PB-205 256
LONG-PATH SPECTROPHOTOMETRIC INSTRUMENTATION
FOR IN-SITU MONITORING OF GASEOUS POLLUTANTS IN
THE URBAN ATMOSPHERE
Arnold Prostak, et al
Bendix Aerospace Systems Division
Ann Arbor, Michigan
October 1970
NATIONAL TECHNICAL INFORMATION SERVICE
Distributed ... 'to foster, serve
and promote the nation's
economic development
and technological
advancement.'
U.S. DEPARTMENT OF COMMERCE
This document has been approved for public release and sale.
-------
Long-Path
Spectrophotometric
Instrumentation for
In-Situ Monitoring of
Gaseous Pollutants
in the Urban
Atmosphere
BSR3O27
Final Report By.
October 1970 Arnold Prostak
Ann Arbor. Michigan Robert H. Dye
Systems Division
-------
BIBLIOGRAPHIC DATA
SHEET
1.; Report No.
3. Recipient's Accession No.
4. Title and Subtitle
Long-Path Spectrophotometric Instrumentation for In-Situ
Monitoring of Gaseous Pollutants in' the Urban Atmosphere
5- Report Date
October 1970
6.
7. Author(s)
Arnold Prostak. Robert H.
Dve
8. Performing Organization Rept.
No> BSR 3027
9. Performing Organization Name and Address
Bendix Aerospace Systems Division
Ann Arbor, Michigan
10. Project/Task/Work Unit No.
11. Contract/gMKOCNo.
CPA 22-69-55
12. Sponsoring Organization Name 'tad Address
Department of Health, Education, and Welfare
Public Health Service
National Air Pollution Control Administration
13. Type of Report & Period
Covered
Final
14.
15. Supplementary Notes
DISCLAIMER: This report was furnished' to the National Air Pollution
Control Administration, in fulfillment of Contract No. CPA 22-69-55.
16. Abstracts
\.
A long-path Spectrophotometric instrument is described for the quantitative measure-
ment of ozone and other infrared absorbing gases (such as sulfur dioxide) in uncon-
fined ambient air. A continuous variable filter wheel is used for wavelength selection
from 7 to 14'u m. The instrument can be used with an active source at a range from 0.4
to over 1.6 km or it can be used to examine passively the radiant emittance from
natural sources. A minicomputer is programmed to control the instrument and analyze
the data in real time. The output of spectra or pollutant concentration is on a
cathode ray tube, a teletypewriter, and/or punched paper tape. The instrument is sen-
sitive to ozone concentrations ranging from 20 microgram/cu. m. .(0..01 ppm) to 2000
microgram/cu.m. (1 ppm), with a measured noise of less than 4 microgram/cu.m. and a
time constant of about two minutes. .
17. Key Words and Document Analysis. 17o. Descriptors
Air pollution
Spe ctrophot'ometry
Monitors
Urban areas
Ozone
Sulfur dioxide
Infrared spectroscopy
Data processing
17b. Identifiers/Open-Ended Terms
17c. COSATI Field/Group
14/02
18. Availability.Statement
Unlimited
19.. Security Class (This
Report)
.ASSIFIET
UN.C1
.ASSIFIET)
Class (Thii
20. Security Class (This
*1jNCLASSIFIED
21. No. of Pages
118
22. Price
FORM NTIS-88 t'lO-70)
USCOMM-DC 40S2B-P71
-------
Final Report
October 197O
Ann Arbor, Michigan
By:
Arnold Prostak
Robert H. Dye
Prepared for:
National Air Pollution Control Administration
Environmental Health Service,
Public Health Service,
Department of Health, Education, and Welfare
Pursuant to: Contract No. CPA 22-69-55
-------
BSR 3027
FOREWORD
The work described in this report was performed between April 1969 and
September 1970. The Government Project Officer was Mr, John S. Nader and the
Assistant Project Officer was Dr. Harold M. Barnes, Jr. , both of the National Air
Pollution Control Administration.
-------
BSR 3027
ABSTRACT
y,
A description is given oiTj| long-path spectrophotoinetric instrument for the
quantitative measurement of ozone and other infrared absorbing gases (such as
sulfur dioxide) in unconfined ambient air. A continuous variable filter wheel is
used for wavelength selection from 7 to 14 \i m. The instrument can be used with
an active source at a range from 0. 4 to over 1. 6 km or it can be used to examine
passively the radiant emittance from natural sources. A minicomputer is pro-
grammed to control the instrument and analyze the data in real time. The output
of spectra or pollutant concentration is on a cathode ray tube, a teletypewriter,
and/or punched papjer tape. The instrument is_sensitive to ozone concentrations
ranging from ZOLj^g/m^ (0. 01 ppm) to 2000 (jijy/rn?/ (1 ppm), with a measured noise
of less than 4; _jig/m* and a time constant of at out two minutes. ., ^ •
-------
BSR 3027
CONTENTS
1. INTRODUCTION 1-1
2. SYSTEM DESCRIPTION 2-1
2. 1 TRANSMITTER 2-6
2.2 RECEIVER 2-7
2.3 ELECTRONICS 2-14
2.3.1 Source 2-14
2.3.2 Receiver 2-14
2.3.3 Central Control Electronics 2-15
2.3.4 Signal Processing 2-21
3. DISCUSSION OF COMPUTER PROGRAM 3-1
3.1 PROGRAM FLOW 3-3
3. 1. 1 INIT Routine 3-4
3.1.2 Auxiliary Routine (AUX) 3-4
3. 1. 3 Interrupt Service Routine (ISR) 3-21
3.1.4 Subroutines 3-22
3.2 DESCRIPTION OF OUTPUT 3-22
4. OFF-LINE DATA PROCESSING 4-1
5. CALCULATION AND MEASUREMENT OF INSTRUMENT
SIGNAL AND NOISE 5-1
5.1 ACTIVE SIGNAL 5-1
5.2 PASSIVE SIGNAL 5-4
5.3 NOISE 5-5
6. RESULTS 6-1
6. 1 INSTRUMENT RESPONSE 6-1
6.2 ACTIVE-MODE SPECTRA 6-1
6.3 PASSIVE-MODE SPECTRA 6-22
6.4 ABSORPTION COEFFICIENT OF OZONE 6-22
6.5 EVALUATION OF OZONE READOUT 6-24
7. DISCUSSION AND RECOMMENDATIONS 7-1
APPENDIX DETAILED OPERATING INSTRUCTIONS A-l
vii
-------
BSR 3027
ILLUSTRATIONS
Figure Title
1-1 Block Diagram of Ozone Monitor
2-1 Transmitter Unit
2-2 Receiver Unit
2-3 Electronics Console
2-4 : Transmittance and Reflectance of Beam Splitter and
Auxiliary Filter 2-8
2-5 Filter Wheel Data Report 2-9
2-6 Filter Wheel Transmittance 2-10
2-7 Automatic Range Control Amplifier 2-16
2-8 Synchronous Detectors, Integrators, and Multiplexer 2-17
2-9 Synchronization Block Diagram 2-19
2-10 Computer Interface Logic 2-20
3-1 Initialize Routine 3-5/3-6
3-2 Auxiliary Routine 3-7/3-8
3-3 Interrupt Service Routine 3-9/3-13
3-4 Subroutines 3-14
4-1 Output From HEXED Program 4-2
4-2 Output From OMAS Program 4-3/4-4
4-3 Output From Regression Program 4-6/4-7
4-4 Output From Coefficient Converter Program 4-8
6-1 Response Curve of Instrument With Transmitter 6-2
6-2 Spectra Obtained With Ozone Monitor 6-3/6-12
6-3 Mean Spectra for ID 60 and 79 6-13
6-4 Polystyrene Spectra 6-14
6-5 Spectra With Ozone Cell in Beam 6-15/6-21
6-6 Passive Spectra 6-23
6-7 Ozone Absorption Coefficients 6-25
6-8 Ozone Monitor Readings 6-27
A-l Oscilloscope Trace From AFC Phase Demodulator A-11
IX
-------
BSR 3027
TABLES
Table Title Page
2-1 Approximate Source Temperature 2-7
2-2 Filter Wheel Wavelengths 2-11/2-12
3-1 Glossary of Computer Program Symbols 3-15/3-17
3-2 Program Branch Explanations 3-18/3-19
6-1 Calculation of K (9.7jjLm) From Spectra 6-24
A-l Test Panel Switch Position Identification A-9
XI
-------
BSR 3027
SECTION 1
INTRODUCTION
The Aerospace Systems Division of The Bendix Corporation has built a long-
path spectrophotometric instrument for the quantitative measurement of ozone in
unconfined ambient air. This work was performed pursuant to Contract No.
CPA 22-69-55 with the National Air Pollution Control Administration, Environ-
mental Health Service, Public Health Service, Department of Health, Education,
and Welfare.
i
The instrument is a single-beam infrared spectrophotometer with a sample
space up to 1. 6 km (1 mile) long and a minicomputer to analyze the results in real
time. The instrument is sensitive to ozone concentrations ranging from 20 |a g/m^
(0. 01 ppm) to 2000 (jg/m3 (1 ppm.)., The instrument can also be used as a spectro-
radiometer to measure the spectral radiant power received from passive sources.
This versatile instrument is designed for use in a wide range of research studies,
but for brevity will sometimes be referred to here as an Ozone Monitor. Figure 1-1
shows a block diagram of the instrument.
1-1
-------
BSR 3027
TRANSMITTER
SOURCE!
RECEIVER
AUXILIARY
DETECTOR /PASSIVE-MODE
CHOPPER
SAMPLED REGION
KM NOMINAL
ELECTRONICS CONSOLE
POSITION FEED-BACK
INTEGRATOR
H
MULTI-
PLEXER
.3
i
i
i
i
i
co
&
00
INTEGRATOR
A-D
CONVERTER
MAIN SIGNAL FLOW
COMPUTER-CONTROLLED
COMMAND SIGNALS
SYNCHRONOUS
RECTIFIER |
I
PHASE-LOCK
\
-
SYNCHRONOUS L.
RECTIFIER ~
H VARIABLE GAIN
AMPLIFIER
1
1
\-
1
t
H VARIABLE GAIN
AMPLIFIER
>\ i «
n
\
"i
i
-H-
i
i
r+-
— 1
H
i
CONTROL
PANEL
DIGITAL
COMPUTER
H
PREAMP
PREAMP
OSCILLOSCOPE
DISPLAY
TELETYPE
JL
PAPER TAPE
READER/PUNCH
Figure 1-1 Block Diagram of Ozone Monitor
1-2
-------
SECTION ?
SYSTEM DESCRIPTION
The source or transmitter unit utilizes aa electrically heated silicon carbide
and alumina source at the focus of a 32 cm (12. 5 in. ) diameter optical system.
The source radiant power is chopped at a frequency of 480 cps. The transmitter
projects a beam of radiant power to the receiver which is normally from 400 to
1600 m distant. The transmission spectrum of this optical path can be obtained,
and material present in this path that possesses specific absorption bands in the
7 to 14 micrometer (Mm) wavelength range of the instrument can be detected and
measured by its spectrum. Figure 2-1 shows a photograph of the transmitter.
The receiver contains 27 cm (10o5 in. ) diameter collecting optics which
focus the received radiant power onto a continuously variable filter wheel.*
This wheel consists of a multilayer interference filter on a germanium substrate,
and the wavelength of thepass-band of this filter varies as one goes around the
wheel. The wheel provided with the oaone monitor has a pass-band width that is
about 0.9% of the center wavelength, and the center wavelength varies from 7 to
14(jm with 180° rotation of the wheel. Thio filter wheel is used as the monochromator
element in this instrument. Relay optics refocus the energy on the mercury-
doped germanium infrared detector that is cooled with a mechanical refrigerator.**
The output of this detector, called the main detector, is processed to provide the in-
frared spectrum and/or readout of absorbing material in the beam.
A dichroic minor in the beam ahead of the filter wheel reflects a band of
wavelengths from 2.8 to 4. 3 fjim to an uncooled indium arsenide infrared detector.
This auxiliary detector is used to provide frequency and phase information about
the chopped radiant power from the distant transmitter. In addition, the auxiliary
detector can be used to obtain atmospheric transmission information in its wave-
length band. If atmospheric scintillation were a problem and if there is an instantaneous
correlation between the received signal in the main and auxiliary channels, then
the signal in the main channel could be normalized with the signal in the auxiliary
channel to reduce the effects of scintillation. Figure 2-2 shows a photograph of
the receiver unit with its cover removed.
*Made by Optical Coating Laboratory, Inc.., Santa Rosa, Calif. 95403; V. L. Yen,
Optical Spectra, May/June 1969, pp. 78-84.
**Malaker Corporation, High Bridge, N. J. 08829.
2-1
-------
BSR 3027
CM
O
N
Figure 2-1 Transmitter Unit
NOT REPRODUCIBLE
2-2
-------
BSR 3027
Figure 2-2 Receiver Unit
NOT REPRODUCIBLE
2-3
-------
BSR 3027
After preamplification in the receiver, the detector signals in the two
channels go through 15 m (50 ft) of cable to the electronics console where they
are further bandwidth-limited, amplified, synchronously rectified, integrated,
digitized, and read by a PDP-8/L digital computer. * The computer further
integrates the signals and, if requested, types out the spectrum on a Teletype,
punches the spectrum on paper tape using a high-speed punch, and/or displays
the spectrum in analog form on a cathode ray tube oscilloscope. Typically, a
50-point (wavelength) spectrum is obtained and punched every nine seconds. The
digital output tape permits an extremely high level of flexibility in subsequently
processing the spectra in an off-line computer. Figure 2-3 shows a photograph
of the electronics console before a high-speed paper tape punch/reader was installed.
The computer in the ozone monitor, using a base-line method, can also
calculate the amount of ozone and/or other infrared absorbing materials in the
beam. To do this, a larger off-line computer is first used to analyze spectra
taken with the ozone monitor. Wavelengths are selected that are useful for meas-
uring the ozone concentration, while simultaneously rejecting interferences. A
statistical analysis based on stepwise linear multiple regression is used.** In
brief, linear regression analysis uses a least-squares method to find a linear
equation that predicts the amount of ozone present from the wavelength readings.
Stepwise regression analysis adds promising wavelengths and ignores or deletes
unpromising channels to reduce the error of the prediction. Those channels are
selected which are well-suited for measuring ozone in the presence of those inter-
ferences which are actually present in the spectra. Regression analysis also deter-
mines weighting coefficients for each selected channel for use in forming a linear
combination of channel readings. The result of this off-line analysis is a list of the
wavelengths and their coefficients to be used in the ozone determination. This list
is then loaded into the PDP-8/L computer where the linear summation and the
ozone concentration are computed.
y = ex + c x + c x + . .. ~ CLx
•where y is the regression sum, the x's are the spectral readings at the selected
wavelengths, the c's are the corresponding coefficients, C is the average concen-
tration of ozone in the beam, L is the optical path length, and x is the mean of the
x's. The ozone monitor computer uses a digital low pass fdlter to smooth the cal-
culated amount of ozone over several spectra and then computes and prints out the
average ozone concentration in the beam. Initially, calibration of the instrument
was based on available ozone spectra, but ultimately the regression analysis and
ozone calibration could be based on independent measurements of ozone concentra-
tion obtained by other methods during the calibration spectral scans.
*Digital Equipment Corporation, Maynard, Mass. 01754.
**N. R. Draper and H. Smith, Applied Regression Analysis, New York, John Wiley
& Sons, 1966.
2-4
-------
BSR 3027
(M
Ob
3
Cxzct* MONITOR :
o
(M
Figure 2-3 Electronics Console
NOT REPRODUCIBLE
2-5
-------
BSR 3027
By using the same type of calibration technique, the ozone monitor can
simultaneously measure other materials having sufficiently strong infrared
absorption bands, such as sulfur dioxide.
The PDP-8/L computer also furnishes instrument control functions such as
providing pulses to step the filter wheel to a new desired position, checking a feedback
signal from the filter wheel to confirm the accuracy of the stepping, calling for
analog-to-digital (A/D) conversion, changing the amplifier gain so as to best use
the dynamic range of the A/D converter, and controlling the multiplexer to switch
between the main and auxiliary channels. Because the instrument operation is
largely computer controlled, changes in operation are readily made by reading
new instructions into the computer with a punched paper tape'rather than by
changing the hardware. For example, the number and locatibns of the wavelengths
that are examined can be very easily changed to meet different conditions.
To use the instrument as a passive radiometer, a chopper in the receiver is
used to chop the incoming radiant power. Frequency and pha'se information for the
synchronous rectifier and system timing is also derived from the chopper.
2. 1 TRANSMITTER
The source is a silicon carbide helix mounted on an alumina tube. * The
electrically heated region is 3/8 in. in diameter and 1 1/8-in. long. It takes a
nominal 200 W of electrical power at 80 V. The source is placed in a cavity in a
refractory material. A stream of cooling air is blown at the mounting end of the
source and around the outside of the source enclosure to ensure that the rest
of the transmitter unit is not appreciably heated. The source power supply contains
silicon controlled rectifiers and has a silicon cell mounted on the secondary mirror
of the optical system. This silicon cell provides optical feedback to keep the source
temperature constant. The source is designed to be run in the temperature range
of 1150 to 1320°C. Table 2-1 shows the approximate source temperature (as
measured with an optical pyrometer**) as a function of the control setting on the
source controller.
The ozone monitor was delivered with a controller setting of 1. 75. A 3600-rpm
synchronous motor drives the eight-blade d chopper in front of the source cavity.
This motor is air cooled with a fan.
*Obtainable from The Carborundum Company, P.O. Box 339', Niagara Falls, N. Y.
14302 as a Gas Igniter, Part No. E-2055-2.
**Mikael Bramson, Infrared Radiation, New York, Plenum Press, 1968, pp. 155-164.
2-6
-------
BSR 3027
TABLE 2-1
APPROXIMATE SOURCE TEMPERATURE
Controller Setting
Source Temperature (°C)
0.50
,0. 75
0.90
1.00
1. 25
1.50
1.75
2.00
2. 25
960
1100
1140
1165
1210
1235
1265
1295
1325
The source is at the focus of a Dall-Kirkham optical system with a primary
mirror diameter of 33 cm and 62. 2-cm system focal length. The optical system
is focused by rotating the secondary mirror.
The transmitter is mounted on a sturdy tripod with adjustments for height,
azimuth, and elevation.
2. 2 RECEIVER
The receiver optics include a Dall-Kirkham optical system with a primary
mirror diameter of 28 cm, a secondary mirror diameter of 15 cm, and a system
focal length of 40 cm. The specified blur circle diameter is less than 0. 25 mm.
Figure 2-4 shows the transmittance and reflectance curves of the dichroic beam
splitter at a 45 ° angle of incidence. The short-wavelength radiation is focused
through a filter onto the room temperature indium arsenide photovoltaic auxiliary
infrared detector. Figure 2-4 also shows the spectrum of the filter. The indium
arsenide infrared detector has a 1-mm-diameter area, a blackbody D* of 3. 15 x
10** cm-Hz^Z/W and a resistance of 56 ohms. The spectral response of indium
arsenide is very low beyond 3.9 ^m.
The long-wavelength radiation is focused on the filter wheel. The rear of
the filter wheel is the wavelength-selecting side. Figures 2-5 and 2-6 show the
manufacturer's data on this filter wheel. The filter wheel is driven by a pulse-
actuated stepping motor. Table 2-2 gives the center wavelengths corresponding
2-7
-------
8373-13
tsj
I
00
o
a
O
0)
a
-------
BSR 3027
SPECTRAL DATA REPORT CIRCULAR VARIABLE FILTER
CUSTOMER BEND IX
PURCHASE ORDER NO. K-4140
OCLI WORK ORDER NO. 14-5738-761
3ATCH NO. 12M-1-62-299
ITEM l.D. 1
THETA
0.
30.
60.
90.
120*
150.
180.
LAMDA 0
6.9767
7*9818
9.1884
1094005
11.6265
12.7917
13.8434
PCT HBW
0.8351
0.9707
0.9237
0.8689
0.8536
0.8236
0.7801
VHbW
0»0329
0.1026
0.0557
0.0209
0.0143
0.0444
0.0878
THETA1 LAMDA'e DEVIATION
0.
30.
60.
90.
120.
150.
180.
6.9018
8.0682
9.2345
10.4009
11.5672
12.7336
13.8999
1*0858
-1.0702
-0.4987
-0.0030
0.5130
0.4563
-0.4064
AVERAGE HBW * 0*8680
COLUMN EXPLANATIONS
THETA
LAKDA 9
HBW
VHBW
THETA'
LAMDA'6
ANGULAR POSITION ON FILTER IN DEGREES
BANDPASS CENTER WAVELENGTH IN MICRONS* AT
BANDPASS WIDTH AT THE HALFPOWER POINTS IN
(AVERAGE HBW)- HBW
ANGULAR POSITION FOR PURPOSES OF BEST FIT STRAIGHT LINE
POSITION THETA
PERCENT OF LAMDA
= CENTER WAVELENGTH VALUES FOR BEST FIT STRAIGHT LINE*
LAMDA'e = K * THETA'
DEVIATION = (LAMDA 8 - LAMDA'9 * 100/LAMDA 6
Figure 2-5 Filter Wheel Data Report
2-9
-------
BSR 3027
UJ
o
<
H
V)
(N
A
7n
?n
0°
I
j
I
TRANSMI
£0.1% B
ABOVE B
TTANCE:
ELOW AND
ANDST015.0;
30°
1
X. tn
>'
TE
NO
60°
MPERATURE'
RMALINCIDE
S
1
K>°
I
1 i
,.
325°C
NCE
120°
,
1
1
1 BO-
ISO"
1
1500 1400 1300 1200 1100 1000 900 800 70(
WAVE NUMBER (CM'1)
Figure 2-6 Filter Wheel Transmittance
2-10
-------
8378-A
F«Plli UCIAL
0
1
2
3
4
5
6
7
a
9
10
11
12
13
1*
IS
16
17
18
IS
20
21
22
23
24
25
26
27
28
29
30
31
32
' 33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
0
1
2
3
4
5
6
T
10
11
12
13
14
15
16
17
20
21
22
23
24
25
26
27
30
31
32
33
34
35
36
37
40
41
42
43
44
45
46
47
50
51
52
53
54
55
56
57
60
61
62
63
6<>
65
66
67
70
71
72
73
74
75
TABLE 2-2
FILTER WHEEL WAVELENGTHS (Page 1 of 2)
CHAN M1CKJN
6.97
6.98
7.00
7.01
7.03
7.0*
7. C6
7.C7
7.C8
7.10
7.11
7.13
7.14
7.16
7.17
7.19
7.20
7.22
7.23
7.25
7.26
7.28
7.29
7.31
7.32
7.34
7.35
7.37
7.38
7.40
7.42
7.43
7.45
7.46
7.48
7.49
7.51
7.52
7.54
7.56
7.57
7.59
7.60
7.62
7.64
7.65
7.67
7.68
7.70
7.72
7.73
7.75
7. 76
7.78
7.80
7.81
7.83
7.85
7.36
7.88
7.89
7.91
FdPUS LT.TAL (.HAN
f-HPOi iKtAL CHAN
OCTAL CHAI» "ItCRUN
62
63
64
65
66
•>->
68
69
70
71
72
73
74
75
76
77
78
79
80
81
B2
83
84
85
86
B7
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
104
110
111
112
113
114
115
116
117
118
119
120
121
122
123
76
77
ICO
101
102
'03
104
105
106
107
110
111
112
113
114
115
116
117
120
121
122
123
124
125
126
127
130
131
132
133
134
135
136
137
140
141
142
143
144
145
146
147
150
1S1
1S2
153
154
155
156
157
160
161
162
163
164
16S
166
167
170
171
172
173
10
11
12
13
7.93
7.94
7.96
7.98
7.99
8.01
8.03
8.04
8.C6
8.C8
8.C9
8.11
8.13
8.14
8.16
8.18
8.20
8.21
8.23
8.25
8.26
8.28
8.30
e.32
8.33
8.35
e.37
8.38
8.40
8.42
8.44
8.45
8.47
8.49
8.51
8.52
8.54
8.56
8.58
8.59
8.61
8.63
8.65
8.66
8.68
8.70
8.72
8.73
8.75
8.77
8.79
8.80
8.82
8.84
8.86
8.88
8.89
8.91
8.93
8.95
8.97
8. 98
124
125
126
127
128
129
130
131
132
133
134
135
136
'137
138
139
140
141
142
143
144
145
146
147
148
149
ISO
151
152
153
154
155
156
157
158
159
160
161
162
163
U4
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
174
ITS
176
ITT
200
201
202
203
204
205
206
20T
210
211
212
213
214
215
216
217
220
221
222
223
224
225
226
227
230
231
232
233
234
235
236
237
240
241
242
243
244
245
246
247
250
251
252
253
2S4
255
256
257
260
261
262
263
264
265
266
267
270
271
14
15
16
17
18
19
20
21
23
24
25
9.00
9.02
9.04
9.06
9.07
9.09
9.11
9.13
9.15
9.16
9.18
9.20
9.22
9.24
9.25
9.27
9.29
9.31
9.33
9.35
9.36
9.38
9.40
9.42
9.44
9.45
9.47
9.49
9.51
9.53
9.55
9.56
9.58
9.60
9.t2
9.64
9.66
9.68
9.6*
9.71
9.73
9.75
9.77
9.79
9.80
9.82
9.84
9.66
9.88
9.90
9.92
9.93
9.95
9.97
9.99
10.01
10.03
10.05
IO.C6
10.08
10.10
10.12
1B6
107
188
It) 9
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
232
234
235
236
237
238
239
240
241
242
243
244
245
246
247
272
273
274
275
27t
277
300
301
302
303
304
305
306
307
310
311
312
313
314
315
316
317
320
321
322
323
324
325
326
327
330
331
.332
333
334
335
336
337
340
341
342
343
344
345
346
347
350
351
352
353
354
355
356
357
360
361
362
363
364
365
366
367
26
27
28
29
30
31
3Z
33
34
10.1&
1C.16
10.13
10.19
10.21
10.23
1C.25
1C. 27
10.29
10.31
10.32
10.34
10.34
1C.38
1C.40
10.42
10.44
1C.45
10.47
10.49
1C.51
10.53
10.55
10.57
1C.58
10.60
10.62
10.64
10.66
10.68
10.70
1C.71
10.73
1C. 75
10.77
10.79
10.81
10.83
1C.84
10.86
10.88
-10.90
10.92
10.94
10.96
1C.97
10.99
11.01
11.03
11.05
11.07
11.09
11.10
11.12
11.14
11.16
11.18
11.20
11.21
11.23
11.25
11.2T
ta
en
CO
o
-------
TABLE 2-2
8378 A
FWPUb OCTAL CHAN MICB'JN
FILTER WHEEL WAVELENGTHS (Page 2 of 2)
CI-4N
mi-US IILIAI. OAN MICC1M
248 370
249 3T1
250 372
251 373
292 374
293 379
11.29
11.31
11.32
11.34
11.36
11.38
264 376 35 11.40
299 377 11.42
296 400 11.43
237 401 11.45
298 402 11.47
259 403 11.49
260 404 ll.il
261 405 11.53
262 406 36 11.54
263 407 11.56
264 410 11.58
265 411 11.60
266 412 11.62
267 413 11.64
266 414 11.65
269 415 11.67
270 416 11.69
271 417 37 11.71
272 420 11.73
273 421 11.74
274 422 11.76
275 423 11.78
276 424 11.80
277 425 11.82
278 426 11.83
279 427 38 11.85
280 430 11.87
281 431 11.89
282 432 11.91
283 433 11.92
284 434 11.94
285 435 11.96
286 436 11.9B
287 437 39 12.00
288 440 12.01
289 441 12.03
290 442 12.C5
291 443 12.07
292 444 12.08
293 445 12.10
294 446 12.12
295 447 12.14
296 450 40 12.15
297 451 12.17
298 452 12.19
299 453 12.21
300 454 12.23
301 455 12.2A
302 456 12.26
303 457 12.28
304 *60 41 12.30
305 461 12.31
306 *t2 12.33
307 463 12.35
308 464 12.36
309 465 12.38
310 466 12.40
311 467 12.42
312 47C 12.43
313 471 42 12.45
314 *72 12.47
315 473 12.49
316 474 12.50
317 475 12.52
318 476 12.54
3i9 477 12.55
320 50C 12.57
3il 501 12.59
322 502 « 12.61
323 503 12.(2
324 50* 12. t4
325 505 12.66
326 506 12.67
327 507 12.69
328 510 12.71
329 511 12.72
330 !12 44 12.74
331 513 12.76
332 514 12.78
333 515 12.79
334 516 12.81
335 517 12.63
336 520 12.84
337 521 12.86
338 522 12.88
339 !23 45 12.89
340 524 12.91
341 525 12.93
342 526 12.94
343 527 12.96
344 530 12.97
345 531 12.99
346 532 13.01
347 533 13.02
3*8 £34 13.C4
349 535 46 13.06
350 536 13.C7
351 537 13.09
352 540 13.10
353 541 13.12
354 542 13.14
355 543 13.15
356 544 13.17
357 545 13.19
358 546 47 13.20
359 547 13.22
360 550 13.23
361 551 13.25
362 552 13.26
363 553 13.28
364 554 13.30
365 555 13.31
366 556 13.33
367 557 48 13.34
368 560 13.36
369 561 13.37
370 562 13.39
371 563 13.41
372 564
373 565
374 566
379 567
376 57C
13.42
13.44
13.45
13.47
13.48
377 571 49 13.50
378 572 13.51
379 573 13.53
380 574 13.54
381 575 13.56
382 576 13.57
383 S77 13.59
384 600 13.60
385 601 13.62
386 602 13.63
387 603 50 13.65
388 604 13.66
389 605 13.68
390 606 13.69
391 607 13.71
392 «10 13.72
393 611 13.74
394 612 13.75
395 613 13.77
396 614 13.78
397 615 13.80
398 616 13.81
399 617 13.83
w
to
co
o
-------
BSR 3027
to each filter wheel position (FWPOS). The octal equivalent of each filter wheel
position is also given. (The PDP-8/L computer accepts the filter wheel position
list in octal. ) Also shown are the present channel locations (CHAN). The wave-
length information in Table 2-2 was obtained by a least-square fit of a cubic
equation to the data in Figure 2-6. The filter wheel reference position (FWREF),
which is the filter wheel position where the computer recognizes the light shining
through the narrow slit in the rim of the filter wheel, was found to be (373)g or
11. 34 pm. The spectral resolution of the filter wheel mounted in the ozone monitor
was not measured, but it will be somewhat larger than shown in Figures 2-5 and 2-6.
This is true because of the range in the angle of incidence upon the filter wheel due
to the f/1. 5 optical system and because of the non-zero size of the source image
on the filter wheel, which can be as large as 1 mm—although in the active mode at
long ranges it will be less, perhaps 0. 25 mm.
Cemented to the outside of the filter wheel is an aluminum rim and flange.
The flange normally does not permit light from a small incandescent bulb to reach
a phototransistor on the other side of the flange. But a narrow slot in the flange
allows the light to reach the phototransistor for a small range of filter wheel angle
when the filter wheel is transmitting in the vicinity of 11. 34 pm. This is used as
a check by the computer to assure that there have been no mistakes in stepping
the filter wheel. If a mistake is found and if the computer switch registers are
properly set, the spectrum is rejected, the instrument spins the filter wheel to re-
acquire the light, and the instrument proceeds to take a new spectrum.
In front of the filter wheel is a 16-blade chopper. When the ozone monitor is
used in the active mode, the chopper positioned with a locking screw so that it
does not block the beam. When the instrument is used in the passive mode, the
locking screw is withdrawn and the chopper motor is turned on. The synchronous
chopper motor turns the chopper at 1800 rpm. The blades are deeply grooved and
painted black so as to simulate a blackbody at ambient air temperature. An
incandescent light is on one side of the chopper and a phototransistor is on the
other side. The light and phototransistor can be mechanically phased relative
to the chopper, and are used to generate timing pulses when the instrument is in
the passive mode.
An aperture close to the back of the filter wheel rejects stray light that
might cause spurious signals. Two pierced paraboloidal mirrors reimage the
radiant power from the back of the filter wheel through a window onto the main
infrared detector with unit magnification. A circular disk between the two mirrors
prevents the detector from receiving chopped radiant power that does not come
from the mirrors and prevents a possible spurious signal when in the passive mode.
2-13
-------
BSR 3027
The disk does not obscure any desired energy as it corresponds to the central
blocked region in the hollow cone of light from the Gassegrainian-type collector
optics.
The mercury-doped germanium infrared detector is mounted on the cold
surface of the refrigerator. It is enclosed in a vacuum jacket to prevent convective
heat input. A 1-mm-thick IRTRAN-2 window serves as the entrance to the vacuum
jacket.
Mounted on the vacuum jacket is a three-position detented wheel. One position
leaves the beam open, another blocks the beam, and the third position inserts a
0. 0012-in. thick sheet of polystyrene in the beam. In normal use, the beam is
unblocked; the other two positions are useful for test and wavelength calibration
purposes, especially in the passive mode. (Filters should be inserted in the beam
between the chopper and the infrared detector, if the infrared radiance of the
filter is not to be chopped. )
The main infrared detector has a cold collar around it to restrict its field
of view to about 30° full angle. It has a 1-mm-diameter sensitive area and was
measured to have a blackbody D* (500 K, 450 Hz) of 3.6 x 10^ cm - Hzl/z/W and a
cold resistance of about 700 kn. It is operated at a constant voltage bias of 5 V that
is current-limited at 88(0. A to prevent damage to the detector when it is warm. The
preamplifier and bias circuit has low impedance so that the,detector responsivity
does not change with varying photon flux, such as can be caused by varying ambient
temperatures or rotation of the filter wheel.
The refrigerator is mounted in a large cylinder that can be accurately
positioned in three dimensions by turning threaded bolts and the adjustment can
then be securely fastened.
2. 3 ELECTRONICS
2. 3. 1 Source
Power is supplied from the 115-VAC, 60-Hz mains and used to operate the
synchronous chopper motor and the source power supply. The source power
supply uses optical feedback to maintain the source at constant temperature.
2. 3. 2 Receiver
Spectral multiplexing is accomplished in the receiver by advancing a filter
wheel from one spectral region to the next with a stepper motor. Control signals
2-14
-------
BSR 3027
are generated in the console by the PDP-8/L computer, and the stepper motor has
the capability of moving forward or reversing at any speed up to 1000 steps/second.
One filter wheel position is monitored by a photodiode, lamp, and amplifier
combination and transmitted to the computer. For other positions, the computer
registers the number of pulse signals sent to the stepper motor and thus keeps
count of the channel position of the filter wheel.
The main preamplifier also provides detector bias. It is a constant voltage
source that is, however, cur rent-limited to avoid burning out the infrared detector
when the detector is at room temperature and has a low resistance.
2. 3. 3 Central Control Electronics
i
The central electronics consist of interface electronics, computer,
oscilloscope, and required power supplies. The following paragraphs explain
the function of each block and give typical schematics.
2. 3. 3. 1 Automatic Range Control Amplifiers
The detected signals from the receivers are accepted at the central
electronics by the automatic range control (ARC) amplifiers. Each amplifier,
as shown in Figure 2-7, consists of three integrated differential amplifiers. Gain
can be controlled by the field effect transistor switches associated with the first
two amplifiers as shown and can take on the following values: 1, 2, 4, 8, 16, 32,
64, or 128, developed by a three-bit binary code from the computer. The third
amplifier is connected as a unity gain inverter.
2. 3. 3. 2 Synchronous Detectors, Integrators, and Multiplexer
Sunchronous detection is accomplished by the action of the switches Si,
82, 83, and 84 (as shown in Figure 2-8) on the split-phase signals from the ARC
amplifiers. The control signals for the switches are produced by the phase-locked
synchronization circuitry to be described in Section 2. 3. 3. 3. The integrals of the
detected signals are produced by amplifiers Al and A2. Switches S5 and S6, under
the control of the computer, reset each integral to zero after each digitized reading
is taken.
Switches S7 and S8, also under control of the computer, select either the
main or the auxiliary integral for delivery to amplifier A3, which offsets the selected
signal to the minus five volt reference level required by the A/D converter.
2-15
-------
8378-2
SI
ts)
Input
from
Receiver
rv^
O Outputs
To
S ynchr onou s
O Detector
.A. J.
r
S2
td
en
O
to
Figure 2-7 Automatic Range Control Amplifier
-------
8378-4
I
I—'
^1
-WV-
From
Auxiliary
ARC
Amplifier
A3
A2
To A/D
Converter
W
en
o
IV
-J
Figure 2-8 Synchronous Detectors, Integrators, and Multiplexer
-------
BSR 3027
2. 3. 3. 3 Synchronization Circuitry
Timing signals are derived from either of the two mechanical choppers
in the system. When the active mode is selected, the auxiliary signal serves as
the phase reference; in the passive mode, a signal from the local chopper in the
receiver is used.
The selected signal is used as the phase reference in the phase-locked
control loop illustrated in Figure 2-9.
The voltage-controlled oscillator shown generates a 3840-Hz pulse train
which is successively divided by two to produce the 480-Hz signals needed for the
synchronous detection of the main and auxiliary signals and the 480-Hz interrupts
to the PDP-8/L.
A 480-Hz. signal is also used to detect synchronously the selected phase
reference signal as shown. The synchronous detector output, after appropriate
filtering, controls the oscillator frequency to maintain phase, lock.
In the event of the disappearance of the phase reference, which can occur
as a result of mis pointing in the active mode, the level detector causes the selection
of the 60-Hz power frequency for use as a timing reference.; This procedure
maintains system timing at the correct frequency until the proper phase reference
signal is restored. :
2.3.3.4 Computer Interface Logic
This section of the central electronics provides the signal conditioning,
level translations, and switching'functions necessary for communication between
the computer and the remainder of the system. Signal data, synchronizing pulses
(filter wheel and chopper reference), and discrete data from the software selection
switches on the front panel are received, properly conditioned, and made available
to the computer under program control. All .computer outputs (except hard copy
which is interfaced in the Teletype unit) pass into this subsystem and are conditioned
to provide the oscilloscope.display, gain control signals for the ARC amplifiers,
integrator dump signal, and filter wheel forward and reverse signals.
These functions, block-diagrammed in Figure 2-10, are accomplished
in the following manner:
1. Signal data input from the receiver is processed by the integrate
and dump technique. The dump is controlled by a device selector
2-18
-------
BSR 3027
Auxiliary Signal
— p^Active Mode
Chopper Ref
Signal
_ , , , ,
Pafl8ive Mode
Power Line
Signal (60 Hz)
in
cb
r-
n
oo
480 Hz
Synchronous
Detector
f\f\
Synchronous
Detector
60 Hz
Binary
Frequency
Divider
Level
Detector
Select 480
I
Select 60
(Line Lock)
Sampled
Low-Pa s B
Filter
Voltage -
Controlled
Oscillator
3840 Hz
i
480 Hz
Interrupts
to Computer
480 Hz
Control Signals
to Main and
Auxiliary Synchronous
Detectors
Figure 2-9 Synchronization Block Diagram
2-19
-------
BSR 3027
PDP^S-*- Buffer
Register
D/A
Converter
' ' ^•••••••••••••i
FDI^O * Buffer ARC Amps ^ Gain
Register Control
£
I
i
A/D p—
Converter-^-IMux | *
To
PDP-8
tepper ^ Wheel
Electronics Motor
\
I-.-
_^i__ _^_^^ TI if , : ii
-, Pump
iDump •
Gating ^
Circuits ^(_
i \
ob PDP-8 »-| Device Selectors 1
5 L J
Figure 2-10 Computer Interface Logic
2-20
-------
BSR 3027
circuit which provides a pulse during the execution of the
appropriate computer instruction. The multiplexer selects
an integrator whose output is digitized by the A/D converter.
2. The synchronizing signals and discrete program control data are
entered into the computer through gating circuits under control
of device selectors. The chopper sync pulse is reshaped to make it
compatible with the requirements of the computer.
3. The oscilloscope display data are passed from computer storage
; ; into a buffer register in the interface under control of a device
; selector. The D/A conversion is performed in the unit on the contents
of the register and the resultant voltage is sent to the oscilloscope
display unit. ;
4. The three bits of information necessary to set gain levels in the
ARC amplifiers likewise are stored in buffer registers. Storage
of the stepper motor pulses for filter wheel positioning is not
required. These come directly from the appropriate device selector
and are lengthened, amplified, and sent to the stepper motor drive
module.
2.3.4 Signal Processing
The demodulated signal must be integrated for a period of time sufficient
to yield the desired S/N ratio. The integration is accomplished partly by analog
and partly by numerical techniques. The analog integration period is 2083 (jsec,
corresponding to one chopping cycle of the 480-Hz chopping frequency. At the
end of the analog integration period, the analog integral is digitized and the
integrator is then issued a reset command by the computer. While the next analog
integration is in process, the computer adds the latest reading to the sum of
earlier readings from the same spectral channel. The total integration time is
determined by multiplying the analog integration time by the number of terms in
the summation, currently 64 (stored in the constant NOCPC), and is easily changed
to other values should a larger signal be required for any reason.
The signal processing performed by the computer can be divided into three
parts: (1) the basic instrument control and display program, (2) the data gathering
program, and (3) the detection processing program.
The chopper reference signal provides the fundamental timing for the
instrument control program. The end of a chopping cycle generates an interrupt
indicating that a priority task must be performed, via the interface unit, which
2-21
-------
BSR 3027
causes the computer to enter the main interrupt servicing program. The analog
input from the interface unit integrator is digitized, and the command is issued to
reset the integrator—again via the interface unit. The computer adds the new
reading to the accumulated sum of readings and increments the running count of
integrator readings by one. If the count has not reached the required number, the
main interrupt servicing program is terminated, and the computer executes a
display program to be described later. Interrupt programming is required to per-
form the essential instrument control tasks (e.g. , moving the filter wheel) in a
timely manner.
If the count of integrator readings has reached the required number, it is
reset to zero, and the computer corrects the sum of accumulated readings for
the ARC gain level in use and places the sum in a list of sums ordered by spectral
channel number. The spectral channel count is then incremented. If the specified
count has not been reached, the computer selects a new filter wheel position and
initiates filter wheel stepping to the specified wavelength for the next channel. The
new wavelength can be any of those available on the wheel, but usually the wavelength
assignments to the channels will be ordered to minimize filter wheel stepping time.
The filter wheel stepping rate is varied by the computer to control the
angular acceleration and deceleration. The maximum stepping rate of 480 steps/
sec is reached in three steps and maintained until the filter wheel is within three
steps of its destination. Each step corresponds to a wavelength shift of about
0. 0175 |o.m.
When the filter wheel has been repositioned, the summation process of the
digitized readings resumes at the new wavelength.
When a spectral scan has been completed by finishing the summation for
the last channel, the computer causes a runback to the first filter wheel position
and starts a new spectral scan.
i %
Each time the filter wheel passes the position at which the reference position
signal is generated (at 11. 34|jm), the computer checks for agreement' with its own
notion of the filter wheel position. If the check fails, the computer will, at the
operator's option, either halt or reinitialize the spectral scanning.
The computer, during each complete spectral scan, determines the magnitude
of the highest reading obtained and uses this number to establish the most desirable
signal gain to be used for the next spectral scan. The desired gain command is
sent via the interface unit to the gain control circuits in the amplifier chain.
2-22
-------
BSR 3027
The oscilloscope provided is driven by the computer with the aid of a D/A
converter in the interface unit. The display program sequentially outputs the
entire list of summed channel readings available from the last complete spectral
scan to the D/A converter, yielding a bar chart display showing the channel
readings for each channel in sequence. The entire display is repeated at the 480-Hz
chopping rate. Each individual channel is displayed in about 15 |jsec, requiring
750 of the available 2083 (j sec to display 50 channels.
When a display cycle is completed, the computer returns to any auxiliary
calculations with which it may.have been engaged at the time of the chopper
interrupt.
The auxiliary programs, data gathering, and detection processing, do not
require execution in strict synchronism with the 480-Hz chopping signal and, if
called for, are executed in the computer's spare time between interrupts.
The user can invoke the data gathering mode by operating the appropriate
control switch. The computer will output a complete list of channel readings, each
consisting of a four-digit number.
The user can invoke the detection processing mode by operating another
switch. In this mode, the computer sums any combination of spectrally derived
channel signals, with selectable polarity and variable weighting constants of any
desired value, to yield a single output. If desired, the computer can simultaneously
evaluate several such equations, each yielding its own output. The number of such
equations that can be processed depends on the number of channels in use. For 64
channels, 10 different equations can be concurrently processed to detect the
presence of 10 different air pollutants, or 10 different ways of detecting one pollutant,
or any other combination. Thus, several pollutants can be detected simultaneously,
and each can be evaluated in several ways.
The single output from each equation is displayed as an auxiliary bar chart
on the oscilloscope display and is typed out, along with an identifying message, to
provide the quantitative information concerning the concentration of the target pol-
lutant or pollutants.
2-23
-------
BSR 3027
SECTION 3
DISCUSSION OF COMPUTER PROGRAM
The Ozone Monitor is designed to gather and process real-time spectra to al-
low remote sensing of atmospheric pollutants, which absorb in the infrared spectral
region. The accompanying PDP-8/L computer and software are responsible for
providing both the control of the instrument and the processing of data.
i • .
: For each complete spectral scan, a linear combination of the channel readings,
called the regression sum, is computed, if called for. The set of coefficients for
this equation is different for each pollutant to be measured. The type of spectra
gathered can be varied to include anywhere from zero to 64 channels spaced ac-
cording to the measurement problem. Both the number of channels and the spectral
location of the channels can be varied by the user. The optimum choice of these
variables is based upon knowledge of the infrared absorption spectra of the indi-
vidual pollutants and the absorption characteristics of possible interferences.
One method of determining the number of spectral channels, the spectral
location of these channels, and the coefficients for each channel is the use of step-
wise multiple regression. This is the method used to provide the coefficients pro-
vided for the Ozone Monitor when it was delivered. An example of the use of this
method is shown in another section of this report. In essence, this method selects
channels and coefficients that can predict (in a least-squares sense) the amount of
ozone independently stated to be present in each spectrum of a training set. One
way of attaining this training set of spectra is to obtain spectra under various
atmospheric conditions and to have an independent measurement of the average ozone
concentration in the path during the time each spectrum is taken. Another way of
attaining this training set is to obtain a number of spectra when the amount of ozone
in the beam is independently known to be very small (essentially zero) and then,
knowing the absorption coefficients for ozone as a function of wavelength, to modify
these spectra to simulate the presence of an arbitrary known amount of ozone. This
latter method was used in the regression analysis performed under this contract.
Other simpler methods can also be used, such as the selection of channels from
inspection of absorption spectra of the target pollutant and possible interferences.
In this case, preliminary coefficients can be obtained by simple calculation based
on a few spectra obtained by the ozone monitor and refined by trial and error.
The regression sum is divided by the average signal (using all channels to ob-
tain the average). The purpose is to give a quotient that is independent of signal
3-1
-------
BSR 3027
strength and proportional to the product of ozone concentration and path length (or
range). This quotient is divided by the range so as to give a result in concentration
units (\i. g/m^).
Large amounts of ozone cause sufficiently large absorption at 9.5 and 9-7 jim
so that the linear approximation that is used, namely that the absorbance A is pro-
portional to the fractional absorptance a (A = a/2.3), no longer holds. For these
large amounts of ozone, a calibration curve can be used to obtain the true amount
of ozone from the indicated amount. Alternatively, the computer itself can be pro-
grammed to calculate the true absorbance, for example, by taking the logarithm of
the transmittance T of the 9.7 jim spectral channel, and then calculating the ozone
concentration from the absorbance (A = -log T).
The PDP-8/L software provides control of all instrument functions, such as
filter wheel positioning, chopper wheel synchronization o*f the channel processing,
data recording, and the outputting of spectral readings and concentration messages.
For discussion purposes, the instrument is said to be operating in the data-
gathering mode when punched paper tapes containing channel readings are being
generated. This output is compatible with the hardware of a larger computer, such
as the IBM-360/50 for use in determining optimum detection coefficients or in per-
forming any other analysis requiring digitized spectral channel readings as input.
The alternative mode of operation, the detection processing mode, uses the same
kind of channel readings to provide for measurement of those pollutants it has been
programmed to detect. In either mode, the function of the instrument is to provide
a number proportional to the signal in each spectral channel. The discussion of the
software will thus follow the internal logic required for instrument control in the
gathering of channel readings. Routines specific to either of the two modes of oper-
ation will be noted as necessary.
The primary output of the PDP-8/L software is a spectrum or an ordered set
of numbers with each proportional to the most recently observed signal in a spectral
channel. This ordered set of numbers is the result of a complete spectral scan,
each of the readings corresponding to one spectral location in the range from 7 to
14 nm. This set of numbers is considered a single observation. A typical obser-
vation might consist of 50 channels uniformly distributed across the spectral range
of the instrument. Both the number and the spectral location of channels comprising
a single observation are controlled by input options to the PDP-8/L software.
At the completion of each observation, the result is plotted against channel
number on an oscilloscope display and is available for use in the detection processing
mode if this user option is in effect. During the execution of this option, the system
provides information concerning the concentration of pollutants. This information
is presented in two ways:
3-2
-------
BSR 3027
1. The Teletype will print a message identifying the pollutant and will type
a number proportional to the average concentration of the gas in the path
length between the transmitter and the receiver.
2. A bar proportional to the average concentration of pollutant will
appear on the oscilloscope display.
If detection processing is not desired, the system will continue to gather and
display spectra while awaiting further instructions from the operator. If, in the
process of observation, the operator notes an interesting spectrum, he has the
option of printing out on the Teletype a numerical representation of that spectrum.
He need merely press the PRINT DATA button on the control panel to exercise this
option. The software will sense this request, forego data gathering at the com-
pletion of its present scan, and transfer control to the Teletype which prints out a
channel- by -channel list of numbers. To keep a log of his observations, the oper-
ator has at his disposal an option that enables him to provide an ID number for each
observation. To do this, he must press the ID INCREMENT button for a moment
and then release it. This will result in the Teletype printing out an ID number
labeling the next observation or set of observations.
Operation in the data-gather ing mode requires the user to turn on the power
to the high speed punch and press the PUNCH DATA button. The software will
then output the channel readings to the punch at the completion of each spectral
scan. The ID option can also be used here to label observations. Experience indi-
cates that it is desirable to record supplementary information such as ambient tem-
perature in the log book under the ID number for that observation.
Only about six or eight channels may be needed for measuring the ozone con-
centration in the air and rejecting a few interferences. (Seven channels have non-
zero coefficients in the processing scheme delivered with the instrument.) In this
case, better results can be obtained by only looking at these few wavelengths, rather
than by spending time looking at a larger number of wavelengths, most of which are
essentially unused. On the other hand, for surveying the spectra of the air in a new
location many spectral channels may be desired. In other words, although a general
purpose list of wavelength channels can be used, and were delivered with the instru-
ment, better results can be achieved on any specific problem by using only those
wavelength channels required by the problem.
3.1 PROGRAM FLOW
The Ozone Monitor software for the PDP-8/L. consists of an initializing rou-
tine (INIT), an auxiliary computation and output routine (AUX), an interrupt ser-
vice routine (ISR) to provide timely control of the instrument, and subroutines for
3-3
-------
BSR 3027
control of filter wheel positioning and error exits. The flow charts shown in Fig-
ures 3-1 through 3-4 exhibit the logical structure of these programs.
Table 3-1 defines the variables which appear on the flow charts. The variables
are listed in order of first appearance to assist the reader in following the logical
flow of the program. The decision branches are numbered approximately sequen-
tially, and at each of these points the computer will be asked whether a specified
flag is zero or whether a certain inequality is satisfied by certain variables. Ad-
ditional assistance in interpretation can be gained from a study of the tabulated
questions which appear in Table 3-2.
3.1.1 INIT Routine
INIT consists of three main branches (Figure 3-1) each of which appears as
a separate column on the flow chart. The first branch is executed once to perform
initializing operations required before the instrument begins routine activities.
The purpose of this branch is to determine the operator instructions as set on the
switch register and to set the program control flags which will return control to
INIT upon completion of ISR activities. The first branch also punches out the filter
wheel position list if this is desired. The non-zero value of INITF, set in the exe-
cution of Branch 1, causes ISR to return control to INIT without first enabling inter-
rupts.
After the first interrupt, which will occur after the completion of Branch 1,
each successive interrupt will occur at the end of Branch 2 until the filter wheel is
mechanically positioned to its reference position. When Branch 2 is completed,
control is passed to location INIT 7, the beginning of the third branch, instead of to
the Interrupt Service Routine. INIT 7 is executed once to set the values of all re-
maining variables and to set INITF equal to zero so that future interrupts will not
return control to INIT 1, the starting location of the second branch. Branch 3 then
enables interrupts and passes control to the auxiliary routine beginning at location
AUX.
3.1.2 Auxiliary Routine (AUX)
The purpose of the auxiliary routine (Figure 3-2) is to perform Teletype
output operations and numerical computations. These activities are carried out
during times when the computer might otherwise be resting from its instrument
control obligations. While awaiting interrupts from the output hardware or the
chopper wheel, the computer is afforded sufficient free time to process the channel
readings for detection processing. Teletype and punch outputs delivered by AUX
include the list of channel readings with ID code appended and the pollutant con-
centration messages.
3-4
-------
8378-3
INIT
IN1T F = 1
INIT Light On
AC = Switch Register
SIMF = Bit 11 of AC
PNHLD =0
AC - Switch Register
AC = Bit 1 of AC
Yes
INITID
No
TEM1 = - NOC
RBL (1) = NOC
TEM2 = NOC
I = 2
J= 1
RBL (I) = FWPL (J)
TEM2 - TEM2 + FWPL (J)
Yes
| TEM2 - TEM2 + 1 |
1
No
TEM1 - TEM1 -f 1
T
INIT1
DWD = Input From Interface Unit
AC = Bit 11 of DWD
Yes
No
Step Filter
Wheel Clockwise!
FWCC = FWCC +1
Yes
FWERR
FWHLD = 0
No
AC = FWHLD
Yes
No
1 Enable Interrupts 1
1
1 I Wait for Interrupt I
-------
OO
I
I IN
= J
I -. I + 1
RHL (I) -
PWD = RBL (1)
PCNC = -(NOC
PCC - -4
PWC - - 14
PNHLD = 1
CRF = 1
^)
J.
Punch Leader
1N1T1D-
PNCF = 1
FWCC = -10Z4
MFLG = 0
FWHLD =1
I
Clear Chopper
Interrupt
I Enable Interrupts I
\Va it To f
"rfffcr'rupt" I
1
FWPOS = FYv'REF
CNCOM = 1
CHAN = NOC
FWHLD = 1
AGC = 0
Output AGC to
Interface Unit
RDIX = -NOC PC
SUML = 0
SUMH = 0
CHNCNT = -NOC
FWDF = 1
FMHLD = 0
CRTF = 0
CHR = 0
Set All
Regression Sum
and Regression
Sum Display
Elements Equal
to Zero
DETF = 0
DETPF = 0
DETTF = 0
CTBP = 0
ADOF = 0
INITF = 0
IDF = 0
TBS = -1
td
co
o
I. Enable Interrupts
AUX
Figure 3-1 Initialize Routine
-------
AUX
fes I
PEX
No
Move Channel
Readings From
CHRDL to RBL
Append ID Code
to RBL
Yes [No
PEX IPNCF = o 1
^^^]T^^^^
Output Next
Character to
Punch
_^r^ Character?^^^^
Yes| No
1 CRTF = 0 I 1 PNHLD = 0 I
Yes
No
PNHLD = 1
_L
] I
FMHLD = 0
Correct Readings
For Current Gain
levels
Compute RBL
Check Sum and
Append tn RBL
JL
_L
DTEMl = . NOE
PEX
Set Punch Control
Counters to
Initial Values
AUX
Multiply Current
Channel Reading
by Coefficient for
Current Channel
and Current
Regression Equation
J.
Add Product to
Current Regression
Sum
_L
DTEMl = DTEMl -I- 1
m
oo
Yes
I No
-------
BSR 3027
Yes
| DETTF = 1 |
DE1
r 3
|DTEM1 = 1 - NOE I ;
1
i"
_fa RGS(DTEMl).
"w" - RQsd). RNGC
1
LPRS = LPR'S. + 2~mT(FRGS-LPRS)
| _
"^
*
| SPECNT = SPECNT +1 | ;
^^•^""SPECNT = oT^Ss^
Yes
Output Messa
and LPRS on
Teletype
. ...^ .,
ge
SPECNT = SPECNO |
...... » i.^.
No
DTEM1 = DTE Ml * 1 I ,4
^jf^DTEMl = 0?*%^
Y«
a- :
No
'
DET 3 »•
| DTE Ml = - NOE |
1
_ . J?
IRGS (DTEM1) = 0 1
| DTEM1 = DTEM1 t 1 |
^^^X^DTEMl = 0?V^^
YesIL .-.., JNo
1DETPF = 01 j ' '
AVX
Figure 3-2 Auxiliary Routine
3-7/3-8
-------
BSR 3027
This portion of Program
entered when an Interrupt
occuri due to Teletype Input,
Punch, Reader, Teletype Output,
or Chopper Reference Signal.
Save AC
and UNK
Yes
I No
Initiate
Analog-to-DigiUl
Conversion
Teletype
"Output Inte r rupt f*
^
iegrator |
RDNC » A/D Covertlon
Result
ISFIN-
Re (tore I
AC and
LINK
Enable
Interrupts
Return Control
to Interrupted
Program
Main Computation
I SUM • SUM + RUNG + OFFSET!
00
00
n
to
Yes,
| ADOF = 1 |
^ *
No
Figure 3-3 Interrupt Service Routine
(Page 1 of 5)
3-9
-------
BSR 3027
Yes j
1 ADOF =
|
'1
V
No
| RDIX = RDIX + 1 |
CO
00
m
CD
RDIX = 0?
Yes
RDIX = -NOCPC
ICHRDL(CHAN)=
SUM/EAGC
No
DISP
Yes
IFMHLD = i
No
[SUM = 0 I
CTBP = CHAN I
CHNCNT = CHNCNT + 1
11
Yes [ | No
CRTF = 1
DETPF = DETF
FWDF = - FWDF
CHNCNT = - NOC
1 CNCOM =
FWHLD =
CNCOM + FWDF; 1
1 1
I
STAT
Figure 3-3 Interrupt Service Routine
(Page 2 of 5)
3-10
-------
BSR 3027
Yes
Y..I
ADERR
tput ACC to I
erface Unit |
Output ACC I
Inter
I
[ADOF » 0|
TEM1 = -NOG
J
Output RBL (TEM 1)
to CRT O/A
Converter
TEM1 = TEM1 + 1
14
TEM1 = 0?
Yei
I No
Output Zero
to CRT D/A
Converter
ITEMI = -NOE
Output DDSPB (TEM1)
to CRT D/A
Converter
TEM1 = TEM1 + 1
oo Yeil
I No
No
Figure 3-3 Interrupt Service Routine
(Page 3 of 5)
3-11
-------
BSR 3027
J?
STAT—«-l
DWD = Input From Interface Unit
SWWD = Input from Switch Register
DETF = Bit 2 of DWD
AC = B2 of SWWD
16
AC = 0?
Yea
No
I STOP = 0 I I STOP = 1 I
Yes
I No
DETPF = 0 1 PNCHG = 0 1
IIT = 0
1
Turn Off Alarm
Indicator Light
1
| AC = Bit 0 of SWWD]
s
Yes
is^X^s^
^ AC = 0? ^V^
No
| PNCHG = 0 | | PNCHG = 1 |
*
CO
ob
'AC = Input From
Interface Unit
AC = Bit 7 of AC
Figure 3-3 Interrupt Service Routine
(Page 4 of 5)
3-12
-------
BSR 3027
Yes
No
AC = Input from
Interface Unit
AC = Bit 7 of AC
20
Yes
No
AC = Bit 8 of DWD
Yes
No
(Multiply DDISI
by4 I
J.
Execute
FWCTL
Subroutine
AC = FWHLD + STOP * CRTF + CTBP + DETPF
22
AC = 0?
Yes
No
I MFLG =il I
MFLG = o
s
m
oo
ISFIN
Figure 3-3 Interrupt Service Routine
(Page 5 of 5)
3-13
-------
BSR 3027
Filter Wheel
Control Subroutine
IFWHLD = 0
REFDIS = | FWPOS - FWREF)
REFDIS = 0?
JL
[TBS - TBS + ij
TBS : 0?
AC = Input From
Interface Unit
AC = Bit 11 Of AC
O)
CO
FWCOM = FWPL (CNCOM)
UIST = FWCOM - FWPOS
TBS -1 I [TBS • D1ST . 4 |
.Mi =|FWPL (CHAN) - FWPOS|
= |Disr|
Error Subroutines
Halt
I No
txit
ADERR
Ve.,
Halt
I No
Exit
Figure 3-4 Subroutines
3-14
-------
BSR 3027
TABLE 3-1
GLOSSARY OF COMPUTER PROGRAM SYMBOLS
Variable Name
Acronym
Purpose
INITF
AC
SIMF
PNHLD
TEM 1
INIT
NOG
RBA Array
TEM 2
I. J
FWPL Array
LINK
PCNC
PCC
PWD
CRF
FWCC
MFLG
STOP
IDF
DPSR
CO DPSL
CO
("1
a> FWHLD
Initialize flag
Contents of the accumulator
Simulate .flag
Punch hold flag
Temporary storage location
Instruction location in INIT routine
Number of channels
Read buffer list
Temporary storage location
Dummy indices
Filter wheel position list
POP-8 hardware overflow bit
Punch channel counter
Punch character counter
Punch word
Carriage return flag
Filter wheel control counter
Main flag
Stop flag
ID flag
Double precision shift right
Double precision shift left
Filter wheel hold
One for initializing; zero otherwise
One for no halts due to instrument
errors; zero otherwise
One for punching operation in progress;
zero otherwise
List of desired filter wheel positions
Keeps track of punching progress
Keeps track of punching progress
Location containing data to be punched
next
One for carriage return next punch
operation
One means enter main computation;
zero if hold
One if operator wants to stop action
zero otherwise
One if ID is to be incremented
Address of routine which multiplies double
precision numbers by one half
Routine multiplies double precision num-
bers by two
One means filter wheel is not where
it belongs; zero otherwise
3-15
-------
BSR 3027
TABLE 3-1 (CONT. )
DWD
CHPCNT
FWREF
CM COM
CHAN
AGC
RDIX
NOCPC
SUM L
SUM H
CHNCNT
FWDF
FMHLD
CRTF
CHR
DETF
DETPF
CTBP
ADOF
TBS
AUX
°? PEX
oo
i~
m CHRDL
Discrete word
Chop count
Filter wheel ref position
Channel command
Channel
Automatic range control
Reading Index
Number of chops per channel
Sum Low
Sum High
Channel count
Filter wheel direction flag
Format hold
Channel reading transfer flag
Complement of highest reading
Detection flag
Detection processing flag
Channel to be processed next
A/D overflow
Time between (filter wheel) steps
Auxiliary routine
Punch exit location
Channel reading list
Instrument condition input from inter-
face unit
Incremented by one for every chop
Channel to .which filter.wheel should
be moved
Current channel number
Index of signal gain provided by system
Counts chops
Number of desired chops per channel
Least significant portion of integrator sum.
Most significant portion of integrator sum.
Keeps track of channel number
Plus or minus one; sense indicates
direction of next .filter wheel rotation
One if format overflow has occurred
One if the current spectral scan is
complete
Maximum reading obtained during
spectral scan
Assigned the value of Bit 2 of switch
register for operator request
Internal flag set to equal DETF when
regression sums are ready for evalu-
ation
Set to channel number when a reading
is ready for contribution to regres-
sion sum;.zero otherwise
Set if A/D conversion overflow occurs
Routine for punching operations
and detection processing
Values entered in main computa-
tion of interrupt service routine
3-16
-------
BSR 3027
TABLE 3-1 (CONT. )
PNCHC Punching
PNCF Punch ready flag
DTEM Detection processing temporary
storage location
NOE Number of equations
DETTF Detection test flag
INTC Low pass constant
MDT Minimum delta temp
RGS Array Regression sums
FRGS Fudged regression sum
LPRS Low passed regression sum
DDSPB (Array) Detection display buffer
DISP
ISR
STAT
OFFSET
SWWD
REFDIS
TBS
DIST
SPECNT
SPECND
CD
oo
r~
oo RNCC
Display
Interrupt service routine
Status
Offset
Switch word
Ref distance
Time between steps
Distance
Spectrum count
Range constant
One if operator has called for paper
tape data output
One if punch is in condition to receive
next characters
Keeps track of the regression equa-
tion being evaluated
Number of regression equations to
be evaluated
One if a complete regression sum
is available for testing
No contrast threshold
One value for each regression
equation evaluated
Alarm information for oscillo-
scope display
Location initiating CRT display
routine in ISR
Electrical offset correction
Operator instruction input from
switch register
How far the filter wheel is from
its reference position
Number of steps required to move
filter wheel to the next desired
position
Number of spectra' to be skipped
between outputs
Range between transmitter and
receiver in kilometers
3-17
-------
BSR 3027
TABLE 3-2
PROGRAM BRANCH EXPLANATIONS
INIT Routine
1. Is initial printout of filter wheel position to be skipped?
2. Has overflow occurred on the most recent addition?
3. Has reading of filter wheel position list been completed?
4. Has filter wheel reached the reference position?
5. Has a filter wheel position error occurred?
6. Is filter wheel properly positioned?
AUX Routine
1. Is a spectral scan completely punched?
2. Is a spectral scan in progress?
3. Can next character be punched?
4. Is punch output required?
5. Can punch output format accommodate readings?
6. Is spectral scan completely punched?
7. Should detection processing be skipped for the channel available?
8. Has the current channel reading been included in each regression
sum being evaluated?
f
9. Has the last channel been included in each regression sum?
10. Have all channels been included in each regression sum?
11. First value of RGS ia the-estimate of the signal level for the
spectrum; is there sufficient signal for valid detection processing ?
12. Does the current value of RGS exceed the alarm level?
13- Have all regression sums been tested?
14. Have all regression sums been reset to zero?
ISR Routine
1. Did the chopper reference signal cause this interrupt?
2. Did the interrupt occur because of a completed teletype output?
3. Do peripheral activities remain to be completed before a reading is
accepted?
3-18
-------
BSR 30Z7
TABLE 3-2 (CONT. )
4. Has initializing been completed?
5. la the filter wheel properly positioned?
6. Is this the largest reading so far?
7. Is the reading too small?
8. IB the reading too large?
9. Has the proper number of readings been summed?
10. Is the channel reading too large for punch output format?
11. Has a spectral scan been completed?
12. Did all readings during the last spectral scan fall within the
range permitted?
13. Is gain at minimum value?
14. Have all channel readings been displayed on the CRT?
15. Ha* alarm display been completed?
16. Has operator requested STOP action?
17. Has the operator instruction for detection processing been given?
(The negative of this question is asked. )
18. Has the operator instruction for data gathering been given?
(The negative of this question is asked.)
19. Are peripheral activities completed?
20. Has times four request been given?
Subroutines
FWCTJL Subroutines
1. Is the filter wheel at its reference position?
2. Has sufficient time elapsed since last step?
'3. Is the filter wheel position reference light signal on?
4. Is the filter wheel in the desired position?
5. Must the filter wheel be moved counter clockwise to reach the next
desired position?
6. Locate the smallest of three values.
7. Is the smallest value larger than three?
9 Filter Wheel and A/D Error Subroutines
eo
r*
0) 1. Should evidence of error of the type indicated be honored?
3-19
-------
BSR 3027
AUX examines various flags to determine if th'e user desires punched or
printed output and if a complete set of spectral readings is ready for outputting.
The first of these variables to loe e'xa-miried in the AUX routine is PNHLD. If a non-
zero value has been previbusly assig'ned to this flag, then an output via the punch is
immediately completed. If,this condition is not met, the routine inquires of the
CRTF, the channel reading transfer flag, whether a completed list of channels is
ready to be transferred to the output buffer list. If so, the transfer is effected.
Only after the processed channel reladirigs ha've been transferred is the question
asked whether the operator' desires an output of the spectrum. This question takes
the form of testing the' fla-g" PNCHG for a non-zero value. This flag will be non-
zero if the user has signified his choice of this option by pushing the PUNCH DATA
button. A similar flag is interrogated to ensure that the output can be accommo-
dated by the display format. Only after the data have passed all of these tests is
the flag PNHLD se't: to a ri6ti-z:erb Value" establishing the condition for punching on
the next pass through AUX. Upon punching the last ch'a'racter on the list, the pro-
gram resets PNHLD to zero. When the user calls for detection processing, this
output is automatically suppressed by assigning PNCHG equal to zero in ISR. In
this case, CRTF = 0 signals the program to transfer control to location PEX where
the detection processing algorithms are evaluated. In a similar manner, the first
part of AUX senses the PRINT DATA request if it has been entered at the console.
As mentioned previously, this r"esuits in the printing of a decimal representation
of the channel readings on the Teletype.
The rest of the routine is concerned with detection processing. Following
the location called PEX, each channel reading is multiplied by the appropriate co-
efficients for each of the agents to be detected arid the products are added to the
running regression sums. The value of DETF, the detection flag, will be non-
zero if the user has called for detection processing. When the regression sums are
complete, the ISR will assign this value to DETPF. If this1 flag is non-zero, the
detection processing thresholds are compared with the current regression sums
and the appropriate rriessa'ges are printed if they ar'e exceeded. The flag DETTF
is initially set to zero in the" INIT routine. After its first examination, its value is
set to one and control is pass~ed to DET 3 where the regres'sion sums are all set to
zero. At the second examination, its value is found to be non-zero and detection
processing continues. While the degression sums are always set to zero in INIT,
it is possible for the user to turn detection processing off during the computation of
sums; this operation reinitializes the values of DETPF and DETTF. Upon a sub-
sequent call for detection processing, the regression sums will be reinitialized
without entering INIT before the results are accepte'd for evaluation.
Each new regression sum becomes part of a weighted average of recent
regression sums in order to accomplish the noise reduction obtained from smoothing
or filtering. The filter type used is a simple numerical exponential filter with a
smoothing constant determined by the quantity INTC.
3-20
-------
BSR 3027
Measurement outputs on the Teletype occur at intervals determined by the
constant SPECNO. The interval will be the time for each spectrum multiplied by
SPECNO.
3.1.3 Interrupt Service Routine (ISR)
All other routine activities, with the exception of the subroutines, are per-
formed by ISR (see Figure 3-3). These remaining routine activities are necessarily
performed at the pace dictated by the chopper. At the completion of every chop
cycle, any other program is interrupted and control is passed to ISR. The address
of the next instruction in the interrupted program and the contents of the accumulator
and link are stored while ISR activities are completed; control is then returned to
the proper location in the interrupted program and the accumulator and link bits are
restored. A similar interrupt occurs when the Teletype has completed a character
output operation. This interrupt operation is very short, and ISR quickly establishes
the nature of the interrupt and control is returned almost immediately to the inter-
rupted program.
The ISR is prepared to handle five kinds of interrupts. An interrupt which
passes control to this program occurs whenever: (1) the chopper generates one,
(2) the punch completes the output of a character, (3) the Teletype completes the
output of a character, (4) the punch completes an input operation, or (5) the Teletype
completes an input operation.
A chopper interrupt immediately causes the integrator to be read, A/D con-
verted, and reset. Then, in accordance with the setting of program control flags,
the program is required to: (1) add the output of the A/D converter to the running
integration sum, or (2) transfer control to the filter wheel control routine, or
(3) return to INIT 1 to complete initialization, or (4) jump to the display routines.
The oscilloscope display of the most recently gathered spectrum is re-
generated at the chopper interrupt rate of 480 Hz.
The last part of the main interrupt service routine checks on the various
operator options selected and reviews the general status of the instrument control
processing activities. The flag MFLG is then set to the appropriate state for the
next chopper interrupt, and control is returned to the interrupted program. Just
prior to this return, the contents of the link and the accumulator are restored to the
values they held before the interrupt.
If the interrupt arose from an I/O flag, the program transfers control to the
appropriate device selectors for input-output operations. Handling of I/O operations
on an interrupt basis is not dictated by the logic of the program; it is a hardware
3-21
-------
BSR 3027
convenience. As such, I/O operations will cause interrupts when the interrupts are
enabled. Alternatively, upon receiving an I/O interrupt, the program could be easily
directed to return control immediately instead of halting. In this case, however,
accidental inputs might alter the operation of the program without warning.
3.1.4 Subroutines
The only major subroutine used in the Ozone Monitor software which is not
a part of the standard PDP-8/L programs is FWCTL ('see Figure 3-4), the filter
wheel control subroutine. This routine is called by ISR when the flag FWHL.D is
found to be non-zero and the main computation has been bypassed (MFLG = 0).
FWCTL is responsible for checking to verify that the filter wheel is stepped to those
positions which appear on the filter wheel list. It also looks for the reference pulse
from the interface1 unit which signifies that the wheel is properly referencing its
positions. If this latter condition is not met, the error subroutine FWERR is sum-
moned to halt action unless the operation has called for a return to INIT by lifting
switch No. 10. The other error subroutine is ADERR, and it is called if the output
from the A/D converter is too large for data processing; it too can be suppressed at
the discretion of the operator.
3. 2 DESCRIPTION OF OUTPUT
When the PRINT DATA button is depressed, the computer will output on the
Teletype the channel readings for the most recently gathered spectrum. These
numbers are in decimal form. The system will stop gathering spectra while this
complete spectrum is being printed.
When the PUNCH DATA button is depressed, the spectrum will be output on
the high-speed punch. This punches at a rate of 50 characters per second which
is five times faster than the Teletype. The tapes punched by this device are in
hexadecimal form. Hexadecimal numbers contain more information in four output
digits than do decimal numbers. The spectra on these tapes can be subjected to
statistical analysis on a large computer to determine the best set of channels and
operating coefficients for use in detecting pollutants in the"presence of spectral
interferences. .Further detail on these statistical programs can be found in the
referenced document* and in Section 4 of this report.
#
Bendix Aerospace Systems Division, "Computer Analysis of Spectral Data, " by
Robert H. Dye and Arnold Prostak, Final Report BSR 2830, Volume 2, Con-
tract No. DAAA15-68-C-0521, December 1969.
3-22
-------
BSR 3027
SECTION 4
OFF-LINE DATA PROCESSING
An IBM-360/50 computer was programmed in FORTRAN IV to handle the
large quantities of spectral data obtained on punched paper tape.
A program called HEXED (for hexadecimal editor) converts the data on paper
tape, which are in the hexadecimal number system, to the decimal number system.
The data are in hexadecimal because they transmit more information per punch
character. The HEXED program also computes a checksum for each spectrum and
compares it with a checksum punched on the tape to reject spectra containing errors
due to tape punch or tape reader malfunction. Figure 4-1 shows an example of
HEXED printed output.
The OMAS (for Ozone Monitor Absorption Simulator) Program computes the
transmittance that would be in each spectral channel if there had been an amount
of ozone in the beam giving a specified peak fractional absorptance. This computed
transmittance takes into account the spectral bandwidth of the Ozone Monitor and is
needed to simulate ozone-containing spectra. Figure 4-2 shows an example of OMAS
printed output.
The stepwise regression program computes a sequence of multiple linear re-
gression equations in a stepwise manner. At each step, one variable is added to the
regression equation. The variable added is the one which makes the greatest re-
duction in the error sum of squares. Equivalently, it is the variable which has the
highest partial correlation with the dependent variable partialed on the variables
which have already been added; and equivalently it is the variable which, if it were
added, would have the highest F value. In addition, variables can be forced into the
regression equation and automatically removed when their F values become too low.
In addition, after all the significant variables are in the regression equation, this
program deletes one variable at each step. The variable deleted is the one with the
smallest F value. Regression equations with or without the regression intercept
may be selected.
4-1
-------
8378-15
ERrtUX LtGFND
IF=0
Lf=l
CK = 0
th = l
0 OR
, 1 INE flf.O UKAY
. LINE l-EEn IN ERKOR
, CARRIACt RETURN OKAY
, CARKIA&E RETURN IN FRkOR
1 PKINTtO UNDFB VA^IABLEt WHbRE, 0 INDICATfcS NO EK*HR, 1 INDICATES
ERROR.
MITE* POSITION L 1ST
9
101
16?
22Z
30*
CASE NJ.
-0.6875
155.2500
140.8125
64.C625
42.1875
CASE NO.
4.5625
142.0625
124.6250
80.0000
32.1875
CASE NU.
-0.6J50
151.1250
136.0000
70.6250
39.6250
CASE MO.
2.1250
1J8.3750
138.0625
86.7500
41.5000
CASE NU.
-0.9375
150.4375
123.0000
84.9375
42.0625
CASE NO.
7.1250
138.1250
128.4375
73.3125
33.6875
CASE NO.
-4.1S75
148.6250
130.6875
73.9375
41.9375
CARD READ:
CARD READ:
CARD READ:
CARD READ:
11
110
168
230
311
1 ID NO. 50
5.25CO
145.4375
132.8750
101.000C
29.0625
2 ID NO. 50
3.5000
133.6875
125.5625-
100.0000
31.7500
t ID NO. 50
4.4375
134.9375
137.6675
97.3750
33.8125
4 ID NO. 50
4.1250
135.1250
138.1875
90.5t25
26.0625
5 II) NO. SO
4.4375
139.8125
119.3750
B9.5000
31.9375
6 10 NO. 50
3.3125
119.3750
124.3125
101.3125
33.0000
7 ID NO. 50
3.6250
136.7500
124.6875
92.75UO
30.0000
OOOB 0001
070E C745
059C 05 2B
0062 C02D
29
118
173
23B
322
1.0625
1J2.8125
U5.500C
10>.lb75
22.8125
B.OOOO
124.3125
117.3750
93.750C
23.5000
-2.0625
115.5625
140.4375
97.4375
21.6250
1.0625
115.9375
129.5000
90.7500
24.2500 •
• 8.0625
119.6250
116.0000
85.9375
23.3750
6.1875
126.1250-
121.5625
84.3750
18.6875
4.4375
110.6250
117.4375
95.8750
22.9375
C069
0821
0510
0032
39
124
179
246
330
21.5625
125.8125
134.4375
89.4375
24.3750
14.6875
11 3.1250
U0.875U
H6.6250
26.1375
17.5625
1M. 0525
134.5000
87.8750
27.0000
16.6875
124.2500
128.4375
80.9375
23.0000
16.2500
111.0625
130.1250
S3. 1250
27.5625
14.5625
115.2500
138.3125
84.0000
36.1875
18.0625
.. 113..7530
133.1250
85.1250
26.0000
OOC7 0240
C799 0813
0439 0464
F088 :
48
13l>
184
254
339
42.3750
109.6750
128.4375
84.4375
21.9375
29.1d75
114.ir)75
132.2500
76.i375
21.3125
42.8125
1 12.3125
134.375U
73.3750
27.S875
29.7500
115.7500
129.0625
79.3750
22.2500
33.8125
112.1253
130.3750
74.7530
20.9375
33.6250
108.1253
142.1875
74.S250
20.5625
40.9375
102.8750
140.1250
73.3000
23.J375
03DC 0618
0786 U826
0301 0354
57
135
189
262
344
06.1250
131.0625
126.3125
75.0000
24.0000
61.8750
123.12*0
135.7500
74.3125
23.7530
64.0625
125.3750
131.0030
69.0625
22.1875
60.0625
115.6875
121.8750
75.6S75
17.8125
54.6250
127.5000
123.8125
77.5000
23.1875
42.1250
120.9375
129.5625
70.0000
25.5625
71.2500
120.3750
137.0625
74.7500
26.6875
08C3
07BF
030F
66
141
195
271
358
96.
125.
121.
5b.
19.
105.
132.
123.
61.
19.
110.
122.
122.
64.
17.
110.
133.
116.
58.
13.
110.
123.
109.
2500
9375
6875
3125
6875
1250
5625
6875
S375
2500
4375
6875
8125
5625
9375
3125
8125
1875
8125
9375
0000
5000
8125
56.1250
19.
98.
129.
125.
63.
24.
115.
134.
124.
60.
16.
096A
08BF
0202
5000
9375
4375
5625
0625
3750
6875
875U
0625
3125
8750
08BE
08HF
01 DC
75 64
146 151
200 205
279 287
367 377
147.0625
125.3750
125.0625
59.6875
12.0625
151.3125
121.8125
131.4375
58.9375
8.7500
135.0625
117.3125
129.5625
58.7500
11.8125
129.5625
130.2500
126.0625
61.8125
12.9375
136.3750
124.3125
112.8750
53.6250
13.0000
152.0000
114.5625
119.0000
62.9375
.8.0625
141.2500
132.1250
120.3125
65.5625
16.0000
OB7E 07AC
077E 07A4
0157 0156
93
157
213
296
367
148.5625
122.3750
109.0625
56.3125
4.1875
143.6250
123.2500
121.1250
60.5000
2.3750
150.6875
129.6875
125.0000
52.1875
5.6250
152.0625
124.1875
117.3125
52.9375
8.8125
147.3125
123.6875
120.6250
55.9375
6.6250
146.9375
112.3750
121.3750
55.5000
6.1875
159.2500
125.2500
104.1250
6O.5000
1.7500
07DE 072C
081E 0654
01C6 0170
144.7500
127.6875
96.5625
50.5625
5.8125
146.8125
132.6875
98.0625
48.1250
5.50OO
131.6250
138.7500
105.3125
53.6875
7.5625
133.3125
149.2500
102.5OOO
51.1875
9.8125
135.5000
121.75OO
95.4375
48.8750
4.0000
140.5625
136.3125
102.1250
49.8750
6.3125
125.0625
124.8125
90.8750
50.1875
6.0000
06F2
05C2
0175
O7C9
O5E1
O08B
w
en
o
ro
-o
A8UVE SET WAS FOUND UNACCEPTABLE. SUM (
64913) DOES NOT AGREE WITH CHECKSUM (
64904).
Figure 4- 1 Output From HEXED Program
-------
8378-16
PEAK ABSORPTION -15.0 PERCENT
CLEAR-TO-AGENT CASE RATIO '
1 BW> 0.14 MICRONS
OZONE
CL* 1*0.13989 M1CROGRAH-KILONETERS PER CUBIC METER
I MAVELENGTH ALPHA PCNT ABSORPTION TRANSMISSION
CHAN FUPOS WEIGHTED TRANSMISSION
1
2
3
4
5
6
^
8
9
10
11
12
13
1*
15
16
17
18
19
20
21
22
23
2*
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
4b
47
40
49
50
51
52
K, ?
7.50
7.55
7.60
7.6S
7.70
7.7S
7.80
7.85
7.90
7.95
8.00
8.05
8.10
8.15
8.20
8.2S
8.30
8.35
8.40
8.45
8.50
B. 55
8.60
8.65
8.70
B.7S
8.80
8.8S
8.90
8.95
9.00
9.05
9.10
9.15
9.20
9.25
9.30
9.35
9.40
9.45
9.50
9.55
9.60
9.65
9.70
9.75
9.80
9.85
9.90
9.95
10.00
10.05
10.10
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.000
0.000
0.000
0.000
0.009
0.012
0.016
0.017
0.024
0.029
0.048
0.059
0.062
0.044
0.030
0.029
0.033
0.041
0.044
0.046
0.056
0.302
2.102
4.301
4.778
2.322
1.779
2.653
2.853
2.6BO
2.448
1.920
1.314
0.918
0.572
0.373
0.190
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
o.ooo
0.000
0.000
0.001
0.030
0.042
0.053
0.058
0.082
0.097
0.164
0.199
0.210
0.1SO
0.102
0. 100
0.113
0.140
0.149
0.157
0.189
1.023
6.900
13.610
15.000
7.594
5.872
8.629
9.248
8.713
7.989
6.322
4.371
3.073
1.92&
1.260
0.645
1.00000
1.00000
1.00000
1.00000
1. 00000
1.00000
1.00000
i.ooooo
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
i.ooooo
1.00000
1.00000
i.ooooo
1.00000
0.99999
0.99970
0.99958
0.99947
0.99942
0.99918
0.99903
0.99836
0.99801
0.99790
0.99850
0.99898
O.99900
0.99887
0.99860
0.99851
0.99843
0.99811
0.98977
O.93100
0.863)0
O.dSOOO
O.92*06
0.94128
0.91371
0.90752
0.91297
0.92011
O.93673
U.95629
a.<»»927
0.96072
O.987*0
0.99355
10
11
12
13
14
15
16
17
18
19
20
21
22
23
2*
25
57
75
84
93
101
110
118
124
130
135
141
146
151
157
162
168
173
179
If.
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
0.99987
0.99950
0.99883
0.99824
0.99874
0.99877
0.99818
0.98445
0.92920
0.88972
0.91764
0.91550
0.92651
0.95261
0.97943
O.99160
-------
-jo
57
58
59
60
61
62
63
64
65
66
67
68
69
70
Tl
72
73
74
75
76
77
78
7V
80
81
82
83
84
85
86
67
88
89
90
91
92
•it
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
ft w. «..»
10.30
10.35
10.40
10.45
10.50
10.55
10.60
10.65
10.70
10.75
10.80
10.85
10.90
10.95
11.00
11.05
11.10
11.15
11. 20
11.25
11.30
11.35
11.40
11.45
11.50
11.55
11.60
11.65
11.70._.
11.75
11.80
11.85
11.90
11.95
12.00
12.05
12.10
12.15
12.20
12.25
12.30
12.35
12.40
12.45
12.50
12.55
12.60
12.65
12.70
12.75
12.80
12.85
12.90
12.95
13.00
13.05
13.10
13.15
13.20
13.25
13.30
13.35
13.40
13.45
13.50
13.55
13.60
.13.65
13.70
13.75
13.80
13.85
13.90
13.95
14.00
0.012
O.CC5
0.003
0.001
0.000
0.000
0.000
0.000
0.0
0.000
0.000
0.000
0.000
0.000
0.0
0.0
o.oou
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
o.o
0.0
0.0
0.0
0.0
0.0
0.0
0.0
o.o
0.0
0.0
0.0
0.0
0.0
0.000
0.000
0.001
0.002
0.004
0.006
0.007
0.011
0.009
0.013
0.013
0.022
0.019
0.027
0.022
0.027
0.038
0.027
0.053
0.030
0.051
0.046
0.032
0.085
0.037
0.046
0.092
0.047
0.101
0.079
0.102
0.126
0.103
0.140
0.066
0.042
0.017
O.C09
0.003
0.001
0.000
0.000
0.000
0.0
0.000
0.000
0.000
0.000
o.occ
0.0
0.0
o.oou
0.0
0. 0
0.0
0. 0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
o.o •
0.000
0.000
0.003
0.005
0.013
0.022
0.024
0.036
0.030
0.043
0.046
0.074
0.063
0.092
0.074
0.092
0.130
0.093
0. 181
0.102
0.175
0.156
0.109
0.290
0.126
0.155
0.313
0.161
0.342
0.268
0.347
0.427
0.350
0.476
0.224
0.99956
O.999B3
0. 99991
0.99997
0.99999
1.00000
1.00000
1.03000
l.OJOOO
1.00000
1.00000
1.00000
1.00000
1.00000
1.00003
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00030
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
0.99997
0.99995
0.999b7
0.99978
0.99976
0.99964
0.99970
0.99957
0.99954
0.99926
0.99937
0.99908
0.99926
0.99908
0.99870
0.99907
0.99819' '
0.99898
0.99825
0.99844
0.99891
0.99710
0.99874
0.99845
0.99687
0.99839
0.99658
0.99732
0.99653
0.99573
0.99650
0.99524
0.99776
27
28
29
10
31
32
33
3*
35
36
37
38
39
40
41
42
<,;
4<
4'
4
4
4
4
i
46
47
48
49
50
195
200
205
213
233
238
246
25*
262
271
279
287
296
304
313
322
330
339
349
358
367
377
387
o. wit*
O.99906
O.99997
1.00000
l.OOOOU
1.00000
l.OOOOO
l.OOOOO
1.00000
l.OOOOO
1.00000
1.00000
1.00000
1.00000
1.00000
0.99992
0.99973
0.99958
0.99929
O.99900
0.99865
0.99845
0.99813
0.99749
W
o
Is)
-J
Figure 4-2 Output From DMAS Program
-------
BSR 3027
The output from this program at each step includes:
1. Multiple R
2. Standard error estimate
3. Analysis-of-variance table
4. For variables in the equation
a. Regression coefficient
b. Standard error
c. F to remove
5. For variables not in the equation
a. Tolerance
b. Partial correlation coefficient
c. F to enter.
From the regression results, one can select a step which gives a suitably low
standard error estimate and use the selected spectral channels and coefficients
from that step.
Figure 4-3 shows an example of the printed output of the regression program.
The coefficient converter program converts the selected coefficients into
double-precision octal form for insertion via punched paper tape into the PDP-8/L
computer. Figure 4-4 shows the printed output of the coefficient converter program.
Another program was used to compute the mean and standard deviations of
each spectral channel in a group of spectra having the same identification number.
Plotting programs were used to plot individual spectra and averaged spectra on a
Calcomp plotter. Some of these machine-plotted spectra are shown elsewhere in
this report.
4-5
-------
8378-17
STEPHISE REGRESSION - VERSION OF SEPTEMBER 68
PROBLEM CODE OZONE
NUMBER OF CASES 47C
NUMBER OF ORIGINAL VARIABLES 51
NUMBER OF VARIABLES ADDED 0
TOTAL NUMBER OF VARIABLES 51
NUMBER OF SUB-PROBLEMS I
INPUT DATA
1
1
I
1
1
-I
1
1
1
2
2
2
2
2
2
2
2
2
.3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
6
6
rO. 00826
1.15635
1.59561
1.47021
1.54305
0.76965
0.70057.
0.27407
0.05031
-0.00840
1.17554
1.61924
1.32979
1.55549
0.78242
0.71220
0.27854
0.05105
0.05544
1.27744
1.51059
1.49768
1.60705
0.97213
0.75264
0.28556
'0.02886
.0.05634
1.29824
. 1.53249
•l**2M2
1.61951
0.98796
0.76490
0.29014
0.02928
-0.00751
1.32680
1.38837
1.55806-
1.61438
0.84849
0.77565
0.25980
0.06758
-0.00764
1.34940
0.06307
1.76681
1.51151
1.53'404
1.51752
1.21341
0.71709
0.29284
0.06983
0.06412
1.79614
1.53467
1.43106
1.53818
1.23355
C. 7 2 899
0.29758
0.07081
0.04253
1J83869
1.44756
1.61236
1. '6 49 58
1.21516
0.171618
'0.31822
0.06683
0.0 4'3 22
1.86864
1 .46929
1.50367
1.671;52
1.23495
0.72785
0.32327
0.06775
0.05331
1.62264
1.41840
1.66694
1.57383
1.16986
0.70582.,
0.32438
0.09086
0.05422
1.65029
0.01276
1.78483
1.32004
1.69172
1.46195
1.23969
0.67654
0.26356
0.0
0.01298
.81445
.34030
.57448
.48524
.26027
0.68777
0.26774
34.01398
0.09721
1. 74*52 7
1.39363
1.51439
1.50300
1.13921
0.73517
0.28)36
0.0
0.09880
1.77370
1.41459
'1.40901
1.52647
1.15777
0.74715
0.293:87
34.01398
40.02478
1.81036
1.35533
1.63390
1.47547
1.17061
0.62698
0.33264
0.0
-0.02520
1.84120
0.25905
1.73903
1.57458
1.59636
1.50Z50
1.36249
0.60746
0.28334
0.0
0.26335
1.76756
1 . 597 8 1
1.50360
1.52722
1.08012
0.61754
0.29283
3.0
0.17948
1.78401
1.49616
1.52578
1.59717
1.05263
0.58*83
0.28360
0.0
0.18138
1.81283
1.51777
1.43668
1.62296
1.06978
0.59432
0.29301
0.0
0.21100
1.58134
1.50625
1.65418
1.55656
1.05573
0.64503
0.26&5&
0.0
0.21459
1.60807
0.50909
1.86517
1.51301
1.50776
1.31028
1.01443
0.50684
0.23653
0.0
0.5175'4
1.895 IB
1.51420
1.46014
1.33199
1.03127
0.51525
0.24313
0.0
0.35467
1.72628
1.61084
1.42629
1.47186
0.93188
0.39113
0.23392
0.0
0.'36045
1.75352
1.61162
•1.38083
1.49579
0.94705
0.39750
0.23741
0.0
0.51435
1.81561
1.47397
1.68722
1.50175
0.94160
0.47605
0.21550
0.0
0.52311
1.84562
0.79443
1.74729
1.50625
1.61513
1.16010
0.90105
0.34 9 1'6
0. 14492
•o.o
0.80761
1.77421
1.42285
1 . 608 16
1.17936
0.91601
0.35492
0.14710
0.0
0.75188
1.62452
1.48022
1.59034
1.19162
O. 90 302
0.38581
0.10633
O.'O
0.76413
1.64904
1.39783
1.58299
1.21102
0.91772
0.39207
0.10789
0.0
0.76965
1.62114
1.40939
1.61588
1.26522
0.82972
0.40622
0.14192
0.0
0.78276
1.64683
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
'O.'O
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
'0:0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
50
5°
'50
50
50
50
50
50
'50
50
50
50
50
50
50
50
50
'50
50
'50
50
50
50
50
50
50
50
50
50
50
'50
50
50
50
'•50
:50
"50
'50
50
50
50
50
50
50
50
50
50
w
en
Figure 4-3 Output From Regression Program (Page 1 of 2)
-------
837817
STEP NUMBER
VARIABLE ENTRD
MULTIPLE R
STO. ERROR OF EST.
RESIDUAL SAMPLE RMS OEV.
ANALYSIS OF VAR IANCE
REGRESSION
RESIDUAL
OF
7
46)
7
.22
0.9729
9.3392
9.2694
SUM OF SQUARES
232079.613
40383.301
MEAN SQUARE
33154.258
87.221
F-RATIO
380.118
VARIABLE
VARIABLES IN EQUATION
COEFFICIENT STO. EKROR f TO REMUVF
(CONSTANT
12
16
19
21
22
26
29
C.O I
36.78831
31.85229
-87.62930
-34.11340
-38.44679
48.86093
45.51961
4.61565
5.89114
5.67714
6.78207
6.27148
5.64439
6.25128
63.526)
29.2137
2)9.2538
25.3003
37.5811
74.9358
53.0224
I
-0
VARIABLE
1
2
3
4
5
6
7
8
9
10
11
13
14
15
17
18
20
23
24
25
27
28
30
31
32
33
34
35
36
37
38
39
40
41
42
4)
44
45
46
47
4R
49
50
RUBLES NOT
IAL CORR.
-0.06186
-0.03771
-0.24840
0.07381
-0.06945
-0.06386
-0.00862
0.07398
0.13057
0.10068
0.03096
0.09630
0.08752
0.03881
0.09671
-0.11398
-0.14447
-0.01845
0.14857
0.07392
0.13770
0.10559
0.05977
0.04208
0.09043
0.13536
0.17141
0.04779
0.07590
0.07688
-0.01840
-0.00553
0.10208
-0.03994
-0.07584
-0.02345
0.09492
-0.08108
-0.04450
-0.09786
-0.05971
-0.08187
-0. 12206
IS EQUATION
TOLERANCE
0.90394706
0.49263632
0.73553079
0.18425030
0.10582834
0.02218916
0.01787509
0.00211882
0.00295757
0.00213170
0.00287401
0.00218262
0.00168840
0.00202292
0.00184999
0.00194075
0.0014804B
0.00155593
0.00157032
0.00161697
0.00271251
0.00165733
0.00361783
0.00640323
0.00217542
.0.00240979
0.00250685
0.00368678
0.00309978
0.00347955
0.00517048
0.00502958
0.00572900
0.00784267
0.00995739
0.02319283
0.01188842
0.02169086
0.02361567
0.02651780
0.08243269
0.32896096
0.45046C02
F TO ENTER
1.7748
0.6578
30.3822
2.5308
2.2392
1.8917
0.0343
2.5427
8.0124
4.7310
0.4433
4.3244
3.5663
0.6968
4.3619
6.0811
9.8476
0.1573
10.4276
2.5383
8.9298
5.2086
1.6563
0.8196
3.8092
8.6224
13.9854
1.0576
2.6772
2.7472
0.1565
0.0141
4.8650
0.7382
2.6730
0.2542
4.2001
3.0575
0.9167
4.4690
1.6531
3.1174
6.9878
w
pS
Figure 4-3 Output From Regression Program (Page 2 of 2)
-------
8378-18
NOC= 50
I
c»
1
2
3
4
5
6
7
8
9
10
11
12
13
14
IS
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
3*
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
4200/
NOE =
O.O2000C
0.020000
0.020000
0.020OOC
0.02000C
0.0200UC
O.U2UOUC
0.02000C
0.02000C
0.0200UC
0.02000C
0.02000C
0.02000C
0.02000C
O.O2000C
0.020COC
0.02000C
0.020COC
0.02UOOC
0.020UOC
O.O200CC
O.O2000C
0.02000C
0.02000C
0.02000C
0.020CCC
0.02000C
0.02000C
0.020000
U.020UOC
0.02000C
0.02000C
O.U2000C
O.U2000C
O.C2000C
0.020UOO
0.020COO
0.02000C
0.02000C
0.02000C
0.020UOC
0.020UCC
O.C20000
0.0200LC
0.0200CC
U.O20COC
0.020CCC
0.0200CC
0.02000C
0.0200CC.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
167772.
1677/2.
167772.
00507534
03537534
33537534
03507534
00537534
00537534
03507534
03507534
03537534
00507534
03537534
03537534
00537534
00507534
03507534
03537534
03507534
00507534
03537534
03507534
33507534
03507534
03537534
03537534
00537534
03537534
00507534
03537534
00537534
03537534
03537534
03507534
03537534
09537534
00507514
03507534
OJ50753*
03507534
03537534
03507534
03507534
00537534
00507534
U3S07534
03537534
03507534
03507534
00507534
03507534
03537534
1
2
3
4
s
6
7
if
•i
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
2t
30
31
32
33
34
35
36
3T
38
39
" 40
41
42
43
44
45
46
47
48
49
50
4400/
0
0
0
0
0
0
0
0
0
0
0
36
0
0
0
31
0
0
-d7
0
-34
-38
0
0
0
48
0
0
45
0
0
0
0
0
0
0
0.
0
0
0
0
0.
0
0
0,
0.
0
0
0
0
.0
.0
.0
.0
.0
.0
.o
.0
.0
.0
0
.78810C
.0
.0
.0
.6522BC
.0
.0
.629266
.0
.1133SS
.4462SS
.0
.0
.0
.86091t
.0
.0
.51<16Ce
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
7S3424.
0.
0.
0.
652335.
0.
0.
-1794647.
0.
-698642.
-787380.
0.
0.
0.
1000672.
0.
0.
932242.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
03030000
03000300
00000300
03000030
03000000
03030000
03033300
03000300
03030000
03000000
03000000
02677420
03000000
00000000
03000300
02372056
00000000
00000000
71116651
03000000
7J253356
74776114
00000000
00000000
03000300
03642337
03000000
00000000
03434621
03003000
00000000
00000000
00000000
00000000
00000000
oooooooo
00000000
00003000
03000000
00003000
03000000
00033000
oooooooo
03033300
03000000
03000000
00030000
0300UOOO
00003000
OOOOOOOO
td
en
o
rs>
Figure 4-4 Output From Coefficient Converter Program
-------
BSR 3027
SECTION 5
CALCULATION AND MEASUREMENT OF INSTRUMENT SIGNAL AND NOISE
5. 1 ACTIVE SIGNAL
The radiant power P reaching the main infrared detector when the filter
wheel is at a given wavelength (9. 5 |j.m, for example) is given by the following
formula for the system in operation with the remote active transmitter unit:
where
(5-1)
Lv = the spectral radiance of the source at 9- 5 |im. For a source tempera-
ture of 1265°C (1538°K), a blackbody 1^ is 0.092 W/(cm2-sr-(jim).
With an emissivity of about 0. 8, the L^ is 0. 073 W/(cm2-sr-|j,m).
A» = the spectral bandwidth of a small area of the filter wheel for collimated
rays. For the filter wheel in the ozone monitor, this bandwidth is
0. 085
At = the area of the transmitter optics. The transmitter primary mirror
diameter is 31.7 cm (12. 5 in. ) and the obscuring secondary mirror
diameter is 14 cm (5. 5 in. ). At = 635 cm .
Ar = the area of the receiver optics. The receiver primary mirror diameter
is 26. 6 cm (10. 5 in. ) and the secondary mirror diameter is 15. 0 cm
(5.94 in. ). Ar = 378 cm2.
R = the range between the transmitter and the receiver. It is nominally
1.61 km at maximum range.
T = the transmittance of the optical system including the atmospheric path
and the chopper. There are six mirrors, each with a reflectance of
perhaps 0. 92 (allowing for a little dirt). The beam splitter has a
transmittance of 0. 92. The chopper has an optical efficiency at the
fundamental chopping frequency of about 0. 39. The uncoated IRTRAN-2
window has a transmittance of about 0. 7. The transmittance of the
atmosphere is perhaps 0. 8. The transmittance of the filter wheel is
0.6 at 9.5 |jim.
5-1
-------
BS'R 3*027
T = 0.:926 X 0.92 X -0.39 x 0.7 X Q.-B X 0.6 = 0.075
P = 0.073 W X 0.085 tm X 65S < X x 0.075
com -sr-nm 1.61 X 10 cm
= 4. 3 X 10"9 W rms
This radiant .power falling on the photo'conductive infrared detector produces
a change in electrical conductance which with the constant bias voltage produces a
change in the electrical current through the detector. This current passes into the
preamplifier which acts as a tran'sresistance to -produce an output voltage V pro-
portional to the input current.
V = P X Resp. X E X R (5-2)
where
P = the chopped radiaaat power reaching the detector calculated, for
example, from Equation 5-1
Resp. = the infrared detector responsivity for 9. 5 jj. radiant power in
mhos/watt and was measured to be 45 x 10 mho/W for the
detector delivered with the Ozone Monitor
Ebias = t*ie ^^as voltage, which is set at 5. 8 V to maximize the S/N ratio
in the output of the detector currently installed
Rr = the transresistance of the preamplifier and is 66. 8 Mtt.
V = 4. 3 X 10"9 W X 45 x 10"3 ^^ x 5.8V x 66. 8 x 106 fi
= 0. 075 V
The output of the preamplifier is filtered, amplified, rectified, integrated, and
converted into a digital signal D.
D = V X 0.4 X 2k X ZjT/v X 1/480 X 1/(RG) X 410 (5-3)
5-2
-------
BSR 3027
where
0.4 = the insertion loss of the bandpass filter and associated
impedance matching voltage divider.
2 = the gain of the computer controlled variable gain amplifier,
where k is equal to the software variable "AGC"
2N/2/TT = conversion from the rms voltage of the AC signal to the
average value of the rectified DC signal
1/480 = the analog integration time in seconds
o
R = the input resistance to the integrator, which is 22. 1 X 10 fi
C = the feedback capacitance of the integrator, which is 0. 022
X 10-6 F
410 units/V = the scaling of the 12-bit A/D converter which covers an
input range of 10 V in 4096 units.
7.5 x 10"2 x 0.4 x 2k X 2V2/TT X 410 ._ , ,k
D - 7= 47. O X i \
480 X 22. 1 X 10 X 0. 022 X 10"
The Ozone Monitor computer program sums 64 such A/D converter readings,
amounting to a dwell time of 0. 133 sec for each channel. This sum is divided by
the gain that was introduced by the range control variable amplifier. For scaling
purposes, this number is further divided by 32 if the channel reading is to be out-
put on the Teletype, or by 2 if the readings are punched by the high-speed punch.
In the latter case, the off-line HEXED program that processes the high-speed
punch output divides by 16 so that the processed high-speed punch readings are
equal to the Teletype readings. Thus, the channel reading in Teletype units (T.t. u. )
is:
T.t.u. = D x 64 x l/2k x 1/32
T.t.u. = 47.6 X 2k X l/2k X 2 « 100
(5-4)
5-3
-------
BSR 3027
The actual signals at 1. 6 kin (1 mile) range at 9. 5-fim wavelength that were
measured in Ann Arbor varied from 80 to 125, depending on the visibility and
presumably on the accuracy of pointing the receiver and the transmitter units.
This is considered to be in good agreement with the calculated estimate of 100.
5. 2 PASSIVE SIGNAL
When the instrument is operated in the passive mode, the radiant power
reaching the main infrared detector when the filter wheel is at, say 9. 5 fim, is
given by:
P = L A J2 A T (5-5)
\ \ r
where
L> = the difference between the spectral radiance of a source which is
large enough to fill the instrument's FOV and the spectral radiance
of the /grooved internal chopper which is assumed to be close to the
local ambient temperature. If the chopper temperature is 295 °K and
the source has an effective radiation temperature 10° warmer or
305 °K, the spectral radiance difference at 9. 5 |im is 1. 68 X 10"4 W/
(sr-|jLm-cm ).
£2 = the solid angle of the instrument's FOV and equal to A^/(L 1. ) , where
A^ is the detector area and f. 1. is the receiver focal length.
0.785 x ID'2 cm2/402 cm2 = 4. 9 X 10'6 sr
T = 0. 924 x 0.92 x 0.39 x 0.7 x 0.6 = 0. 108
/
P = 1.68 X 10"4 - — -^ - X 0.085 IJL X 4. 9 X 10 sr x 378 cm2 X 0. 108
cm -sr-jim
= 2. 8 X 10~9 W rms
Using this power and Equations 5-2, 5-3, and 5-4, a signal of 65 units would be
predicted or 6. 5 units/degree. Measurement of hot and cold black-painted tanks
gave a responsivity of about 5 units/degree.
5-4
-------
BSR 3027
Before instrument delivery, instrument alignment was performed only to
maximize the signal from a 400-m-distant transmitter. Instrument performance
with extended sources could probably be improved somewhat by realigning and
optimizing the detector position for an extended area source.
5. 3 NOISE
The detector noise at the preamplifier output was 0.60 mV as measured with
a 6-Hz bandwidth centered at 480 Hz. The noise of the preamplifier itself, with
a 600-kfl resistor in place of the detector, is much less, 0. 07 mV. The noise
bandwidth, BW, of the Ozone Monitor is:
BW = -~ = r-4— = 3.8 Hz
2T 2 X 0. 133 sec
where T is the integration time per channel. The detector noise in this 3. 8-Hz
bandwidth would be:
0.60mV x 7^— = 0.48mVrms
If the output noise were entirely due to detector noise (which is the optimum
situation), the output noise would be given by Equations 5-3 and 5-4 operating on
this voltage. The result is 0. 64 Teletype units of noise. The actual measured
noise (standard deviation from the mean of the channel readings) with the beam
blocked is 1. 2 Teletype units. This includes all noise sources in the receiver
unit. A typical measured noise on a sunny day is 5 units which includes atmos-
pheric effects such as scintillation, pointing effects due to vibration and wind
loading, source fluctuations, and all other noise sources.
5-5
-------
BSR 3027
SECTION 6
RESULTS
6. 1 INSTRUMENT RESPONSE
Figure 6-1 shows the signal spectrum received from the transmitter when the
transmitter was about a meter away from the receiver. This spectrum at essentially
zero range is the response curve of the instrument and includes the source spec-
trum, the varying filter wheel transmittance with wheel position, and the infrared
detector response, as well as atmospheric absorption in the meter or two of air in
the optical path. The beam was attenuated with a pierced sheet of aluminum foil so
as to reduce the optical signal to normal levels. The spectrum shown is the average
of 39 consecutive spectra.
6. 2 ACTIVE-MODE SPECTRA
Figure 6-2 shows, on the left, average single-beam spectra as obtained by the
Ozone Monitor. About 30 to 60 spectra are included in each average. The.path
length varies from 0. 42 to 1. 67 km. On the right are shown the corresponding
normalized spectra obtained by dividing the single-beam spectra by the instrument
response curve shown in Figure 6-1 and arbitrarily normalizing the peak to 100.
Each individual spectrum was gathered in about 8. 5 sec, which includes about
2 sec for moving the filter wheel and a dwell time of 0. 133 sec at each of 50 channels.
Figure 6-3 shows two examples of digital listings of typical mean spectra.
Figure 6-4 shows a spectrum taken with a sheet of 30(j.m polystyrene film in
a 420-m-long beam. Also shown for comparison are the wavelengths where poly-
styrene has absorption bands.
Figure 6-5 shows spectra with a 1-m-long gas cell in the 1. 67-km path length
beam. The gas cell diameter was larger than that of the Ozone Monitor optical
elements and had polyethylene film windows. Ozone was introduced into the cell from
an ozone generator. In this figure, the spectrum peak has been arbitrarily normal-
ized to 75 to suggest the attenuation caused by the two polyethylene windows. The
ozone measurements shown were made using a spectrometric potassium iodide
oxidation method and were made by Dr. Harold M. Barnes, Jr. , of the National
6-1
-------
BSR 3027
o
o
flVG DflTfl
ID= 87
S °7TOO
8.00
n
a
9.00 10.00 11.00 12.00
WflVELENGTH (MICRON).
13.00
1.00
15.00
Figure 6-1 Response Curve of Instrument with Transmitter
6-2
-------
8378-21
flVC OflTR
10= 32
NORMPLIZEO DPTfl
10= 32
22 JUN, 2150 EST. CtfAR
420 m. ANN ARBOR
.00 8.00
9.00 10.00 11.00 IZ.OO 13.00
MflVELENCTH (MICRON)
IV.00 1 .00
w
CO
OJ
o
g
'S'.oo B'.OO
9 00 10.00 I .00 12.00
WAVELENGTH (MICRON)
13.00 IV.00
Figure 6-2 Spectra Obtained With Ozone Monitor
(Page 1 of 10)
-------
8378-21
HVG DRlfl
10= 50
NORMALIZED QfiTR
10- 50
°J.OO 8.00
9.00 10.00 11.00 12.00
MRVELENGTM (MICRON)
6 JULY. 1312 EST
.1670m. .ANN ARBOR
BRIGHT, SUNNY. CLEAR
13.00 '.4.00 15.00
td
o
IN)
9.00 10.00 11.00 12.00 13.00 IV.00
MPVELENGTH (MICRON)
Figure 6-2 Spectra Obtained With Ozone Monitor
(Page 2 of 10)
-------
i
Ul
8378-21
fiVG ORTP
10= 51
§.
9 00 10.00 1 .00 12.00
MflVELENGTH (MICRON)
NORMRLIZEO ORTR
10= 51
6 JULY. 1608 EST
1670 m, ANN ARBOR
13.00 11.00 15.00
w
01
co
o
to
f ooa. oo 900 10.00 I'I.OQ 12.00 13.00 Hi.oo
WAVELENGTH (MICRON)
Figure 6-2 Spectra Obtained With Ozone Monitor
(Page 3 of 10)
-------
8378-21
RVG OflTfi
ID= 52
NORMRL1ZEO ORTH
10= 52
6 JULY. 2000 EST
1670 m. ANN ARBOR
r oo a.oo
9.00 lOlOO 11.00 12.00
HflVELENGTH (MICRON)
13.00 IU.OO 15.00
td
C/5
o
ro
-o
°7 00 8.00
9.00 10.00 11.00 12.00
HRVELENGTH (MICRON)
13.00 14.00
Figure 6-2 Spectra Obtained With Ozone Monitor
(Page 4 of 10)
-------
8378-21
PVG ORTfl
10= S3
NORHPLIZED ORTfl
10= 53
6 JULY. 2330 EST
1670 m. ANN ARBOR
t'.OO 9.00 10.00 l' .00 12.00 13.00 lil.OOIS.00
MPVELENCTH (MICRON)
W
cn
°7Voo*e'.oo
9.00 10.00 11.00 12.00 13.00 lil.OO
HflVELENGTH (MICRON)
Figure 6-2 Spectra Obtained With Ozone Monitor
(Page 5 of 10)
-------
8378-21
I
oo
RVG OfiTH
10= Si)
NORMALIZED OflTR
10= Stt
7 JULY. 920 EST
1670 m, ANN ARBOR
8.00 S.OO 10.00 11.00 12.00 13.00 IV.00 IS.00
HRVELENGTH (MICRON)
en
°?.00 8.00
9 00 10.00 11.00 12.00
MHVELENGTH (MICRON)
13.00 IV.00
Figure 6-2 Spectra Obtained With Ozone Monitor
(Page 6 of 10)
-------
8378-21
AVG OflTR
10= 60
NORMRLIZEO DRTH
10= 60
7 JULY, 2200 EST
1670 m. ANN ARBOR
I
NO
°?.00 8.00 9.00 10.00 11.DO 12.00 13.00 IV.OO IS.00
MflVELENGTH (MICRON)
W
CO
50
w
o
to
9'.00 10.00 I'l.OO 12.00 i'3.00 ill. 00
HPVELENGTH (MICRON)
Figure 6-2 Spectra Obtained With Ozone Monitor
(Page 7 of 10)
-------
8378-21
I
t—'
o
RVG DflTfl
ID= 61
NORMflLIZED DflTR
10= 61
9JULY. WOEST
1670 m, ANN ARBOR
OVERCAST. HAZY. HUMID
9.00 10.00 11.00 12.00
WfWELENGTH (MICRON)
13.00 14.00 IS.00
td
o
ro
7 oo e.oo
9.00 10.00 11.00 12.00
WAVELENGTH (MICRON)
13.00 114.00
Figure 6-2 Spectra Obtained With Ozone Monitor
(Page 8 of 10)
-------
8378-21
RVG DPTO
10= 77
NORMALIZED DflTfl
10= 77
9 JULY, 1700 EST
1670 m, ANN ARBOR
CLOUDY. AFTER RAIN
.00 a.oo
l'.00 10.00 I .00 12.00 IS.00
URVELENCTH (MICRON)
11.00 IS.00
7 00 8.
V
W
05
50
OJ
o
tVI
00 10.00 11.00 12.00 13.00 IU.OO
MRVELENGTH (MICRON)
Figure 6-2 Spectra Obtained With Ozone Monitor
(Page 9 of 10)
-------
8378-21
flVG OflTR
10= 8W
g
c
c
n
-J
er
"8.
s
7 oo a.oo
9 oo 10.00 11.00 li.ooli.oou.oo
HRVELENGTH (MICRON)
NORMRLIZED OflTR .
8H
15 JULY, 1330 EST
1080m. CINCINNATI
ts.oo
td
OJ
o
9.00 10.00 11.00 12.00 13.00 14.00
URVELENGTH (MICRON)
Figure 6-2 Spectra Obtained With Ozone Monitor
(Page 10 of 10)
-------
8378-22
NUMBER OF CHANNELS:
NUMBER OF CASES:
10 NUMBER: 60
50
CHANNEL
NUMBER
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Id
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
4?
44
45
46
47
48
49
50
MEAN
-0.
3.
1.
4.
16.
43.
81.
138.
143.
135.
137.
139.
122.
118.
110.
124.
121.
118.
117.
133.
133.
136.
126.
140.
141.
136.
117.
130.
122.
117.
101.
101.
92.
84.
71.
71.
59.
54.
50.
44.
38.
29.
19
26
17
19
17
8
4
4
07127
36952
99890
50110
78947
24671
80373
14912
28618
61732
38816
E8487
41776
395E3
75768
603C7
33224
72917
87939
9C899
15351
154tl
51316
9C899
65461
74561
89563
89035
87C61
578S5
,30921
046C5
,09320
92873
86732
73794
,00548
17544
,36842
99452
,7t535
,6589V
.250CO
.98355
.28C70
.58443
.5011C
.33333
.65570
.63377
SID OEV
2.93848
2.93682
3.02428
3.65923
3.24084
3.26397
4.16710
4.54918
4.6C900
3.34667
4.66624
4.95977
4. 31493
4.23038
4.16713
4.08316
3.44872
3.71371
3.65566
4.04578
3.74710
3.90491
3.92982
4.46961
4.96258
4.60327
3.89141
4.34112
4.34677
3.90378
3.71298
4.31325
3.96050
3.26750
3.50341
3.36028
3.41895
3.4d303
3.34503
2.71280
2.67C27
3.27970
2.33274
3.C9710
2.tll26
3.40476
2.91609
3.15988
3.21207
2.97200
NUMBER OF CHANNUb:
MJMBEK OF CASES:
ID NUMBER: 79
50
25
CHANNEL
NUMBER
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
IB
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
3d
39
40
41
42
43
44
45
46
47
46
49
50
MEAN
-0.69250
3.74750
1.83750
1.560CO
11.28750
15.97CCO
34.23250
54.55500
59.15250
55.07750
56.85500
51.17000
45.24750
40.555CO
38.215CO
43.99500
41.70250
27.07COO
17.515CO
30.63250
25.81250
26.98000
36.10750
48.17250
56.97250
57.04750
53.44250
54.80750
52.86000
47.050CO
35.73750
43.23750
41.075CO
35.31750
31.40000
30.020CO
25.36750
24.70250
23.33000
19.89250
14.75500
11.31000
8.275CO
10.275CO
7.49000
8.47250
7.37500
3.72750
2.455CO
1.075CO
STO OEV
1.71694
2.53353
2.57138
2.15870
2.90743
2.51047
2.31535
2.81 703
2.24486
2.41767
2.70065
2.39978
2.04502
2.30696
2.10531
2.42404
2.57936
1.82131
1.81764
2.83454
2.34555
1.98437
2.38986
1.89368
2.52301
2.46785
1.81124
1.79090
2.05847
2.27375
4.38455
3.01450
1.52571
1.88857
2.90182
1.60C85
2.14310
2.65222
2.91690
2.03745
2.36214
2.54001
2.36125
1.76804
2.17688
3.01197
2.16333
2.44070
2.12305
1.43852
td
CO
Co
O
-------
8378-25
RVG DPUfl
ID= 30
NORMflLlZED DRTft
10= 30
A
POtYSTYRENE BANDS
"7 00 8.00 9.00
10.00 11.00 12.00
HRVEIENGTH (MICRON)
13.00 14.00 15.00
7.00 a.oo
9.00 10.00 11,00 1J.OO
WRVELENGTH (MICflON)
13.00 1U.00
cn
00
O
ts»
Figure 6-4 Polystyrene Spectra
-------
8378-20
flVG OATH
.10= 55
NORMflLIZEO DRTfl
10= 55
7 JULY, 1120 EST
0 OZONE
i
i—"
in
7.00 B.OO
9.00
\
10.00 11.00 12.00
URVELENGTH (MICRON)
13.00 11.00
15.00
td
en
O
tv
7 oo a. oo
'.00 10.00 1 .00 12.00
MRVELENGTH (MICRON)
i oo
TM.OO
Figure 6-5 Spectra with Ozone Cell in Beam (Page 1 of 7)
-------
8378-20
RVG DfiTfl
J0= 67
NORMflLIZEO ORTfl
10= 67
9 JULY. WISEST
0 OZONE
8.00
9.00
10.00 11.00 12.00
HflVELENGTH (MICRON)
13.00
14.00
*
cn
IS.00
°7.00
8.00
00 10.00 11.00 12.00
HRVELENCTH (MICRON)
W
CO
OJ
o
13.00 .14.00
Figure 6-5 Spectra with Ozone Cell in Beam (Page 2 of 7)
-------
8378-20
I
I—«
-J
flVG DflTfi
10= 78
.00 8.00
NORMPLIZEO DRTfi
ID= 78
9 JULY. 1715 E5T
17ngta?K 1600 m OZONE
9.00 10.00 11.00 12.00
WAVELENGTH (MICRON)
13.00 14.00 15.00
W
CO
o
M
oo a.oo
9.00 10.00 11.00 12.00
HRVELENGTH (MICRON)
13.00 11.00
Figure 6-5 Spectra with Ozone Cell in Beam (Page 3 of 7)
-------
8378-20
RVG DPTft
ID= 75
NORMALIZED DRTfl
ID= 75
9 JULY. 1625 EST
110/i g/m3x 1600m OZONE
00
>0.00 11.00 12.00
HflVELENGTH (MICRON)
15.00
W
cn
o
ro
8.00
9.00 10.00 11.00 12.00
HflVELENGTH (MICRON)
13.00
111.00
Figure 6-5 Spectra with Ozone Cell in Beam (Page 4 of 7)
-------
8378-20
RVG OflfR
10= 73
NORMALIZED ORTR
10= 73
9 JULY, 1500 EST
170 p gln?x 1600m OZONE
7.00 8.00 9.00
. 10.00 11.00 12.00
WAVELENGTH (MICRON)
13.00 It. 00 . IS.00
CO
7.00 8.00
9 00 10.00 11.00 12/OO
HHVELENGTH tMICROW)
1 .00 1M.OO
Figure 6-5 Spectra with Ozone Cell in Beam (Page 5 of 7)
-------
8378-20
flVG DflTR
ID= 61
NORMALIZED DflTP
10° 61
7 JULY. 2230 EST
378/tgfm3x 1600m OZONE
i
NJ
O
mo.
(—*
s-
g.
7 00
8.00
9.00
10.00 11.00 12.00
HPVELENGTH (MICRON)
13.00
11.00
15.00
cn
CO
o
7 oo a.oo
9.00 10.00 11.00 12.00
HRVELENGTH (MICRON)
13.00
U.OO
Figure 6-5 Spectra with Ozone Cell in Beam (Page 6 of 7)
-------
8378-20
PVG ORTR
10= 79
NOBHRLIZEO OfiTR
10= 79
9 JULY. 1800 EST
385/ig/m3x 1600 m OZONE
e.oo
9 00
10.00 1 .00 12.00
HRVELENGTH (MICRON)
13.00 111. 00 1 .00
td
CO
OJ
o
8.00
9.00 10.00 11.00 12.00
HRVELENGTH CMICRON)
13.00 1M.OO
Figure 6-5 Spectra with Ozone Cell in Beam (Page 7 of 7)
-------
BSR 3027
Air Pollution Control Administration. The ozone absorption bands at 9- 5 and 9. 7 fjm
are clearly visible in the ozone -containing spectra. The cause of the erratic ab-
sorption at 10. 8 jjrn is not known.
6. 3 PASSIVE -MODE SPECTRA
Figure 6-6. shows two. spectra obtained with the instrument operated in the pas-
sive mode. Here the source, was. a rectangular black-painted cavity in the wall of a
water tank. The size of the cavity was a little larger than. the diameter of the re-
ceiver optics, and it was placed ab.out a meter away. The water in the hot source
(spectrum shown on the left) was 14. 0°C above and in the cold source (spectrum
shown on the right) was 12. 7°C below the room temperature of 25. 5°C.
6.4 ABSORPTION COEFFICIENT OF OZONE
The following calculations were performed to obtain the absorption coefficient
Kof ozone at 9. 7 ^m from the spectral measurements made by the Ozone Monitor
and the ozone concentration measurements made by Dr. Barnes using the potassium
iodide method. The data used are presented in. Figure 6-5;.
Calculations were made at the 9. 7 -jam ozone abs.orption band rather than at
the stronger 9. 5-jjm band, because the width of the wider 9- 7-|jm band is large
compared with the. spectral resolution of the ozone monitor and, hence, the results
are not very sensitive to instrument bandwidth or spectral calibration.
For each spectrum, it is necessary to find by a baseline method what the output
signal would have been in spectral channel 21 (9.69 Hm) if1 no ozone had been present
in the beam. It has previously been found by regression analysis that spectral channel
28 (10.40 fjm), where ozone does, not appreciably abs.orb, is a good predictor of the
output signal in spectral channel 21. From the spectra which had ambient air in the
test cell, bearing the ID numbers 55 and 67, it was found that if the channel 28 out-
put signal were multiplied by 0. 986 it would predict the channel 21 output signal in
the absence of ozone, Po. P is the actual signal in channel 21 with an ozone con-
centration path length product CL in the cell. The transmittance T is given by:
The results of these calculations are shown in Table, 6-1.
It can be seen from Table 6-1 that the K from ID 78 should be eliminated as
being far out of line with the others. The small fractional; absorptance in ID 78 and
the small ozone concentration both tend to produce large relative errors in their
measurement. The average of the remaining four measurements of K (9. 7 fji m) is
1 . 04 m2/g.
6-22
-------
8378-23
flVG OflTfl
10= 82
RVG DflTfl
10° 83
I
is)
to
> oo a.oo
9.00 10.00 1 .00 12.00 13.00 14.00 IS.00
MflVELENGTH (MICRON)
w
CO
OJ
o
6.00 9.00
10.00 11.00 12.00
URVELENGTH (MICRON)
13.00 l«.00 IS.00
Figure 6-6 Passive Spectra
-------
BSR 3027
TABLE 6-1
CALCULATION OF K (9.7|un) FROM SPECTRA
Spectrum
ID
78
75
73
61
79
Channel
28
57.9
37.7
62.5
93.0
54.8
PO
57. 1
37.2
61.6
91.6
54.0
P
51.3
30.7
45.7
57.8
25. 8
T
0. 898
0.825
0.742
0.630
0.478
KCL
0.11
0. 19
0.30
0.46
0.74
CL
(g/m2)
0.027
0. 176
0.272
0.604
0.616
K
(m2/g)
4.0
1.08
1.1
0.76
1.2
Dr. R. F. Calfee of The National Oceanographic and-Atmospheric Administration,
Boulder, Colorado, reports a K (9. 7 pm) of 0. 29 m^/g, while from the results of
Anding's* survey a K (9. 7(im) of 0.42 m2/g can be calculated. Figure 6-7 shows
the absorption coefficients of ozone as a function of wavelength as supplied by
Dr. Calfee.
The calibration of the Ozone Monitor is based on independent simultaneous
measurements of ozone concentration which are essentially equivalent to measure-
ments of the ozone absorption coefficient.. -The accuracy of the Ozone Monitor
readings is thus, in part, dependent upon obtaining correct ozone absorption
coefficients.
6. 5 EVALUATION OF OZONE READOUT
Several extended runs were made to evaluate the fluctuations or noise in the
ozone concentration readout. These runs were made in Ann Arbor and in Cincinnati
The University of Michigan, Willow Run Laboratories, "Band Model Methods for
Computing Atmospheric Molecular Absorption, " by D. A-nding, Report No. 7142-21-T,
AD 815 481, February 1967.
6-24
-------
BSR 3027
o
•
CO
o
CO
I/)
o
co
0=-
UJO
Oo
co
co
«x
o
oo
o
o
. /
°Z.OO 8.00 9'. 00 10.00 11.00
WfiVELENGTH (MICRON)
12.00
13.00
Figure 6-7 Ozone Absorption Coefficients
6-25
-------
BSR 3027
and at ranges of 1. 67 and 1. 08 km. Fifty spe.ctral channels were observed during
these runs which is a sub-optimum use of time, inasmuch as only seven channels
are used in the regression equation. The Ipw-pa.ss time constant for the readings
was 145 sec (2.4 min). The standard deviation of the readings varied between
2 and 4 ng/m3 which cor responds, to .about 1% absorptance in the 9.5 |im spectral
channel due to ozone. This deviation is due in part to instrument effects such as
noise, also includes atmospheric effects such as scintillation, and in addition in-
cludes any actual fluctuations with time, in the ozone concentration in the optical
path. Regression and factor analyses of the spectra showed there were no sig-
nificant interferences present.
The expected noise in, the ozone readout No can also be calculated from the
measured noise of the individual channel readings, N., and a knowledge of the
coefficients used for each .channel. The .coefficients used for each channel are
scaled to be 10 times those given by Figure 4-3. The coefficients used are 367. 9,
318. 5, - 876. 3, - 341. 1, - 38,4. 5, 488. 6, and 455. 2. Because the square of the
output noise is the sum of .the squares of the npise in .each channel, one finds the
square root of the sum .of the squares (RSS) of the coefficients and it is 1310. To
obtain the output noise, the channel noise is multiplied by the RSS coefficient, divided
by the first regression sum [RGS(l)] (which is the average channel signal, and
which normalizes the results to constant signal strength), divided by the range (R)
in kilometers, divided by four (which was done to make the .output readings conform
to values of the ozone absorption coefficients deduced from;Dr. Barnes1 measure-
ments rather than from Dr. Calfee's coefficients), and divided by ^ 16
(to allow for low-passing the output results over 16 spectra;). This calculation is
performed for the spectrum with ID 60 shown in Figures 6-3 and 6-2.
N. X RSS
RGS (1) XvR X 4 X N/~76"\
3.7 X 1310
80. 7 x ..!., 6.x 4X4.
, .
= 2< 3
Figure 6-8 shows the ^ozone measurements ma.de by the Ozone Monitor with
an ozone -containing cell in the beam plotted against the ozone concentrations re-
ported by Dr. Barnes. One can see the nonlinearity of the fractional absorptance
with ozone concentration when the fractional absorptance becomes large. Fig-
ure 6-7 can be used as a calibration curve (for 1. 6-km range) to convert indicated
ozone concentrations to actual ozone concentrations; or, the PDP-8/L computer
can be programmed to correct for the nonlinearity in, fractional absorptance.
6-26
-------
BSR 3027
400
200
a..
LU
Z
O
o 100J
CO
CM
100 200 300
OZONE Ug/m3) - Kl METHOD
400
500
Figure 6-8 Ozone Monitor Readings
6-27
-------
BSR 3027
The relative error of the data points from the "best fit" line is approximately
15%. This scatter is, of course, 'due to: imprecision in the potassium iodide (KI)
method of measuring ozone, sampling errors, as well as possible fluctuations in the
Ozone Monitor reading. Given the uncertainty in the KI method, it is impossible to
draw firm conclusions about the Ozone Monitor, but assuming that the precision of
the KI method is ± 10%, the precision of the Ozone Monitor readings is also about
10%.
6-28
-------
BSR 3027
SECTION 7
DISCUSSION AND RECOMMENDATIONS
The Ozone Monitor, as delivered, has met the contract specifications for
the instrument. Much more data will, of course, need to be gathered with this
instrument, especially in polluted urban environments. These additional spectra
will better define the interferences likely to be encountered. A new selection of
wavelengths and coefficients will then permit the rejection of these interferences.
Several possibilities for further improvement should be explicitly pointed
out. One is an increase in the range. By only looking at seven channels rather
than the current general-purpose 50, an increase in duty cycle of seven times can
be achieved. By low-passing the ozone readout over 128 more rapidly acquired
spectra, the S/N ratio can be increased by *J 128/16 or N/~8~ which is 2. 8. Also, to
the extent that one is willing to increase the instrument time constant above the
present 2. 3 minutes, additional S/N improvement can be gained. This improved
S/N ratio can be traded-off for increased range, and it is estimated that the range
can be increased to 3. 2 km (2 miles) or perhaps even to 4. 8 km (3 miles) with no
change in the size of the optics and no appreciable degradation of performance in
the measurement of the lower ozone concentrations. (A change in the phase-lock
circuitry may be necessary for this increased range.)
Another possibility is an increase in resolution. If it is desired to obtain
higher-resolution spectra, this can be done by stepping the filter wheel with fewer
steps between spectral channels. The resulting spectra with overlapping channels
can be computer-processed (off line) to remove the effect of the instrument response
function which degrades the inherent resolution to the resolution of the filter wheel.
This computer-processing is described in some detail in the reference cited in
Section 3.2.* and is also discussed in the reference cited below. ** It is estimated
that an increase in resolution to 0. 03 pm is achievable by this technique.
*R. H. Dye and A. Prostak, op. cit.
**P. A. Jans son, R. H. Hunt, and E. K. Plyler, "Resolution Enhancement of
Spectra, " Jour. Opt. Soc. Am. 60, 596(1970).
7-1
-------
BSR 3027
Additional studies made with this instrument should continue to include inde-
pendent measurements of the amount of pollutant in the beam, so as to better
determine the readings that the Ozone Monitor is expected to produce. This is
analogous to the problem of defining "ground truth'1 for remote sensing instruments.
It now appears that satisfactory, re suits could be achieved with an uncooled
infrared detector such as a pyroelectric detector in place of the currently installed
mercury-doped germanium detector. The measurements on the Ozone Monitor indi-
cate that its performance exceeds the contract specification and is not now limited
by detector noise. As a result, a detector with a larger noise equivalent power
could be used with no relaxation of the performance specifications.
Use of an uncooled detector could reduce.the-size, weight, and cost of an in-
strument and eliminate the additional complication caused by the detector refriger-
ator.
When further field: studies with the present instrument define firm operational
requirements and a-limited number of-useful spectral channels, a less versatile but
simpler, smaller, lighter, and cheaper instrument can be devised.
7-2
-------
BSR 3027
APPENDIX
DETAILED OPERATING INSTRUCTIONS
A. 1 TRANSMITTER UNIT
A. 1. 1 Operation
The main power switch turns on the chopper motor and two fan motors and
also supplies power to the temperature controller. The temperature controller
has its own switch. For turn-on, the main power and temperature controller
switches should be turned on. For turn-off, first turn off the temperature con-
troller and wait about five minutes for the source to cool, while the fans and chop-
per are still running. Then turn off the main power switch. This will prevent
heat damage to the chopper wheel and other components.
A. 1. 2 Maintenance
There are two fuses in the transmitter, one before the main power switch
and one in the temperature controller.
A polyurethane foam filter is located behind each of the two fans. The
filters should be periodically removed when they become dirty, cleaned by washing
in water, and reinstalled.
The two mirrors can become quite dirty visually before they need recleaning.
They should be cleaned with clean absorbent cotton and alcohol. The window over
the silicon cell for the temperature controller should be cleaned occasionally.
This silicon cell is mounted on the secondary mirror and can be reached after re-
moving the tubular baffle.
A. 1. 3 Repair
If the silicon carbide source burns out, it should be replaced with another
80-V source. The source is approached from the bottom of the cylindrical en-
closure after the back cover and lower fan are removed. When removing the back
cover, it is necessary to unplug the connector that supplies power to the back fan.
The sealing material around the heater is chipped away. After the new heater is
installed, a piece of the Fibrefrax high-temperature cloth is cut to shape, dipped
A-l
-------
BSR 3027
in Rigidizer, and pressed into,place. Some of the high-temperature cement is
poured over the cloth to provide, after hardening, an air seal. After about 10 min-
utes of air hardening, the heater can be momentarily turned on to provide heat to
accelerate the curing. The heater should not be turned on for more than a few
seconds at a time, .unless it is in its normal position where it can be seen by the
optical feedback detector mounted in the baffle on the secondary mirror.
A. 2 RECEIVER UNIT
A. 2. 1 Operation
Connect the cables from the main rack to the receiver unit including the
ground connector attached to one of the connectors. Attach the cooler control box
to the cooler and fan using the two cables supplied. Plug in the power cord for the
control box. Point the receiver toward the transmitter. The three-position wheel
in front of the vacuum jacket on the cooler should normally be in the clear position.
The internal chopper wheel should be locked in the open position with the thumb-
screw all the way in for .active operation or unlocked with the thumbscrew all the
way out for passive operation. It is necessary to remove the outer fiberglass
cover and a filter panel to gain access to the three-position wheel or the thumb-
screw.
If more, than two weeks has elapsed since the last pump-out of the vacuum
jacket, it will probably be necessary to repump the vacuum jacket. It is necessary
to remove the outer fiberglass cover and a filter panel to gain access to the vacuum
jacket.
ONLY CONNECT THE MAIN DETECTOR TO THE
PREAMPLIFIER WHEN THE "SYSTEM1* OR "RECEIVER
ELECTRONICS" SWITCH ON THE CIRCUIT BREAKER
PANEL IS TURNED OFF.
i : • !
Turn on the switch in the cooler control box when a cold detector is desired
for operation. Cool-down time is approximately 15 minutes.
A. 2. 2 Maintenance
Occasionally, it will be desirable to remove, wash, and reinstall the four
foam filters on the receiver unit.
When the cooler stops operating properly it will have to be returned to the
Malaker Corporation for repair. (See Malaker Cooler Manual). The time between
repairs is claimed to be greater than 500 hr operating time. It is suggested that
A-2
-------
BSR 3027
the flanged cylinder holding the cooler be removed from the receiver. Then re-
move the blower from the flange and remove the cooler from the outer cylinder by
removing the three nylon set screws at the rear of the cylinder and the three screws
at the front of the cylinder. The cooler will then come out of the cylinder. The
vacuum jacket should be left on the cooler when shipping the cooler for repair.
Side-to-side and up-down adjustment of the detector for optical realignment
is obtained by moving the large cylinder on the vertical plate to which it is attached.
This is accomplished by removing or loosening the bolts on the bolt circle holding
the large cylinder and adjusting the three set screws on the side of the cylinder
that positions the large cylinder. After the adjustment is completed, the bolts
should be replaced and tightened.
In-out adjustment of the detector is made by sliding the small flanged
cylinder that holds the cooler. Three thumbscrews and four regular screws work
in opposition to control the in-out movement of the small cylinder.
Focusing of the collecting telescope is accomplished by loosening the nylon
set screws on the secondary mirror threaded shaft, loosening the locking ring and
rotating the threaded shaft. This.moves the secondary mirror in and out. The set
screws can slightly tilt the secondary mirror.
To allow adjustment and refocusing, most of the parts are not pinned.
Therefore, after disassembly the parts cannot be easily returned to their present
relative positions. Consequently, the bolts holding the various optical elements and
the cooler should not be loosened unless realignment is desired.
After 500 hours of use, the oil in the stepper motor should be changed. The
oil should be drained (by inverting the motor) and the motor refilled with 5 ml of
approved* oil. Failure of the motor to step properly at high speed, especially in
cold weather, may be due to too much oil. The motor can be partially drained of
oil without harm.
FWREF is found as follows. The built in polystyrene film (or any other
material with known absorption bands) is inserted in the beam. A filter wheel po-
sition list is used that has closely spaced (one-step) increments in the vicinity of a
known polystyrene absorption band and that also includes a filter wheel position on
the other side of FWREF to assure that the slit in the filter wheel rim will pass by
USM Spec 66 BR Lubricant, USM Corporation, Gear Systems Div., 5033 Balch St.
Beverly, Mass. 01915 or Humble Oil Spinesso-34.
A-3
-------
BSR 3027
the light. Spectra are then taken. The minimum signal in each spectrum should
be at the filter wheel position-shownun'Table 2-2 as corresponding to the wavelength
of the known absorption band. If the minimum"is not at the correct filter wheel po-
sition, FWREF is changed slightly so as to bring the spectrum minimum to the
correct filter wheel positions. The known absorption band,on which the calibration
is based should be near a wavelength of interest (e.g., 9.7 IJL for ozone measure-
ment). Extremes of ambient temperature cause the interference films on the filter
wheel to expand or-contract and will probably require a change in. FWREF to main-
tain the wavelength calibration.
Lubrication of the tripods* is required twice a year. Put one drop of light
lubricating oil on the threads of all clamp screws; on1 each side of the joint where
legs are hinged to tripod plate, and on hinge joints-at both ends of leg struts. Do
not apply any oil to lower legs or strut tube. The horizontal and vertical pivot and
anti-friction bearings on the pan heads are permanently lubricated. Twice a year
apply one drop of light lubricating oil to the threads of clamp screws. Apply one
drop of lubricating;oil to hand wheel shaft bearing-of the gear drive.
A. 3 MAIN ELECTRONICS RACK
A. 3.1 Operation
Plug the Teletype power cord into the back of the switch panel and join the
connectors on the Teletype signal cable. Plug the main power cable into a source
of 117 V, 60 Hz single-phase AC power. Throw the main power switch on. Turn
on the oscilloscope and the cooler switch (on the cooler control box). All the other
switches on the switch panel can be turned on except for the chopper switch if the
instrument is to be using the active source unit. For passive operation, with the
internal receiver chopper unblocked, the chopper switch should be turned on. The
computer should be turned on and the mode switch set to active or passive as ap-
propriate.
The manual for the PDP-8/L computer supplied by the Digital Equipment
Corporation should be consulted for details of computer operation. In brief, the
group of 12 keys in the middle of the computer comprise the SWITCH REGISTER.
The switches in the SWITCH REGISTER are grouped in threes, corresponding
within each group to the digits 4, 2, 1 from left'to right. The sum of the digits
selected in any group gives the octal number for that group'. For example, the
first switch and the third would be 4 + 1 or 5. If'every other switch in the switch
register were up starting with the first one, the number would be (5252)8, and if
all were up the number would.be (7777 )g. The number (0200)g, corresponding to
the starting address of the OZONE MONITOR main program, is made by lifting
A-4
-------
BSR 3027
the second switch in the second group. Some references in this manual are to a
specific switch starting from the left side of the SWITCH REGISTER numbered
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, and 11.
The Ozone Monitor program ("everything" tape) is loaded with the binary
loader. The starting address of the Ozone Monitor program is 0200. Therefore,
set the computer switch register to 0200. Depress LOAD ADDRESS switch. The
number 0200 should appear on the Program Counter lights. Depress START
switch. The computer RUN light should turn on. If the detector is cold, normal
operation can commence. If an error halt occurs, the computer can be reinitialized
by depressing CONT (Continue) or by reloading the start address (0200).
A slit is provided in the rim of the filter wheel. A light source and a
phototransistor are arranged such that when radiation of a certain wavelength in
the image passes through the filter wheel, the light from the lamp passes through
the hole in the filter wheel and actuates the photodiode. This is fed to the computer
in the form of a reference pulse which the computer expects at the same point in
every scan. The computer stops the operation if it fails to receive the pulse
when it is expected. If either the lamp of phototransistor is changed in position
or the detector image on the filter wheel is changed, the reference pulse will occur
at a different position and should be checked by observing something with a known
spectral absorption, such as the built-in polystyrene film, to assure that it is in
the same position. If the entire spectrum is shifted up or down in wavelength, the
position of the reference pulse should be changed accordingly in the software by
depositing a new channel number into memory location 0137. The present channel
number corresponding to the filter wheel reference pulse is 0373g. If switch
register switch no. 10 is lifted, the program automatically (in effect) "continues"
if a filter wheel error is encountered.
As a guide to receiver alignment, it has been found that on a somewhat hazy
day at 1.6 km (1 mile) range, the auxiliary channel signal (oscilloscope channel 1,
switch position 11) was 0.50 V peak to peak. The signal for filter wheel position
275g was 1. 2 V on the bar chart. (To look at this single bar, load 0001 in location
0176g, and load 0275g in location 41008- To return to tlie 50-channel bar chart,
load 0062g in location 0176g, and the first filter wheel position (0011) in location
41008.
To maximize the receiver signal when the oscilloscope can not be seen from
the receiver, the hand-held multimeter can be used. Using a BNC-to-alligator clip
connector, attach the multimeter leads to the channel 1 output on the control panel
and observe the main integrator output (oscilloscope channel 1, switch position 9)
on DC volts. A single strong channel should be used, either by using a one-channel
filter wheel position list or by lifting computer switch no. 2. Align the receiver to
maximize the main integrator output.
A-5
-------
BSR 3027
To turn off the instrument, first press the STOP switch on the computer.
Then turn off the computer with the key switch.
CAUTION
DO NOT TURN OFF THE POWER TO THE COM-
PUTER WHILE IT IS EXECUTING (RUNNING).
THIS MAY CAUSE THE OZONE MONITOR PRO-
GRAM IN THE COMPUTER MEMORY TO BE
CHANGED, NECESSITATING A RELOAD OF THE
PROGRAM "EVERYTHING" TAPE INTO THE
COMPUTER.
A.3.1.1 Control Panel
The control panel to the right of the oscilloscope has two rows of five
pushbutton switches to communicate with the computer. The lights behind most
of these switches are computer controlled.
Top Row
DISABLE COMPUTER INTERRUPT - This switch at the top left is pressed
ON when it is desired to disable computer interrupts from the ozone monitor hard-
ware. This would be done if one wanted to use the computer as a normal PDP-8/L
computer without interaction with the Ozone Monitor equipment. For normal ozone
monitor use, this switch should be OFF as indicated by the light behind it being
OFF.
LO PASS - When this switch is ON and if the detection processing switch
is ON, the indicated ozone concentrations, are low-passed as provided in the Ozone
Monitor program in the constant INTC where INTC is currently 4, producing a low-
pass time constant of 2^ or 16 spectra. This reduces the noise by 0.8 X \/F& or a
factor of 3. 2. If the LO PASS switch is OFF, there is no low passing.
DETECTION PROCESSING - This switch when ON causes the computed
ozone concentration to be output on the Teletype - currently after every fourth
spectrum.
PRINT DATA - This switch when ON causes the spectrum to be printed
on the Teletype in decimal notation. While the Teletype is printing the spectrum
at 10 characters/second, additional spectrum gathering ite suppressed. The fifth
blank switch button is a spare that can be used when modifying the Ozone Monitor
computer program to provide additional capability.
A-6
-------
BSR 3027
Second Row
TIMES 4 - This switch increases the resolution and size of the spectrum
bar chart that can be displayed on the oscilloscope by a factor of 4. The magnified
bar chart display is also four times closer to the maximum signal that can be
displayed without numerical overflow. Numerical overflow at the D/A converter
causes a large positive number to be displayed as a negative bar in the downward
direction. Hence, the TIMES 4 switch should be ON for weak signals, such as
are received at 1.6-km range, but should be OFF for strong signals, such as are
received at 0. 4-km range.
LINE LOCK - This switch is not currently used. The light in the switch
is ON when the instrument is in the active mode and the auxiliary signal is too
small to supply timing information. Hence, the circuitry changes from signal
lock to line lock. (Occasional noise pulses may appear large enough to lock onto
so that the light may flicker off. )
INCREMENT I. D. - This switch is pressed and released to increase the
identifying number associated with a spectrum or group of spectra.
PUNCH DATA - This switch, when depressed, causes the spectra to be
punched on the high-speed punch. Obtaining the spectra in this way does not slow
down the rate of gathering spectra.
The last blank switch is not currently used.
The leftmost connector receptable is controlled by the left 24-position
switch on the interface panel and is normally connected to the left oscilloscope
channel (Oscilloscope Channel #1). The middle connector receptacle is con-
trolled by the right 24-position switch on the interface panel and is normally con-
nected to the right oscilloscope channel (Oscilloscope Channel #2). The rightmost
connector receptacle is controlled by the three-position switch on the interface
panel and is normally connected to the oscilloscope trigger synchronizing signal
input. The Tektronix Oscilloscope Instruction Manual contains detailed informa-
tion on oscilloscope operation and maintenance.
A. 3. 1.2 Analog-to-Digital Converter
In normal operation the A/D Converter word length switch is left on 12
and the power switch is left ON. The three other rocker switches are not normally
used.
A-7
-------
BSR 3027
A. 3.1. 3 Interface Panel
The left 12-position switch on the interface panel is normally on position 1
for spectrum bar chart display. The other positions are used for setup and trouble-
shooting. The three-position sync switch should be on BAR chart sync for bar
chart display.
The right 24-position switch and the power supply multiple position
switch below it are not normally used but are useful for troubleshooting.
Table A-1 shows the circuit locations that can be easily examined on the
oscilloscope by using these two 2'4-position switches. Of course, any other loca-
tion can be examined by using the oscilloscope probe.
A. 3.1. 4 Teletype Controls
. The switch on the right side of the Teletype unit has three positions:
LINE, LOCAL, and OFF. The LOCAL position allows the operator the control of
the instrument for punching out tape or for any other reason (1. e., merely using
the typewriter). The LINE position assures a direct link with the computer for
reading in programs, or punching or typing out responses (data collection, program
parts, detection processing, etc.). Switches on the punch panel on the left of the
Teletype control the tape-punch operation - on or off, back space, and release. On
the lower left is a switch with three positions: START, SfTOP, and FREE. These
control the tape read-in mode to the computer. The tapeiis secured mechanically
into the reader by a latch in such a way that the tape advance sprockets must be in
the tape perforations.
A. 3. 2 Phase Lock Check
The phase lock circuit should be checked after setting up at a new location
as follows.
-j
Confirm that the remote chopper is locked to the local line frequency by
putting the oscilloscope on line sync and observing the auxiliary signal (channel 1,
switch position 11). The aux signal should be -stationary .on the screen and not
drift.
Confirm that the phase lock circuit is locked on the auxiliary signal by
putting the oscilloscope on line sync and simultaneously looking at 60-Hz sine wave
(channel 1, switch position. 17) and a nominal 60-Hz square wave obtained from the
phase lock circuit (channel 2, switch position 17). The two waves should remain
in constant relation to each other. If there is drift of the square wave relative to
the sine wave,: the phase lock circuit is not locked on the auxiliary signal.
A-8
-------
BSR 3027
TABLE A-1
TEST PANEL SWITCH POSITION IDENTIFICATION
Position
i
i
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Channel 1
B12B2 Bar Chart
J6B Main Detector "I" Bias
SE20 Filter Wheel CCW
Rotation
J6C Main Detector Signal
B20V2 Main A. R.C. 1
B20U2 Main A. R. C. 2
B16N2 Main (-) Complement
B19V2 Main Demodulator
B19K2 Main Integrator
B19L2 Multiplexer
J6E Aux. Det. Signal
A20J2 Aux. A. R.C. 1
A20H2 Aux. A. R.C. 2
A19R2 Aux. Demodulator
A19U2 Aux. Integrator
A4U2 5 CR Trigger
SW3-1 Line Sync
A6K2 Receiver Chopper
A14M2 AFC Voltage
A16M2 AFC Phase Demodulator
A16H2 AFC Integrator
B16J2 AFC Sample
B16K2 AFC Dump
Channel 2
Power Supplies
J6A Main Detector "E" Bias
SE22 Filter Wheel CW Rotation
A6J2 Filter Wheel Index Pulse
A6P2 Filter Wheel Complement
A6V2 A. R.C. Gain (Log)
B16U2 Main (+) Complement
A19N2 AUX ( + ) Complement
B6N1 Dump Complement
B19J2 Select Main Complement
Spare
Spare
SW5-2 Phase Reference
Spare
Spare
A4H2 Ramp
A15V2 60
A15S1 120
A15P2 240
A15L1 480
A15H2 960
A15E1 1920
Spare
A-9
-------
BSR 3027
To attempt to have the phase lock circuit lock on the auxiliary signal,
first block the beam. The phase lock circuit should now lock on the local line fre-
quency. Then unblock the beam and the phase lock circuit should lock on the
auxiliary signal. If it does not, the voltage-controlled oscillator (VCO) control
(on card 14 A in the interface unit) may need slight adjustment. This control is
very touchy and should only be moved a very small amount.
When the phase lock circuit is locked on the auxiliary signal, examine the
automatic frequency control (AFC) voltage (channel 1, switch position 19). The
AFC voltage shouldbe 0 ± 1 volt. If it is out of this range, the VCO control should
be slowly and deliberately adjusted to bring the AFC voltage within the above range.
Only a very small adjustment should be necessary.
Finally, the trace on oscilloscope channel 1 switch position 20 should look
as shown in Figure A-1 (a). If the oscilloscope trace looks as shown in Figure A-l
(b), then the AFC offset voltage (marked "0" on card 16) should be adjusted until
the vertical breaks occur at extrema of the wave form.
A. 3. 3 Maintenance
The polyurethane foam dust filter on the top of the cabinet should be oc-
casionally removed, washed, and replaced.
A. 3. 4 Repair
The following circuit diagrams are provided in accompanying envelopes.
They will permit troubleshooting and repair, should it become necessary. The
diagrams are identified by title and drawing number.
System Cabling Diagram BSX 10815
Circuit Breaker Panel BSX 10813
Control/Display Panel BSX 10802
Interface Chassis BSX 10803
Phase Lock BSX 10807
Line Receivers and ARC D/A BSX 10812
Automatic Range Control
Amplifiers BSX 10805
A-10
-------
BSR 3027
a. Good
n
oo
b. AFC Offset Needs Adjustment
Figure A-l Oscilloscope Trace From AFC Phase Demodulator
A-ll
-------
BSR 3027
Auxiliary Preamplifier BSX 10814
Main Preamplifier BSX 10808
Receiver Electronics BSX 10801
Fotofet Amplifier BSX 10809
Synchronous Demodulators,
Integrators, and Multiplexer BSX 10806
Stepper Regulator BSX 10804
D/A Converter BSX 10811
To gain access to the wire-wrap pins on the plug-in card receptacles in
the interface chassis, it may be necessary to remove some of the screws that
fasten the metal frame (that holds the card receptacles) to the slides. The metal
frame can then be rotated 90° using two remaining screws (one on each side of
the frame) as a pivot. In this position, the wire-wrap pins-are accessible.
A. 4 USING THE COMPUTER SOFTWARE
All control of the system is initiated at the computer console and is imple-
mented by commands from the user to the computer through the control panel
switches. The brief description here of the switches on the computer console
should enable the operator to use the system successfully. A more detailed ex-
planation can be found in the DEC Small Computer Handbook* or in the PDP-8/L
Users Handbook. **
On the bottom of the PDP-8/L operator console is a row of 22 keys. The
middle group of 12 keys comprise the SWITCH REGISTER. The switches in the
SWITCH REGISTER are grouped in threes, corresponding within each group to the
digits 4, 2, 1 from left to right. The sum of the digits selected in any group gives
*DEC Small Computer Handbook, Digital Equipment Corporation,
Maynard, Massachusetts.
**PDP-8/L Users Handbook, Digital Equipment Corporation,
Maynard, Massachusetts, 1968.
A-12
-------
BSR 3027
the octal number for that group. For example, the first and third switches would
be 4 + 1 or 5. If alternate switches were in the up position starting with the first
one, the number would be (5252)g; if all were in the up position, the number would
be (7777)g. The number (0200)g, corresponding to the starting address of the
Ozone Monitor main program, is made by lifting the second switch in the second
group from the left. Specific switches are numbered 0, 1, 2, 3, 4, 5, 6, 7, 8, 9,
10, and 11 starting from the left side of the SWITCH REGISTER.
Before the computer can respond to operator control for data gathering or
detection processing, it must be loaded with a binary tape containing the programmed
instructions for this application.
A. 4. 1 Loading the Binary Tape
Turn the computer power key on. Put the Ozone Monitor "everything" tape
in the high-speed tape reader. Turn the reader on. Set the Switch Register to 7777
and depress the LOAD ADDRESS switch. The number 7777 should appear on the
Program Counter lights. Put switch number 0 in the down position and press the
START switch. The computer should now be reading the contents of the tape into
its memory as evidenced by the physical reading of the tape in the reader. If
nothing happens when the START switch is depressed, the operator should repeat
the above steps. If the tape is still not being read, it will be necessary to reload
the BINARY LOADER.
To make small changes in the program in the computer by overwriting the
program from the "everything" tape, make a change tape using the punch on the
Teletype in the LOCAL mode. The format of this change tape is analogous to that
used in using the computer switch register to change the program. It is
XXXX/YYYY (CR) (LF)
ZZZZ (CR) (LF)
etc.
where XXXX is the address in which YYYY is to be deposited. ZZZZ is deposited
in the next address location. CR and LF are "carriage return" and "line feed. "
To load the change tape, place the tape in the Teletype reader. Set the Switch
Register to (7400)g and depress LOAD ADDRESS. Turn on the Reader and depress
START.
A-13
-------
BSR 3027
A. 4. 2 Loading the Binary Loader
The set of standard DEC tapes supplied with the computer contains one labeled
BINARY LOADER. This tape contains instructions that enable the PDP-8/L to read
binary encoded programs (the Ozone Monitor program is an example). Put the
Binary Loader tape in the Reader. Set the Switch Register to 7756 and press the
LOAD ADDRESS switch. Put the first switch (number 0) in the down position and
press START. Additional details of the Binary Loader routine can be found in the
software manuals supplied with the computer, and it is- suggested that the user
consult this documentation. If the Binary Loader does not load properly, the Rim
Loader should be checked. This short series of manually loaded instructions is
described in the software manual which should be consulted in the event of failure
to load the Binary Loader. Successful loading' of the loaders is then followed by a
repetition of the steps listed above for loading the Ozone Monitor binary tape.
A. 4. 3 Initializing;the Program
The software is ready to use upon successful loading of the binary tape, as
indicated by the number of 0000 appearing in the accumulator after the tape has
been read into the computer's memory. To begin operation, place the number
(0200)s in the Switch Register and press LOAD ADDRESS. This number, which is
the starting address of the Ozone Monitor main program, should appear on the
Program Counter lights. To start the instrument, press the switch labeled START.
This operation enables the software to begin exercising its control over the instru-
ment and its data. This control will be apparent to the operator because he will be
able to see and hear the filter wheel stepping to and stopping at the positions at
which it is to take readings. The instrument is now ready to be operated in any of
the ways previously described. While in the data-gather ing mode, the user may
desire a punched paper tape containing the list of filter wheel positions being used.
To obtain this hard copy, press STOP, load 0200, the starting address of the main
program, power the high speed punch, lift up switches 1, and 11, and press START.
Occasionally it may be necessary to stop the instrument. This can be done
in either of-two ways: (1) depress the STOP switch, or (2)'lift up switch number 2.
The advantage of the second method is that the last spectral scan completed will
remain displayed on the oscilloscope screen. To resume data taking, push CON-
TINUE in the first; case or depress switch number 2 in the-second case.
It may prove convenient during system checkout to suppress the error halts
that result from the inability of the filter wheel to find its reference point on a
spectral scan or from signals too large for processing (A-D overflows). This
will result in the generation of a HALT instruction by the software and suspension
of all instrument activity. To prevent delay, the software;can be made to ignore
A-14
-------
BSR 3027
these errors by lifting switch number 11. If the error halts are not suppressed in
this manner, the system will stop gathering data. To resume activity, the spectrom-
eter has to be reinitialized. This will be done automatically if switch number 10
has been lifted. The upper-right light on the control panel will go on when the
software is setting up the initial conditions for operation. When the program is
halted by one of the error sub-routines, the address of the error routine will
generally appear in the program counter lights. This permits diagnosis and cor-
rective action should this error occur too frequently.
The following instructions should be followed to generate the bar chart
spectral display on the scope:
1. Set the Channel Selector No. 1 on the control console to position
No. 1. This position allows monitoring of the output from the
A/D converter. Also set the switch to BC, bar chart synchronization.
2. Trigger the scope on EXT sync and set the time scale to 0. 1
millisecond/div.
3. Channel No. 1 should be viewed at a gain of about 1 volt/div.
4. The scope settings can be varied from those given here to suit
the viewing preferences of the operator.
When viewing spectra at long range, it may be desirable to multiply the
channel readings by a factor of four to increase the oscilloscope digital resolution.
For this purpose, a TIMES 4 button can be pushed on the operator console. If
one merely increased the scope gain, the height of the observed spectrum would
increase, but the digital resolution would not be increased. As a result, one would
be observing truncation error and the finite number of digital steps allowed.
Another button on the control console is the upper-left button. This button,
which prevents the interface unit from generating interrupts to the computer,
should be pushed in when the computer is being used in a noncontrol capacity. Such
a use would be testing the PDP-8 or peripheral devices with standard DEC
MAINDEC Maintenance Routines. In this case, one would not want interrupts to
disrupt the program.
The value of C is in |j.g/m , providing that the value of 4 X L is placed in
the double-precision constant RNGC (for range constant). The 4 comes from the
factor necessary to convert the values of the ozone absorption coefficient K
obtained from Calfee to those obtained by Barnes and the L is the range in kilo-
meters. The number 4 X L should be in octal with the octal point between
A-15
-------
BSR 3027
the two words. For a range of one mile or 1. 609 km, .4 X L =;(6. 4375)j0 =
(6. 340)g. Then RNGC =0006 3400. For the concentration readings to be
correct, this number must be changed e,ver,y,time .the .range is changed.
A. 4. 4 Control Factions of Switch Registers
Following is a summary of the special-purpose switches of the Switch
Register:
Switch No.
10
11
Function
PUNCH POSITION LIST. If lifted, the program
punches on paper .tape .the filter wheel position list
when in the INIT subroutine.
STOP ACTION. If lifted, the computer stops the
filter wheel at its present location. The integrator
continues to integrate signals. The last complete
spectrum gathered remains displayed on the scope.
FILTER WHEEL ERROR HALT SUPPRESSION.
If lifted, the program returns control to the INIT
subroutine when filter wheel errors are encountered.
HALT SUPPRESSION. If lifted, the program sup-
presses halts which occur from A/D format over-
flows and filter wheel errors.
A-16
------- |