can remotely measure transmittance
or opacity is best seen on Figures 3.7 and 3.9, the synthetic target test results.
Although these were thin synthetic targets, their transmittance was well known
and constant which was not the case for the telephotometer measurements on real
plume s.
Since that data shown is averaged over a number of lidar shots for each
point such random error sources as oscillogram reading, atmospheric inhomo-
geniety, and lidar alignment and aiming should have been averaged out.
This
leaves only the systematic error or system inaccuracy caused by afterpulsing
which depends on target reflectance and is shown on Figure 3. 7 by the amount
that the curved line s deviate from the line of perfect agreement.
The following
general statements can be made about the maximum measurement error in lidar
determined transmittance, T L' assuming maximum target reflectance:
A.
The error in T L is always positive, such that T L is always an upper
bound transmittance.
B.
The error in T L is less than + 12 percent for T greater than 0.50.
c.
The error in T L is less than + 2.5 percent for T greater than 0.80.
In the previous section a comparison was made between the lidar and tele-
62
-------
photometer measured transmittance values during the field tests on real smoke
plumes.
In general, the lidar measurements agreed with the telephotometer
measurements.
Due to bad weather conditions, there is not as much data and
data overlap as desired but agreement between the two techniques is still evident,
particularly at the Western Penn Station in Springdale where plume transmittance
was high and the lidar could be expected to yield good results.
Again, the meas-
urements made at Albright Station, showing telephotometer determined transmit-
tances of about 8 percent with lidar determined transmittances of about 15 to 17
percent, agree well with what is expected of the lidar system from the synthetic
target test results at these low transmittance levels.
Finally, it is clear that, in general, plume-to- sky contrast (plume visibility)
is unrelated to plume transmittance.
Plume-to- sky contrast depends not only on
the transmittance of the plume (the amount that the plume attenuates the luminance
of the sky seen through the plume) but also on the direct scattering of sunlight by
the plume.
Thus, plume-to- sky contrast is quite dependent on ambient lighting
conditions while plume transmittance is not.
63
-------
6. RECOMMENDATIONS
After having used the mobile lidar system extensively, some desirable system
modifications have become apparent.
The reasons for these modifications along
with proposed solutions are discussed in the following paragraphs.
Since the lidar signal return, for a homogeneous scattering atmosphere, behaves in-
herently as l/r2, it is necessary to transfer the data from a polaroid oscillogram to a
double logarithmic plot to make the necessary linear extrapolations prior to com-
puting the far side to near side signal ratio, A/B.
This is a time consuming pro-
cess done with dividers and can take 20 to 30 minutes for one oscillogram before
If the signal were not l/r2 in nature but constant with range,
the results are known.
the data reduction could be done directly on the oscillogram taking the ratio of the
average signal level before and after the plume.
It is possible to add an amplifier
to the system, following the photomultiplier tube and in front of the oscilloscope
preamplifier, whose gain is proportional to time squared and thus, range squared.
2
This would have the effect of cancelling the l/r signal dependence yielding an out-
put constant with time (range) for a homogeneous scattering atmosphere.
The
design of such an amplifier has been considered in some detail and appears quite
feasible.
6
It could have a 3. 5 x 10 Hz frequency response with acceptable noise
and would operate over a range of from 100 to 500 meters.
The amplifier would
be physically small, consume negligible power and could be bypassed by flipping
This type of r2 com-
a switch, leaving the system in its original configuration.
pensation would be universally applicable to any photomultiplier detector design
used with the system and would not require re-working the photomultiplier tube
64
-------
and power supply circuitry or involve the neces sary photomultiplier tube circuit
design compromises which are required when considering varying the gain of the
tube to accomplish the r2 .
correctlon while preserving linear tube response over
the wide signal dynamic range encountered.
A second desirable modification would be the elimination of afterpulsing, thus
eliminating the dominant source of error in the lidar measurement.
Much effort
has been expended already, as outlined in Sections 2 and 3 and detailed in Appen-
dix B, on photomultiplier tube gating without success in the elimination of after-
pulsing.
This has led to a clearer understanding of the afterpulsing mechanism
as well as the magnitude of the effect of afterpulsing on transmittance measure -
ment error.
As a result, two separate solutions to the problem have emerged.
Of the two, one would be simplest to implement requiring few basic system changes
but would be less assured of success, while the other would require optical and
mechanical modifications to the receiving system but should reduce afterpulsing
effects to a negligible level.
The first solution mentioned consists of purchasing a recently available photo-
multiplier tube commercially made with an internal screen grid adjacent to the
photocathode.
This gating option has not previously been available and most cer-
tainly will stop afterpulsing.
The question is whether the applied electric field
will be sufficiently uniform that charge redistribution on the photocathode will not
take place avoiding damaging the tube's on-response following realistic off-gating
pulse durations.
If nec e s sary, with thi s new type of tube step s could be taken to
prevent charge redistribution on the photocathode.
This solution should probably
be tried first since it is the easiest to implement and would involve the least
65
-------
system modification.
The second solution involves placing a fast optical shutter in front of the photo-
multiplier tube, disallowing the troublesome plume return pulse from ever getting
to the photocathode.
This can be done using a Pockel cell shutter and would employ
the current photomultiplier tube.
In principle, this technique should nearly elim-
inate afterpulsing but it was not seriously considered prior to this because of cost,
complexity and the difficulty of satisfying all of the optical requirements of the
Pockel cell when used with the lidar receiving system.
Preliminary designs
coupled with some detailed discussions with manufacturers of Pockel cells have
indicated a feasible Pockel cell configuration which should allow attenuation of the
plume return signal by a factor of at least 100.
Such an attenuation should reduce
afterpulsing to the point of its error contribution being negligible.
Thi s technique
can then be considered as a back-up to the previous technique of employing a
gridded photomultiplier tube.
Following the completion of either or both of the above modifications it W) uld
be desirable to evaluate the effect of these changes using the synthetic targets at
approximately 200 to 300 meters range.
This would not be difficult since the
targets exist
and have been calibrated, and if the r2 compensation has been in-
stalled data reduction will be much les s time consuming.
Of course, a limited
number of shots would probably be made using the existing lidar configuration for
side by side comparison and these would be time consuming to process.
Finally, it is recommended that the problem of thermal lensing of smoke plumes
be investigated both analytically and experimentally.
A first order analysis should
66
-------
be made for the case of laminar plumes to indicate the effective focal length of
such lenses and the variation of this focal length across the plume allowing for
temperature gradients.
This would then be inserted into a simple optical model
of the lidar transmitted pulse and receiver field of view allowing first order
numerical evaluation of the lensing effect.
Experimental confirmation of these
results could then be gotten by systematically observing highly transmitting, hot,
laminar plumes and comparing the far side signal return both through and immed.
iately adjacent to the plume.
As has been mentioned earlier it does not look rea-
sonable to analyze turbulent plumes but systematic lidar observations could be
made, perhaps with varying receiver fields of view, both through and adjacent to
the plume yielding a better measure of the frequency of and factors contributing
to lack of far side l/r2 performance.
67
-------
7. ACKNOWLEDGEMENTS
The authors are grateful for the cooperation of the Philadelphia Electric
Company and the Allegheny Power and Light Company for making their power
station facilities available for the field testing done during the evaluation of the
mobile lidar system.
68
-------
8. REFERENCES
1.
Conner, W. D. and Hodkinson, J. R., "Optical Properties and Visual
Effects of Smoke Stack Plumes", Public Health Service Publication
No. 999-AP- 30, Cincinnati,
Ohio, 1967.
2.
Evans, W. E., "Development of Lidar Stack Effluent Opacity Measuring
System", Final Report of Stanford Research Institute Project No. 6529,
Menlo Park, California, July 1967.
3.
Bethke, G. W., Cook, C. S., and Mezger, F. W., "Laser Air Pollution
Probe", Final Report, ARC Contract No. 68-19, January 1970.
69
-------
APPENDIX A
THE LIDAR RANGE EQUATION
It is assumed that the laser light source is pulsed)with light
pulses which are short in length compared to the measurement ranges
of interest.
It is also as sumed tha t the las er beam is collimated to a
divergence which is smaller than the lidar receiving system angle of
acceptance (field of view).
It is further assumed that the transmitted
laser beam is coaxial or nearly coaxial with the receiving system, such
that the propagating laser beam is located entirely within the receiving
system field of view during the period (ranges) of interest.
Because the
laser beam is located entirely within the receiver field of view, only
the total power (PL) and energy (E) of the laser beam need be considered)
and not the beam intensity.
The total power of incident light (Pi) scattered in direction 8 by
illuminated scattering length (.t) of gas 1S
P s(8) = Pi t . ks(8)
( I)
where ks(8) is the directional scattering coefficient.
Now referring to Figure A-I, we see that a pulse of light which
starts from the laser at time t and continues to be emitted until time
o
tL' has a length L, where
A -1
-------
L..L.I
<.=>
z
-------
L = c (tL - to) = c t L
( 2)
As this short light beam of length L propagates away, light is continuously
scattered back to the lidar system.
However, FigureA-l illustrates
that the scattered light observed by the lidar system at anyone instant
in time, tr, is scattered only from incident light located between ranges
r 1 and r2 which have a distance difference of L/ Z.
Thus the illuminated
length of gas that scatters light back towards the lidar system at any
one time, tr, is
i, = rZ - r 1 = L/ Z.
(3 )
The light which left the lidar laser with power PL arrives at the
range of interest (r) with reduced incident power Pi due to atmospheric
extinction.
Thus
Pi = P L exp (-r k e)
( 4)
where ke is the total extinction coefficient of air from all sources
as suming extinction to be constant with range.
Finall y ,
P L = E = E / tL
(5 )
where E is the total laser energy per pulse.
N ow combining equations
1 through 5, we have
P s (e) = (c E . k s (e) / Z) exp (- r k e) .
(6 )
The power of back-scattered light collected (P c) by the lidar
receiving objective is
Pc = Ps(180o) . r exp (-rke)
A-3
( 7)
-------
where r is the solid angle subtended by the lidar receiving objective from
the scattering range r.
But
r = 27T(l - cosy} ~ TTy2 ~ A/r2
(8)
where Y is the angle subtended by the radius of the lidar receiving objective
from scattering distance r, and A is the area of the receiving objective.
Since it has been assumed that the light pulse is short in length compared
to the measurement ranges of interest, we can assume with negligible
error that r is the same at anyone instant (tr) from all portions of the
illuminated scattering volume.
Finally, the power of scattered light
which reaches the detector (P d.> is
P d = Eo Pc
(9 )
where Eo is the total efficiency of all optics in both the transmitting and
receiving sections of the lidar system.
Thus combining equations 6 through
9 we have
Pd =
~ 7z Eo E'ks (1800). exp (-2r ke) .
( 10)
If Pd is in watts and the detector has a sensitivity S in amps/watt,
then the voltage drop, V, produced by detector output current I across
detector load resistor R is
V = IR = P d S R
(11 )
The directional scattering coefficient consists of both molecular
(Rayleigh) scattering, and particulate or aerosol (Mie) scattering.
ks (8) = n( r} . O"R (8) +
A-4
m(r) . O"M (8)
(l2)
-------
Here, n is the molecular concentration (density) in air, 0: (9) is the directional
R
Rayleigh cros s section, m is the particulate concentration in air ,and 0" (8) is
M
the suitably size-integrated directional Mie theory-calculated cross section for
particulates in the air.
Similarly, the total extinction coefficient is defined as molecular (Rayleigh)
and particulate or aero sol (Mie) total scattering plus any molecular absorption
at the specific wavelength of interest.
ke = n( r) , aR +
nI1( r) . aM + ka
(13)
Here a is the total Rayleigh cross section, k is the absorption coefficient of
R a
air at the laser wavelength and O"M is the total particulate (Mie) cross section.
As is clear from equation 13, k is not constant with range.
e
Consequently, we
must replace the expres sion (r k ) in equations 4, 6, 7, and 10 with
e
r
(>
r ke->aR J n(r)d r +
o
r
aM J m(r}dr + rka.
o
( 14)
Finally, combining equations 10, 11, 12, and 14, one obtains the range equation
v = cRSE EA [n(r} , J"R (1800) + m(r} 'aM (1800)1
o 2r2 'J
r r]
. exp-2 fa rn(r}dr+ J"MJm(r}dr.
L R Jo 0
(15)
Here we have dropped the absorption coefficient of equation 14 because
A-5
-------
we assume that the laser wavelength does not coincide with a significant
atmospheric line or continuum.
The atmospheric density. n can be expressed analytically if one
curve-fits to a suitable standard atmosphere.
For lidar system elevations
that do not cover a range of more than a few thousand feet above sea level,
the following expres sion is useful to about 50,000 feet scattering altitude:
n = n ~ . [exp (-2.3026 rZi sin cp) - Z'I r2sin2 cpl
o Tpo J
( 16)
Here, no is the value of n at temperature To and pressure Po' while T
is air temperature and p is air pressure near the lidar system, cp is the
lidar system angle of elevation above horizontal, and Z' and Z" are
suitable constants.
If the lidar system is at sea level and r is in cm, then
6 -14 -2
setting Z' = 2.31 x 10 cm and Z" = 3.0 x 10 cm gives n values
which agree with the AFCRL 1959 standard atmosphere to within ~o. 6%
and < 3% at ~ 30,000 feet and ~ 50,000 feet altitude s above the lidar;
respectivel y.
If the lidar system is at 2000 feet altitude, then the same
asswnptions give agreement to within s;; 1 % and < 6% at s;; 20, 000 feet and
s;; 50,000 feet altitudes above the lidar, respectively.
Of course for q;«
o
90 , one is not concerned with such high altitudes even when r is large.
A-6
-------
APPENDIX B
- PHOTOMULTIPLIER TUBE GATING INVESTIGATIONS
In order to make lidar-determined smoke plume t ransmi t t ance measurements,
it is necessary to faithfully record the atmospheric backscattering both just in
front of and just behind the plume.
There is, of course, no trouble in obtaining
the near side signal right up to the point in time where the laser pulse intersects
the plume.
However, the intense light scattering from the plume itself (during
the time that the laser pulse is in the plume) complicates the problem of faith-
fully recording the clear air light scattering immediately on the far side of the
plume.
This backscattering from the plume can be as high as 40 db above the
ambient clear air scattering for the case of a dense white plume (or for test pur-
poses, a white wall).
Two separate after effects from this intense plume signal can occur during the
time period just following the emergence of the laser pulse from the plume. Both
of these effects tend to increase the apparent amplitude of the far side clear air
return, yielding transmittance values for the plume that are too high.
The first effect is simply caused by the frequency response of the detector and
amplifying electronic s in going from the high level plume signal to the low level
far side clear air signal.
Since the slowest frequency response occurs
in the
amplifying electronic s, turning off the detector during the intense plume return
will avoid overdriving the electronics and will allow a proper far side measure-
ment.
The second effect is called after-pulsing and is less certain in origin although
its effects are obvious.
Each strong discrete light pulse (plume signal) produces
B-1
-------
not only a corresponding discrete output from the photomultiplier tube (PMT)
anode, but also one or more secondary output pulses which usually occur 0.2 to 1
microsecond after the primary pulse.
These secondary pulses are somewhat
broader and decay much more slowly than the main pulse.
The explanation for
this effect is as follows:
The PMT cathode photoelectrons (created by backscat-
tered light from the intense plume return) leave the cathode and are accelerated
towards the first dynode.
At some point along their path these electrons acquire
sufficient energy to ionize the residual gas in the tube or to eject ions from the
tube structure upon impact.
These relatively slow ions then are accelerated back to
the
photo-cathode where they cause secondary electron emission.
The second-
ary electrons then are amplified down the dynode chain and wind-up as an after-
pulse signal at the anode. Those electrons resulting from light ion impact at the
photocathode cause the earliest after-pulsing while heavier and slower ions cause
the later after-pulsing.
The amount of after-pulsing then depends on the intensity
of the reflected signal from the plume or target.
To eliminate at least the first (slow electronics recovery) effect and if possi-
ble also the afterpulsing effect, investigations were made to determine the best
method for gating off the PMT during receipt of the strong smoke plume return.
The various requirements imposed on the off-gate method to be used included the
need for a large on/off ratio (preferably> 30 db), a rapid off-gate response and
rapid on-recovery response « 100 nanoseconds), and the ability to off-gate for at
least 200 nanoseconds.
The real thickness of smoke plumes combined with the
total titne of the complete laser pulse, all under field operating conditions, imposes
the last requirement.
B-2
-------
EXPERIMENT AL T EST METHODS
Several experimental methods were used to measure the above-mentioned
gating characteristics as a function of the gating method being tried.
In all case s,
the PMT operating voltage was set to the value anticipated for lidar operation.
Also, a Tektronix #551 oscilloscope with H preamplifer was used for all the re-
sponse time measurements, this combination having a manufacturer - stated
risetime of 25 nanoseconds.
The pulse generator provided pulses of ~ + 100
volts, with 15 nanosecond rise and fall times.
PMT gating response time and the smaller on/off ratios ( ~ 25 db) were meas-
ured by off-gating the PMT while it was exposed to a cw light source.
In these
cases, a Corning #2412 red filter was used and the light level was adjusted to
yield PMT anode outputs less than 0.1 of the divider chain current.
The latter
precaution insured linear PMT response under cw operating conditions.
The larger on/off ratios ( > 25 db) which resulted from off-gating the PMT
were measured with a pulsed red light source as follows: A He - Ne laser (6328 A)
was beamed at a rotating prism (500 rps) which swept the reflected laser beam
across an 18 foot distant aperture placed in front of the PMT.
An opal glas s
diffuser between the aperture and the PMT produced nearly uniform illumination
of the PMT photocathode as the laser beam moved across the aperture, thus min-
imizing the effect of any spatial nonuniformities in the photocathode sensitivity.
Whenever it was necessary to reduce the light intensity on the PMT by a known
amount, calibrated neutral density filters were placed in front of the aperture.
For most of these measurements, the aperture was set so the light pulses were
about O. 3 to 0.4 microseconds long (FWHH).
For these measurements, the PMT
B-3
-------
anode maximum current, maximum current times pulse width (charge/pulse),
and average current were all kept within linear response ranges for the PMT.
The linearity between PMT anode pulse amplitude and input light pulse intensity
was determined using this same experim.ental set-up at (a) constant light level
while varying PMT voltage and (b) constant PMT voltage while varying light level
via calibrated neutral density filters.
Finally. afterpulsing plus any electronics effects were measure d with the PMT
in the lidar system.
The lidar return from a white wall plus the laser output
energy monitor signal were displayed on the lidar oscilloscope, both with and
without off-gating the PMT during the white wall return.
This very strong and
brief light return approximated that which would be returned from a dense white
(high reflectance) smoke plume, a very severe test.
Since the wall was opaque,
the signal level ideally should have gone to zero immediately after the wall return
signal.
Any remaining non-cw PMT signal was then due to a combination of PMT
afterpulsing plus any effects originating in the receiver electronics.
OFF-GATING METHODS AND RESULTS
Our investigations have included off-gating the Amperex 56 TVP PMT via (a)
the focus electrode, (b) the photocathode (by pulsing an external wire mesh), (c)
dynode 2, (d) dynode 6, and (e) dynodes 2 and 6 simultaneously.
The RCA 7265
PMT fits the same socket as the Amperex tube, with most of the pin connections
being the same as well as having a very similar voltage distribution requirement.
Thus
the divider circuit was only slightly modified for the RCA 7265 PMT which
was also investigated while off-gating via (a) the focus electrode, (b) the photo-
B-4
-------
cathode (using an external mesh), and (c) dynodes 2 and 6 simultaneously.
The principal results of all these investigations are summarized in Table B-1
which lists the PMT used, the pulsed element, the pulse voltages for best gating
characteristics (within + 100 V to -100 V limits), the 0% to 90% turn-on response
time at the end of a 200 nanosecond(off-gating) pulse period, and the on/off gat-
ing ratio for both the primary (plume) signal and for the resulting afterpulse
signal.
Since the on-response time increases with pulse width for some of the
gating methods, a 200 nanosecond pulse period is used for comparison in Table
B-1, that being the shortest off-gating period that is usually practical under field
conditions while looking at real smoke plumes.
During these investigations, for each PMT driving voltages were used which
yielded PMT gains approximately those which were later found to be needed for
lidar use.
(Lidar use required 1500 V for the Amperex 56 TVP tube and 1900 -
2000 V for the RCA 7265 tube).
The PMT voltage divider used for all these in-
vestigations had resistor (R) ratios as follows: focus electrode adjustable,
cathode -Dl = 3R, Dl - D2 through DlO - Dll = R, Dll - Dl2 = 1. 36R, D12-D13 =
2R, D-13 - D14 = 3R, and D14 - ground = 4R.
Based on the afterpulse mechanism suggested earlier, it would appear that
off-gating a PMT by reversing the electric field at or near the photocathode sur-
face would eliminate any afterpulsing (to the extent of the on/off ratio) since
ionization then could never occur in the region beyond the photocathode. For this
reason and also because of earlier work and suggestions by Evansl, the initial
attempts were to off-gate the Amperex 56 TVP PMT first via the focus electrode
B-5
-------
and then via an external metal mesh over the photocathode.
The results of these
plus subsequent investigations are described in the following subsections.
Afterpulsing and Focus Electrode Gating of the Amperex 56 TVP
As shown in Table B-1, when off-gating was accomplished using the focus elec-
trode of the Amperex 56 TVP PMT, the best results were obtained by employing
a -100 V pulse.
Under these conditions, the afterpulsing was reduced by about
8 db.
However, the PMT recovery time at the end of the off-gated period was
too slow to be useful for the plume opacity measurement application.
Figure Bla shows the non-gated Amperex 56 TVP afterpulsing resulting from
the lidar being shot at a white wall, while Figures BIb and BId show the reduced
afterpulsing obtained when the source (wall backscatter) signal was gated off with
a -100 V pulse to the PMT focus electrode.
As seen from Figure Bla, the prin-
cipal afterpulse maximum (A) is 0.67 JJ. sec
after the source pulse (W), while
a secondary afterpulse (A ') is 1. 3 J,J. sec
after W.
The principal afterpulse is
about 40 db less than the source pulse wouki be if the PMT did not saturate.
Between A and W (Figure BI) there is a reduced level of signal which is still non-
zero.
Since an opaque wall permits no lidar return from greater ranges, a per-
fect detector would show zero return after W.
Table B-1 also shows the on-response time to be an undesirably slow 630 n sec
when a -100 V 200 nsec wide gating pulse was applied to the PMT focus electrode.
This is the narrowest gate width practical for field use on real plumes.
These
on-response times were found to increase considerably with increasing pulse
width and with increasing absolute value of pulse voltage on the focus electrode,
B-6
-------
G
W A AI W A
(c) (d)
M
b:I M
I
-..J
PMT PMT
(a)
M
PMT
G
(b)
.<. --T
"
'''''->(
rr.:.
.--r~ ~""_._---+
I ! '
r~ !
~ ~+...++- +t... .....+
.Hi \.... I j
. ...
M
PMT
...l-
~'+- '"
G
w
A
AI
Figure Bl. Lidar oscillograms showing the results when the PMT detector is an Amperex 56 TVP with focus
electrode gating and with the following conditions: Bla is white wall target and no gate, BIb and BId are white
wall target with gate, and B1c is no target but with gate. In each oscillogram, W is position of wall return,
A and A' are afterpulse peaks, M is the laser energy monitor trace, PMT is the PMT lidar return (negative
going signal), and G is the differentiated off-gating pulse showing gate position. In all cases, the gate pulses
are -100 V amplitude and 200 nsec wide, oscilloscope sweep speeds are 0.5 JJ. sec/cm, and the PMT traces
are 0.5 V /cm except for 0.05 V /cm for PMT of Figure BId.
-------
Table B - 1.
Photomultiplier Tube Off-Gating Characteristics
On-response On/off ratio (db)
Pulse time (n sec.) Source Afterpu1se
Tube Used Element Pulsed voltage (a) signal maximum
Focus electrode + 100 V 610 n s 11db 0 db
Focus electrode - 100 630 '" 23 "'8
b 60 930 18
Photocathode +
Photocathode - 40 2000 8.2
Amperex Dynode 2 + 100 '" 40 20
56 TVP (c) Dynode 2 - 100 15
Dynode 6 + 100 '" 30 23
Dynode 6 - 100 '" 30 25
Dynode s 2 & 6 + 100 ,..- 3 5 45 0
Dynodes 2 & 6 - 100 '" 40 41 ""'0
---- ---.--. --- -----------.- ..--.- --.--.---.
Focus electrode + 100 ,..- 130 14 0
Focus electrode - 100 '" 130 11 0
RCA Photocathode b + 100 1800 > 13
7265 (d) Photocathode - 70 1000 7. 0
Dynodes 2 & 6 + 100 ,..- 5 0 45 0
Dynodes 2 & 6 - 100 '" 50 36
-_.~--~._-_. ._h. - -. h
(a) Gating pulses were 200 nanoseconds wide.
(b) The photocathode was pulsed capacitively by pulsing an external wire mesh
which was held flat against the face of the tube.
(c) Amperex tube was operated at 1300 and 1500 volts.
(d) RCA tube was mostly operated at 1900 and 2000 volts.
B-8
-------
and to decrease somewhat with increasing total operating voltage on the PMT.
For example, for 1500 V on the PMT and -100 V pulses, the PMT on-response
time at the end of the off-gate period increased from 350 nsec for 100
nsec
pulse widths, to 630 nsec for 200 nsec pulse widths, and to 1100 nsec for 500 n
sec pulse widths.
Figure B1c illustrates the effect of the slow on-response re-
covery of the Amperex PMT when focus electrode gating was employed.
Since
here the lidar shot was into clear air, the PMT output should have resumed its
1/r2 fall-off immediately at the end of the 200 nsec gate.
Afterpulsing and Focus Electrode Gating of the RCA 7265
When off-gating at the focus electrode of the RCA 7265 PMT, the best results
were obtained with + 100 volt pulses.
Unlike the Amperex tube, this RCA PMT
showed a usefully rapid 0% - 90% on-response (recovery) time of
"" 130 nsec
after any gating pulse ~ 1000 nanoseconds wide.
Unfortunately, gating at the
RCA PMT focus electrode had no attenuating effect on its afterpulsing.
These re-
suIts are summarized in Table B-1.
Figure B2 is a lidar oscillogram showing afterpulsing by an ungated RCA 7265
PMT.
This oscillogram resulted from a lidar shot at a white wall, with the wall
return being located at W on the oscillogram.
The principal afterpulse maximum
(A) is about 0.3 #J. sec after W, followed by a strong secondary afterpulse (A ')
about 1 #J. sec after W.
Here too, the strongest afterpulse maximum was about
40 db less than the source pulse would have been if the PMT did not saturate.
Between Wand A, and betwen A and A' there are reduced signal levels which
are still non- zero.
Although the timing and detailed shape of afterpulsing from
B-9
-------
Figure B2.
M
PMT
Lidar oscillogram showing RCA 7265 photomultiplier tube (PMT)
afterpulsing. Trace M is the positive-going laser energy monitor
output and trace PMT is the negative-going PMT output. The lidar
shot is at a white wall target with its off- scale return at W. and
the resulting afterpulse peaks at A and A '. Oscilloscope sweep
speed is 0.5 J.L sec / cm.
B-lO
-------
the RCA PMT are different than that from the Amperex PMT, the magnitude of
afterpulsing from both PMT' s is comparable.
External Grid Gating of the Amperex 56 TVP
When pulsing the Amperex PMT photocathode by means of an external mesh
grid held flat against the face of the tube, as shown in Table B -I the best off-
gating was obtained with a + 60 volt pulse.
However, the PMT recovery time at
the end of the off-gated period was far too slow to be useful.
Consequently, the
afterpulse on/off ratio was not measured.
As with the Amperex tube focus electrode gating already discussed, when the
Amperex PMT was capacitively gated at its photocathode via an external mesh
the on-response (recovery) times were found to increase considerably with in-
creasing pulse width and with increasing absolute values of pulse voltage.
The
effect of pulse width on PMT recovery time is shown in Figure B3 which shows
PMT output while exposed to cw light and being gated off with +60 volt pulses
applied to the external mesh.
For this type of oscillogram, the PMT output
(upper trace) is a negative-going signal which shows some ringing as each pulse
(lower trace) starts and ends.
The effects of three pulse widths (200, 500, and
1000 nsec
are superposed, with the slow on-recovery of PMT sensitivity being
apparent at the end of each pulse.
In this way, 0% - 90% recovery times from
+ 60 V pulses were found to be ,..., 0.23, 0.9, 2.5, and> 3.5 IJ sec for 0.1, 0.2,
0.5, and 1. 0 IJ sec wide pulses, respectively.
External Grid Gating of the RCA 7265
As listed in Table B-1, the results of off-gating at the photocathode of the RCA
B-ll
-------
Figure B3.
-0
Oscillogram showing Amperex 56 TVP photomultiplier tube (PMT)
output while exposed to cw light and being gated off with pulses
capacitively applied to its photocathode via an external mesh grid.
The three overlapping lower traces (P) show the +60 V gating pulses
of 200, 500, and 1000 nsec width which are applied to the PMT.
The three partly overlapping upper trace s (PMT) show the PMT
output and its recovery to full sensitivity after the off-gating pulses
end. The PMT traces are negative-going, with their zero at O.
Sweep speed is 0.5 JJ. sec / cm. and vertical sensitivities are 5 mV /
cm for PMT and 50 V /cm for P.
B-l2
-------
7265 PMT by pulsing an external mesh grid held flat against the face of the tube,
show the PMT recovery at the end of the off-gated period to be too slow to be
useful.
Thus, the afterpulse on/off ratio was not measured.
In this case as with the Amperex tube, the recovery times were found to in-
crease considerably with increasing pulse width and with increasing absolute
values of pulse voltage.
Even for -70 V gating pulses (PMT at 2000 V), the 0% -
90% recovery times were found be"", 0.53, 1. 0, 1. 4, and 1. 6 J.I, sec for 0.1, 0.2,
0.5, and 1. 0,.,. sec wide pulses, respectively.
Dynode Gating of the Amperex 56 TVP
Primarily because of the excessively slow recovery of PMT sensitivity after
off-gating via both focus electrode and external mesh over the photocathode, other
means were sought for gating the PMT.
Consequently, off-gating pulses were
applied to dynodes 2, 6, and 2 + 6 with the results summarized in Table B-1 and
detailed below.
When dynodes 2 and 6 were separately pulsed with.:!:. 100 volts, the Amperex
tube was found to be gated off with on/off ratios of 15 - 25 db (see Table B-1).
In all of these cases, the on- (and off-) response times were measured as ~ 40
nanoseconds for off-gate periods of 200 nanoseconds.
Subsequently, .:!:. 100 volt off-gate pulses were simultaneously applied to both
dynodes 2 and 6 of the Amperex 56 TVP tube, resulting in the following on/off
ratios for various operating conditions:
PMT volt s Pulse volts On/off ratio
1300 + 100 V 46. 2 db
1300 - 100 39. 1 db
1500 + 100 45.4 db
1500 - 100 40. 7 db
1900 + 100 43.5 db
1900 - 100 40. 7 db
B-13
-------
The PMT on- (and off-) response times were found to be ~ 40 nsec for off-gate
periods ~ 500 nsec.
This response can be seen in Figure B4 which shows the
response by the PMT to off-gating in the presence of the cw light source.
At the
high oscilloscope sensitivity needed to show PMT cw output with fast response
(93 ohm cable and load), PMT anode stray pickup from the gating pulse produced
10 mV spikes in the PMT anode signal corresponding to the beginning and end of
each gate pulse.
These pickup spikes do not interfere with lidar useage of the
detector due to their very transient nature and due to the fact that lidar shots use
oscilloscope sensitivities 10 to 200 times lower than used in Figure B4.
Figure
B5 also shows the fast response obtained by off-gating the Amperex 56 TVP via
dynodes 2 plus 6, this figure being an oscillogram of a no target lidar shot taken
through the clear aperture of the synthetic target holders (see Section 3).
A s indicated in Table B -I, the PMT afterpulse maxima were not influenced by
gating off the source signal via dynodes 2 plus 6.
However, lidar system evalu-
ation tests using known synthetic targets (see Section 3 of the report) show that
target transmittance measurements made during the afterpulse minimum between
the source (target) signal and the first afterpulse maximum (Wand A of Figure
Bla), are much more accurate with dynode 2 plus 6 off-gating than without such
off-gating.
As discussed in the report, this difference may be due to the ungated
intense target return overdriving the detection system electronics.
Finally, the question arose as to whether the afterpulsing was superlinear,
such that its amplitude would fall off faster than that of the source intensity if the
source intensity were reduced.
To check this possibility, additionallidar meas-
urements were made of afterpulsing while varying the source (wall return)
B-14
-------
ANODE SOURCE OFF
SIGNAL SOURCE ON
to
I
......
U1
GATING
PULSE
Figure B4.
0.005 v/em
100 v/em
0.1 JL see/em
Oscillogram showing response by the Amperex 56 TVP photomultiplier tube to simultaneous
off-gating at dynodes 2 and 6 in the presence of a cw light source. The anode trace has an
oscilloscope risetime of 25 nsec, while the gating pulse trace has an oscilloscope risetime
of 13 nsec.
-------
Figure B5.
l.1li
n
..... .""".:i''''''-''.
~
. r:.iII
==
;u;
,~i II
II II II
Ii
I
PMT
G
Lidar oscillogram showing response by the Amperex 56 TVP photo-
multiplier tube (PMT) to simultaneous off-gating at dynodes 2 and
6. This lidar shot was made through the clear aperture of the
synthetic target holders into clear air while being gated off with
a 300 nsec wide + 100 V pulse. Here, M is the positive-going laser
energy monitor trace, PMT is the negative-going PMT lidar return,
and G is the differential off-gating pulse showing gate position.
Oscilloscope sweep speed is 0.5 JJ sec / cm.
B-16
-------
intensity and off-gating the Amperex tube at dynodes 2 plus 6.
While operating
the lidar laser at a constant output, the afterpulsing reeJUlting from the off-gated
white wall return was investigated employing various masks (1. 0, 0.1 and
0.01 area) over the lidar receiving system objective.
Varying the receiving aper-
ture area over a factor of 100 resulted in no significant change in the ratio of
source intensity (on PMT) to afterpulse peak signal or the ratio of source inten-
sity (on PMT) to afterpulse minimum located before the afterpulse peak (see Fig.
Bla).
Thus, afterpulsing appears to be a linear function of the source pulse, as
the proposed afterpulse mechanism would suggest.
Dynode Gating of the RCA 7265
As shown in Table B -I, when off-gating the RCA 7265 PMT by simultaneously
pulsing both dynodes 2 and 6, the PMT on/off ratio was large enough (45 db at 1900 V
and + 100 V pulses) and the PMT speed of recovery from the off-gate condition
was fast enough (about 50 nsec) to be useful.
As with similar gating of the Am-
perex tube, the afterpulse maximum was not reduced by dynode gating.
DISCUSSION AND CONCLUSIONS
It has been determined that off-gating the Amperex 56 TVP PMT at either the
focus electrode or at the photocathode (via an external mesh grid), and also the
RCA 7265 PMT at the photocathode (via an external mesh) all result in a fast off-
gate response, but much too slow a subsequent on-recovery to be useful for lidar
transmittance measurements of real plumes under field conditions.
This slow
on-recovery of the PMT's after off-gating at or near the photocathode is probably
due to the slow redistribution of charge on the semiconducting trialkali photo-
cathode after the off-gate ends.
The longer the off-gate pulse lasts, the longer
B-17
-------
the photocathode is exposed to altered electric fields, resulting in the accumu1a-
tion of greater deviations from the normal charge distribution.
The difference
in this recovery time characteristic between the Amperex and the RCA tubes
when using focus electrode gating, is probably due to their differences in focus
electrode design and its closeness to the photocathode.
Although the RCA 7265 tube showed fairly fast off- and on- response when off-
gating at the focus electrode, its on/off ratio was not very good and the off-gate
did not reduce the afterpulsing.
The dynode methods of off-gating both the Amperex and RCA tubes all show
both fast off-gate response and subsequent fast on-recovery because the dynodes
have highly conducting metal substrates and are well removed from influencing
the electric field around the semiconducting photocathode.
It was decided to use simultaneous + 100 volt pulsing of both dynodes 2 and 6
to off-gate the PMT because this method produced the fastest off- and on-
response and the largest on/off ratio when off-gating.
Given this dynode 2 plus
6 method of off-gating, the Amperex 56 TVP tube was selected because its after-
pulsing became increasingly strong only at > 550 nsec after the target (plume)
return, while the RCA 7265 tube showed strong afterpu1sing starting much closer
(about 200 nsec) to the target return signal.
Thus, the Amperex PMT permits
the far- side-of-p1ume measurement to be made imme diately ( < 550 nsec) after
the plume signal where afterpu1sing effects are relatively weak.
REFERENCES
1. W. E. Evans, IIDeve10pment of Lidar Stack Effluent Opacity Measuring System",
Stanford Research Institute report, SRI Project 6529, July 1967.
B-18
------- |