United States
Environments! Protection
Agency
Environmental Sciences Research
Laboratory
Reeearch Triangle Park NC 27711
EPA-600/2-79-121
July 1979
Research and Development
Development of an
SO2 Monitor for
Mobile Sources
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL PROTECTION TECH-
NOLOGY series. This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from point and non-point sources of pollution. This work
provides the new or improved technology required for the control and treatment
of pollution sources to meet environmental quality standards.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-79-121
July 1979
DEVELOPMENT OF AN SO. MONITOR
FOR MOBILE SOURCES
by
Darrell E. Burch, Pamela S. Marrs Davila,
Francis J. Gates and John D. Pembrook
Ford Aerospace and Communications Corporation
Aeronutronic Division
Newport Beach, California 92663
Contract No. 68-02-2448
Project Officer
Roy Zweidinger
Emissions Measurement and Characterization Division
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U. S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences Research Laboratory,
U.S. Environmental Protection Agency, and approved for publication. Approval
does not signify that the contents necessarily reflect the view and policies
of the U.S. Environmental Protection Agency, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.
-------
ABSTRACT
An instrument has been designed and built to monitor the concentration
of S02 in the exhaust of mobile sources. A grating assembly disperses the
ultraviolet energy from a deuterium-arc source and passes five narrow
spectral intervals. Three of the intervals (set A) coincide with strong
absorption features in the S02 band near 3000 A; the other two intervals
(set B) coincide with weak absorption. A spinning reticle alternately
transmits energy passing through set A and B to a photomultiplier detector.
A dc output signal proportional to the concentration of S(>2 in the sample
cell is produced by appropriate electronics.
A pump and manifold assembly permit operation over a wide range of
sample flow rates. At high flow rates, the 90% sample turn-over time is
approximately 0.7 sec. The sample cell contains a multiple-pass optical
system adjusted to 12 passes, giving a sample path length greater than
5 m. The rms noise level corresponds to approximately 0.05 ppm of S02
when the electronic time constant is 1 sec.
This report was submitted in fulfillment of Contract No. 68-02-2448
by Ford Aerospace and Communications Corporation, Aeronutronic Division
under the sponsorship of the U.S. Environmental Protection Agency. This
report covers the period September 30, 1976 to January 31, 1978, and work
was completed as of September 30, 1978.
1X1
-------
CONTENTS
\
Abstract iii
Figures v
Abbreviations and Symbols vi
1. Introduction 1
2. Conclusions 3
3. Recommendations 5
4. Layout and Optical Diagram 7
5. Sample Cell 15
6. Pumps and Manifold 22
7. Spectroscopic Principles of Detection 26
8. Electronic Processing of Detector Signal 31
9- Performance, Calibration, and Discrimination 33
-------
FIGURES
Number Page
1 Photograph of the Optics Assembly, the Manifold
Assembly, the Pump Assembly, and the UV Source
Power Supply 8
2 ' Photograph of the Optics Assembly with Front Hinged
Panels Closed 9
3 Photograph of the Optics Assembly with Front Hinged
Panels Opened 10
4 Optical Diagram of the S02 Monitor 11
5 Sectional View of the Sample Cell 17
6 Multiple-Pass Optical System 22
7 Flow Diagram of Pump Assembly 24
8 Flow Diagram of Manifold Assembly 25
9 Spectrum of S02 from 2600 to 3300 Angstroms 27
10 Partial Spectrum of S02 Band Showing Spectral Slits
A and B 28
11 Block Diagram of Signal-Processing Electronics 29
VI
-------
ABBREVIATIONS AND SYMBOLS
AGC —automatic gain control
cfh —cubic feet per hour
cfm —cubic feet per minute
cm —centimeter
1pm —liters per minute
m —meter
ppm —parts per million by volume
T —transmittance at a particular wavennumber (or wavelength) that
would be measured with a spectrometer having infinite resolving
power
T. —average transmittance of the absorbing gas (S0_) in spectral
intervals designated by A
T —average transmittance of the absorbing gas (S0_) in spectral
intervals designated by B
V_ —voltage component of amplified detector signal at carrier frequency,
360 Hz
V —voltage component of amplified detector signal at alternator
frequency, 30 Hz
V' —Va/Vc, normalized voltage. Normalization is made so that V' =
1 when there is 100% modulation of the beam at 30 Hz
Vll
-------
• SECTION 1
INTRODUCTION
Most of the sulfur in gasoline is oxidized upon combustion in an engine
to S02< Concentrations of this gas in raw exhaust are typically between 5 ppm
and 50 ppm. At these concentrations, the S02 is not considered a serious
pollutant unless it is oxidized further to form 803, which reacts rapidly
with.-H2-0- to form toxic sulfuric acid. Catalytic converters are known to
oxidize some of the S0£ to the more harmful 803. Unfortunately, the process
by which the 802 is converted to 803, and ultimately to acid, is difficult
to study quantitatively because there is no convenient, real-time method of
measuring the acid concentration in the exhaust. It has been found, however,
that much can be learned about the behavior of the catalytic converters and
about the entire 802 conversion process by monitoring the concentration of
802 in the exhaust.
The instrument described herein has been designed and built for this
purpose, to monitor in real-time the 802 concentration in engine exhaust.
The low concentration requires that the instrument be quite sensitive and
that the discrimination against other exhaust gases be good. Rapid changes
in the 802 concentration have been observed previously, and it is important
that these changes be recorded. This leads to another important requirement,
quick response, for the monitor.
The instrument achieves high sensitivity by making use of the very
strong 802 absorption band near 3000 A while employing a multiple-pass optical
system in the sample cell to obtain a long absorption path in a small volume.
An instrument built previously by usl for the EPA to monitor 802 ^n stac^
exhaust made use of the infrared absorption band of 802 near 4 ym. This
infrared band is too weak to provide the high sensitivity required for the
present instrument. Excellent discrimination has been achieved by employing
a correlation method that makes use of the structure in the uv absorption
band. A low-volume sample cell designed for high gas flow rates used in
conjunction with high-capacity pumps makes the quick response possible.
1. Burch, D. E., and D. A. Gryvnak. "Infrared Gas Filter Correlation
Instrument for In-Situ Measurement of Gaseous Pollutants." Prepared
by Philco-Ford Corporation for EPA under Contract No. 68-02-0575.
EPA Report No. EPA-650/2-74-094. Also, Burch, D. E., and D. A.
Gryvnak. "Cross-Stack Measurement of Pollutant Concentrations
Using Gas-Cell Correlation Spectroscopy." Chapter 10 of Analytical
Methods Applied to Air Pollution Measurements, Stevens, R. K. and
W. F. Herget, (eds.). Ann Arbor, Ann Arbor Science Publishers Inc.,
1974.
-------
The important design goals are summarized as follows:
Sensitivity ranges, full scale: 5, 10, 20, and
100 ppm.
Higher concentrations can be measured by
decreasing the number of passes of the
multiple-pass cell.
Noise-equivalent-concentration: =0.05 ppm
Sample turn-over time (90%): =0.5 sec at maximum flow rate.
Interference by other gases in automotive exhaust should not
produce significant error in the S0« measurement.
-------
SECTION 2
CONCLUSIONS
It has been demonstrated that an instrument that makes use of the strong
S02 absorption features near 3000 A can be designed and built with the sensi-
tivity and accuracy required to measure the concentration of S02 in automotive
exhaust. Good discrimination against interference by other gases in the
exhaust can be achieved by employing the correlation technique used in the
instrument described herein. This technique, when used properly, provides
very little sensitivity to changes in the reflectivity of mirrors, the trans-
mittance of windows, or variations in source intensity. Electronic circuits
developed and used previously by us to process the detector signals of gas-
filter correlation instruments work equally well with the present instrument,
which employs a grating polychromator with a spinning reticle in place of a
gas-filter cell.
The minimum detectable concentration is limited by photoelectron noise
generated in the photomultiplier. The noise level measured at the output of
the photomultiplier is approximately proportional to the square-root of the
radiant energy in the ultraviolet beam incident on the photomultiplier. It
follows that increasing the energy level by such means as increasing the
source intensity or improving mirror reflectivity would result in an increase
in S/N, the signal-to-noise ratio, that is proportional to the square-root
of the energy level.
The maximum S/N for a given S02 concentration is obtained in the present
instrument when the multiple-pass sample cell is adjusted to 12 passes. The
necessity for this low number of passes is due to the low effective reflec-
tivity (between 0.80 and 0.85) of the uv-enhanced aluminum coating on the
mirrors of the multiple-pass optical system. The actual reflectivity may be
higher than this value, but each reflection adds slightly to the distortion
of the images. This additional distortion has the effect of lowering the
reflectivity by decreasing the amount of energy entering the entrance slit of
the grating assembly. Mirrors with a dielectric reflective coating can be
made with better reflectivity than the mirrors used near 3000 A, the spectral
region of interest. The reflective coating that is more efficient in the uv
was not used because its very low reflectivity in the visible makes it diffi-
cult to align the optics with visible light.
Short turn-over times for samples in the sample cell can be achieved by
flushing the gas in a low-turbulent stream perpendicular to the monitor beam.
If the sample cell is shaped properly and the mirrors and windows are mounted
properly, the noise produced on the detector signal by the turbulence is not
-------
significant. Little or no error in sampling is produced by adsorption of SC>2
on the sample cell surfaces because they are coated with non-reactive Kel-F.
Teflon coatings adsorbed S02 more readily than Kel-F. Anodized aluminum sur-
faces should not be exposed to S02 samples because of the very rapid adsorption.
-------
SECTION 3
RECOMMENDATIONS
The instrument described herein has been designed and built to meet a
combination of several desirable specifications as a monitor for S02 in mobile
sources. Some of the important features are: relatively small size, short
response time, variable sample flow rate, very good discrimination, good
accuracy, low minimum detectable concentration, and wide dynamic range.
Specifications for some of these features necessarily have been sacrificed
to improve the specifications of other features. Thus, any S0£ monitor to
be patterned after this one undoubtedly should be modified according to the
specific types of measurements to be made with it. A few design changes have
become apparent as a result of assembling and testing the present instrument,
but even more can be learned by using it for a few weeks under normal testing
conditions.
Without changing the basic design, a few relatively minor changes could
be made to improve the signal-to-noise ratio by increasing the amount of
ultraviolet energy reaching the detector. The biggest gains can be made by
employing mirrors with a dielectric reflective coating that has a much higher
reflectivity than the uv-enhanced aluminum presently being used. This change,
however, would complicate the visual optical alignment because the dielectric
mirrors reflect very poorly in the visible. This disadvantage could be over-
come by employing mirror mounts that make it possible to remove a mirror and
replace it with one of the same radius of curvature without losing optical
alignment. Mirror mounts could be adjusted with mirrors having good visible
reflectivity; the mirrors then could be replaced with ones having good uv
reflectivity. Some of the flat mirrors should be replaced by quartz prisms
that are oriented to produce total internal reflection, which can be quite
efficient. The prisms should be anti-reflection coated on the surfaces
through which the beam passes.
By enlarging the cross-sectional area of the sample cell, the multiple-
pass mirrors can be enlarged, thus allowing more energy to be transmitted.
The sample cell could be lengthened to increase the electrical signal produced
by a given S02 concentration. Either of these two changes necessarily in-
creases the sampling volume, which increases either the necessary flow rate
or the sample turn-over time, or both. A grating assembly with longer focal
length mirrors and/or a larger grating might be required to take full advan-
tage of larger mirrors in the sample cell.
Some gain in signal-to-noise ratio and instability can probably be
achieved by replacing the side-looking photomultiplier with an end-looking
-------
one that has a higher quantum efficiency. Stability could also probably be
improved by employing a lens between the exit grid and the photomultiplier so
that the grating is focused on the photocathode. This would reduce some of
the effect of the inherent non-uniform sensitivity of side-looking photo-
cathodes.
The size and complexity of the instrument can be decreased if the speci-
fications are relaxed on either the minimum detectable concentration, the
discrimination or the sample turn-over time. As an example, the grating
assembly with the grid that produces relatively high spectral resolution
could be replaced with a pair of uv interference filters. One filter would
transmit a spectral region that includes most of the strong absorption by
the entire S(>2 band; the other would transmit an adjacent spectral interval
in which the S02 absorption is weak. By alternating the filters in the
monitoring beam and comparing the transmitted energy levels, the S02 con-
centration could be determined. This simplified method makes use of the
absorption by most of the absorption band, but it does not make use of the
structure in the band (see Figures 9 and 10), with the result that the
discrimination is reduced and the zero-setting of the instrument is more
subject to drifts caused by variations in source intensity or by changes in
the temperatures of the filters or other optical components.
Gas-filter cells used in conjunction with an interference filter should
also be considered as a simpler method. However, gas-filter correlation
techniques are not expected to provide discrimination and stability as good
as that provided by the grating assembly in the present instrument.
The flow through the sample cell is not as free of mixing as was expected
and should be investigated further to determine ways of decreasing the turn-
over time. Different shapes for the inside of the sample cell should be
tested at typical sample flow rates. The gas flow should be kept parallel to
the faces of the multiple-pass mirrors (i.e. perpendicular to the monitoring
beam) if high flow rates are required. This avoids excessive turbulence that
may be produced by gas flowing around the mirrors.
-------
SECTION 4
LAYOUT AND OPTICAL DIAGRAM
Figure 1 illustrates the pump assembly, the manifold assembly, and the
optics assembly. The pump assembly is connected to the manifold assembly
with stainless steel tubing, and the manifold is connected to the inlet and
outlet of the aluminum sample cell with the same type of tubing. The uv source
power supply is also shown in Figure 1. Compressed gas enters the manifold
after particulates are removed by a filter mounted on the pump assembly. A
wide range of sample gas flow rates is obtainable with the two pumps mounted
on the pump assembly and the flow-meters and controls available on the mani-
fold assembly. The gas pressure is indicated at key points in the flow path
by pressure gauges. After leaving the manifold, the gas flows into the sample
cell and through the sampling volume where the monitoring beam is perpendic-
ular to the direction of the gas flow. The gas absorbs ultraviolet radiant
energy from the monitoring beam and then exits from the sample cell. The
sample gas is then exhausted through the manifold or passed on to other
instruments for further analysis.
Figure 2 is a front view of the optics assembly with the front hinged
panels closed and with the available components and controls labeled.
Figure 3 is a front view of the optics assembly with the front hinged panels
opened and with components and controls labeled appropriately. These figures
are discussed in more detail in this section.
Figure 4 is an optical diagram of the S02 monitor. Ultraviolet energy
from the deuterium arc source is focused by lens LI onto the field lens L2,
which serves as the entrance window of the sample cell. The optical diagram
within the sample cell is too complex to illustrate completely for 12 passes
of the cell. The distance between mirror C2 and either mirror Cl or C3 is
42.7 cm, giving a 512 cm path for 12 passes. The number of passes can be
changed by integral multiples of 4 by use of the adjustment screws located
behind mirrors Cl and C3. Mirrors Cl, C2, and C3 are bonded to stainless
steel back-up plates with the appropriate holes and grooves required for the
adjustments.
After passing through the multiple-pass optical system in the sample
cell, the beam emerges through the exit window, lens L3. The beam is then
incident on flat mirror XI, which is mounted 45° from the vertical. Mirror XI
rotates the beam 90° and directs it downward to flat mirror X2. Mirror X2
is mounted 45° from the vertical, and this mirror rotates the beam 90° and
directs it toward spherical mirror X3. Before entering the sample cell, the
central ray of the beam is 15.0 cm above the baseplate of the optics assembly.
-------
Figure 1. Photograph of the optics assembly, the manifold assembly,
the pump assembly, and the UV source power supply.
-------
TIME CONSTANT SWITCH
c
ANALOG
PANEL
METER
DIGITAL
PANEL
METER
DIGITAL OUTPUT
JACK
FIXED ANALOG
OUTPUT CONNECTOR
UV SOURCE POWER
CONNECTOR
UV SOURCE
POWER SWITCH
VARIABLE ANALOG
OUTPUT CONNECTOR
ZERO BALANCE SWITCH
AND POTENTIOMETER
SAMPLE CELL
PRESSURE GAUGE
LINE CORD
CONNECTOR
Figure 2. Photograph of the optics assembly with front hinged jpaneIs closed.
-------
CALIBRATION CELL
LIFTING ROD
INDIVIDUAL
RANGE POTENTIOMETERS
INDIVIDUAL
DIGITAL METER
SPAN POTENTIOMETERS
GRATING
ASSEMBLY
TEMPERATURE
CONTROLLER
± 15 VDC
POWER SUPPLY
MOTOR TRANSFORMER
HIGH VOLTAGE
PM POWER SUPPLY
HV ADJ
POTENTIOMETER
REFERENCE CARD
METER FUNCTION
SWITCH
MASTER SPAN
POTENTIOMETER
RANGE CARD
MAIN ELECTRONICS CARD
SIGNAL
INPUT
POTENTIOMETER
Figure 3. Photograph of the optics assembly with front hinged
panels opened.
-------
r
Cl (")
ARC
SAMi'I I. Cl 1.1. -
SAMJM.I on n;orrn-:)
J
1-
-- PI i r %
~~ ~~ \
\==^:~~~~^
•UAT r v, ~- Q
f * "H
•-K),0'.
1 30 r:x-
1KAT1NC ASSEMBLE-
GRID
1— SPINNING
:.--+_ T RETICLE
Figure 4. Optical diagram of the SC>2 monitor.
-------
Upon emerging from the sample cell, the central ray is 11.9 cm above the base-
plate. Mirrors XI and X2 lower the beam 5.1 cm; thus the beam is 6.8 cm above
the baseplate of the optics assembly after reflection from mirror X2. Mirror
X3 forms an image of the source on the 0.3 mm wide by 2.3 mm high entrance
slit of the grating assembly.
Spherical mirror Gl collimates the beam of energy after it has passed
through the entrance slit. The combination of concave mirror G2 (f = 61 cm)
and convex mirror G3 (f = -53.2 cm) focus the radiant energy dispersed by the
grating onto the grid. The dispersion at the focal plane where the grid is
located is approximately three times as great as it would be if mirrors G2
and G3 were replaced by a single concave mirror identical to mirror Gl
(f = 40.6 cm). This increased dispersion makes it possible to use a grid
with wider spacings between the openings. In the optical diagram of Figure 4,
energy of only a single wavelength is represented in the path from the
grating to mirror G2 to the grid. Energy of a wavelength slightly different
from the wavelength represented is diffracted from the grating at a slightly
different angle and is therefore brought to a focus beside the point of focus
of the represented wavelength. The grid is a thin piece of stainless steel
shim stock with five slits cut in it. Two 1.5 mm wide by 4.6 mm high slits
pass the two B spectral intervals in the S02 absorption band where weak
absorption occurs. The three 1 mm wide by 4.6 mm high slits pass the three A
spectral intervals in the S02 absorption band where relatively strong
absorption occurs. After passing through the slits, the radiant energy is
chopped at 360 Hz>and energy from either the three A intervals or the two B
intervals is incident on the CsSb cathode of the photomultiplier, which has a
modified S-5 spectral response. The side-looking photomultiplier is sup-
ported by a mount attached to the grating assembly and extends into the
cylindrical chopper. The photomultiplier is centered inside the cylinder, and
the 24 mm wide x 8 mm high photocathode is perpendicular to the beam of
radiant energy. A reticle that functions both as a chopper and as a 30 Hz
alternator is bonded to the cylindrical chopper. The reticle chops the incom-
ing signal at 360 Hz and also alternately passes the three A intervals, then
the two B intervals at 30 Hz. The reticle slits are aligned with the grid
slits and are oversized so that this spinning component will not contribute
significantly to instabilities, noise, or zero drifts.
The photomultiplier dynode chain has a ten-stage linear configuration.
The signal from the photomultiplier is processed electronically to produce
a dc output signal proportional to the concentration of S02 sample gas. The
signal-processing electronics are described in detail in Section 8.
A calibration cell that is filled with a mixture of S02 and N2 can be
moved easily into the beam between mirror Nl and the entrance window to the
sample cell. The calibration cell is not shown in Figure 4. The optical
pathlength through the calibration cell is approximately 4 mm. The cell can
be filled to any desired pressure up to 1 atmosphere. Valves and fittings
fastened permanently to the calibration cell make it convenient to change
the amount of gas in it without removing it from the instrument.
Also not shown in Figure 4, but included with the instrument, is a
permanently mounted tungsten-iodide bulb. The tungsten-iodide bulb is mounted
12
-------
below the beam between lens LI and mirror Nl and provides visible light to
align the optical components in the sample cell and in the grating assembly.
The deuterium arc source emits very little visible energy, making it difficult
to align optics, particularly after the beam has been reflected a few times.
A flat mirror, N2, and a lens, L5, are placed directly above the bulb and
in the path of the ultraviolet energy between lens LI and mirror Nl. Thus,
from the movable mirror, the beam of visible energy traverses exactly the
same path as the ultraviolet energy from the arc source.
A small window on the end of the sample cell opposite the entrance and
exit windows makes it possible to view images formed on mirror C2 of the
multiple-pass system to determine the number of passes of the monitoring beam.
A prism is mounted over the viewing window for viewing the images on mirror
C2 from the rear of the optics assembly. The number of passes can be adjusted
from outside the sample cell without opening the cell or disturbing the
sample.
Figure 2 is a front view of the optics assembly. A digital and an analog
panel meter are mounted on the large, front hinged panel. The analog panel
meter displays either the dc output signal, the input signal, the reference
signal, or the carrier level, depending on the position of the meter function
switch mounted on the main electronics card. The dc output signal is always
displayed on the digital panel meter regardless of the position of the meter
function switch. The analog dc output signal is monitored by connecting a
strip-chart recorder to either the fixed or the variable output connector
mounted on the lower panel. The output of the fixed output connector is
10 volts for a full-scale reading. The full-scale output of the variable
output connector can be adjusted from about 1 millivolt to 10 volts by
adjusting a potentiometer and the recorder gain switch on the main electronics
card.
The range switch mounted on the front hinged panel is used to select
either the 5, 10, 20, 50, or 100 ppm range. When the range switch is in
the off position, the dc output of the instrument is zero. The time constant
switch is also mounted on the front hinged panel; time constants of 0.1, 0.3,
1.0, 2.0, and 5.0 seconds are available. A 10-turn zero-balance potentiometer
and a zero switch are used in combination to electronically cancel any
residual 30 Hz modulation on the carrier when S02 is not present in the sample
gas.
AC power is supplied to the electrical components by connecting a line
cord to the "AC IN" male connector and by switching the power switch to the
on position. When the power switch is in either the standby or the on
position, power is supplied to the heaters bonded to the sample cell and to
the power resistors mounted on the sample cell inlet and outlet fittings.
After the cell has stabilized at 55°C, the cell can be maintained at
this temperature with the power switch in the standby position while other
electrical components are inoperative. This feature allows the cell to be
maintained at 55°C when tests are not being performed and thus the operator
does not have to wait 2 to 3 hours for the cell temperature to stabilize each
time before data are taken. The source power switch located on the lower front
panel is used to turn the UV source fan on after the UV source power supply
13
-------
is connected to the source power outlet mounted on the lower front panel.
This outlet was installed as a safety feature to prevent the source from being
turned on when the cooling fan is not operating.
Figure 3 is a front view of the optics assembly with the front hinged
panels open. The main electronics card and the range card are mounted on
the inside surface of the front hinged panel. The reference card is mounted
on the inside surface of the right side panel. The high-voltage power supply
for the photomultiplier, the chopper motor transformer, the +15 Vdc power
supply, and the temperature controller are mounted on the power supply chassis.
The chopper motor is mounted on the stainless steel grating assembly and is
located below the power supply chassis. The bottom of the sample cell is
secured to the baseplate of the optics assembly with two screws. The cell is
also secured to the partition that separates the grating assembly from the
sample cell. Figure 4 is a top view of the optics assembly showing the loca-
tions of the grating assembly, the sample cell, the chopper motor, and the
photomultiplier.
Both the sample cell and the grating assembly rest on spacers that hold
these two components approximately 1.5 cm above the base plate. The spacers
under the sample cell provide insulation to minimize the flow of heat from
the sample cell to the base plate. The three spacers under the grating assem-
bly are arranged in such a way as to not put any stress on the grating assem-
bly if the main base plate is distorted slightly.
14
-------
SECTION 5
SAMPLE CELL
REQUIREMENTS AND GENERAL FEATURES
The two specifications for the instrument that are important factors in
the design of the sample cell are the high sensitivity and the quick turn-
over of gas within the sampling volume. The gas flow through the sampling
section may be varied, depending on the type of study being performed. If the
concentration of SC>2 in the gas entering the gas line at maximum flow rate
changes suddenly from one concentration level A to a second concentration
level B, it is desired that the concentration within the sampling volume will
have changed 90% of the way from level A to level B within 0.5 seconds after
the concentration starts to change within the sampling volume. This excep-
tionally good time resolution requires a gas flow system with a minimum
amount of mixing. The flow rate must be sufficiently high to displace all of
the gas in the sampling, volume quickly enough to allow for a certain amount
of mixing that can not be avoided. Turbulence produced by the high flow rate
must not produce significant noise on the signal from the photomultiplier
used to detect the monitoring beam.
The second difficult specification calls for the signal-to-noise ratio,
S/N to be> 2 for an S02 concentration of 0.1 ppm. This high sensitivity
requires that the monitoring beam pass through at least a few meters of
sample gas. Thus, this imposes certain minimum limits on the size of the
sampling volume. This limitation, in turn, requires a relatively high flow
rate of the sample gas in order to displace the gas within the sample volume
in the required time. The long path of the monitoring beam is obtained by
employing a multiple-pass optical system in the sample cell. The sampling
volume is made as small as practical while maintaining the mirrors in the
multiple-pass optical system large enough to pass an adequate amount of
ultraviolet energy through the monitoring system.
15
-------
CELL CONSTRUCTION AND GAS FLOW
A cross-sectional view of the sample cell is shown in Figure 5. The
sample cell is 42.7 cm long in the dimension perpendicular to the plane of
the section illustrated. The gas sample enters one end of the distribution
tube illustrated in the upper portion of Figure 5. The 1.59 cm I.D. stain-
less steel distribution tube contains 33 holes evenly spaced in a row in the
positions indicated. After leaving the distribution tube through these holes,
the gas flows around the distribution tube, through the 0.6 mm throat, and
through the cavity in the sample cell perpendicular to the monitoring beam.
The cavity in the sample cell is shaped to produce a minimum amount of mixing
of the gas before it has passed through the volume traversed by the monitor-
ing beam. After passing the monitoring beam, the gas continues on through
another throat 2.3 mm high and enters the collection tube in the lower
portion of the sample cell. The gas leaves one end of the collection tube
and is either exhausted or passed on to other instruments for further
analysis. This type of gas flow produces very little turbulence in the
volume traversed by the monitoring beam. At a high gas-flow rate of 200
liters per minute, no additional noise on the detector signal attributable
to turbulence can be observed.
The function of the distribution tube is to distribute the gas flow
uniformly along the length of the sample cell. Ideally, portions of a small
volume of gas that enter the end of the distribution tube at a given time
should pass through the holes near one end at the same time that other
portions of the same volume element pass through the holes near the middle or
near the other end. The mixing and the dwell-time in the distribution tube
are reduced to negligible amounts by a tapered piece of rod placed inside
the distribution tube, as illustrated in Figure 5. The O.D. of the rod is
approximately 0.1 mm less than the I.D. of the distribution tube; the rod is
tapered so that the cross-sectional area of the gas volume in the distribu-
tion tube decreases in the direction of gas flow.
A few tests were performed with smoke to determine the efficiency of the
distribution tube and the mixing inside the sample cell. The ends of the
sample cell were replaced with transparent plexiglass so that a laser beam
could be passed through the length of the cell at different locations. Light
scattered by smoke within the sample cell was observed through-the plexiglass
end plates. While air was being pumped through the sample cell at a typical
flow rate, smoke was injected into the stream of flowing gas just ahead of
the inlet to the distribution tube. The smoke could be started or stopped
abruptly. As a test of the distribution tube, the laser beam was directed
along the cell approximately 1 cm below the throat near the tube, and the air
flow was adjusted to between 100 and 150 liters per minute. When the smoke
flow, was stopped abruptly, the smoke disappeared from the laser beam in less
than 0.1 sees, with no distinguishable difference between the times it dis-
appeared from each end. This result indicates that the distribution tube per-
forms its function quite well.
The holes in the collection tube are larger than those in the distribu-
tion tube in order to reduce the pressure drop and maintain the absolute
pressure in the sampling volume only slightly above the outside air pressure.
16
-------
DISTRIBUTION TUBE
TAPERED ROD
URETHANE INSULATION (2.5 em THICK)—
33 HOLES (1.5 mn DIA) EVEKLY SPACED
DISTRIBUTION TUBE (1.6 ca I.D.)
— GAS INLET
THROAT 0 .6 mm
— URETHANE INSULATION (1.3 cm THICK)
FIN
VOLUME TRAVERSED
BY MONITORING BEAM
STAINLESS STEEL PLATES (1.2 nm THICK)
1.5 cm
THROAT 2.3
COLLECTION TUBE (2.2 cm I.D.)
GAS OUTLET
HOLES (2.6 mm dla) EVENLY SPACED
Figure 5. Sectional view of the sample cell. The left-hand portion shows
the tapered rod in the distribution tube. The main body of the cell is
approximately 31 cm high and 43 cm in the direction parallel to the
monitoring beam.
17
-------
The spacings between the holes were varied along the length of the collection
tube so that the gas flow is nearly uniform throughout the length of the cell.
If the holes were spaced uniformly, more gas would move across the end of the
cell from which the gas leaves the collection tube than across the opposite
end. The proper spacings of the holes were determined experimentally by
observing light scattered by smoke from a laser beam located between the
sampling volume and the 2.3 mm throat. Various combinations Of holes in the
collection tube were covered and the smoke observed. The distribution of the
open holes used in the final design produced a flow-pattern so that the
smoke disappeared from both ends of the cell at very nearly the same time
after the smoke injected into the airstream had been stopped.
The fin placed just upstream from the monitoring beam divides the gas
stream into two equal parts. This fin was not included in the original
design, but the flow tests performed with smoke indicated that there was more
turbulence and mixing within the cell than had been anticipated. Addition of
the fin reduced the apparent mixing and therefore reduced the gas turn-over
time.
The main body of the sample cell is made from two nearly symmetrical
pieces of aluminum that have been milled out and placed together as indicated
in Figure 5. The two pieces are held together rigidly by 12 bolts. Silicone
rubber provides a seal at the junction of the two main side pieces and at the
junctions of these pieces with the end plates. A vacuum-tight seal is not
required because the gas sample is flushed continuously and the pressure is
very near atmospheric. The pressure in the sample cell is slightly above
atmospheric when the gas is pumped through in the normal manner as discussed
in Section 6. If the gas is drawn through by a pump on the downstream side of
the sample cell, the pressure is slightly below atmospheric. The difference
between the sample pressure and atmospheric pressure depends on the flow rate
and on the amount of constriction by the valves, filter, etc. between the
sample cell and the atmosphere. The end plates for the sample cell, not shown
in the figures, support the multiple-pass optical system and the windows
through which the monitoring beam enters and leaves the sample cell. Gas fit-
tings also mounted to the end plates connect to the distribution tube and the
collection tube so that these two tubes can be removed easily and cleaned, if
necessary.
The aluminum body of the cell is anodized to prevent corrosion. Sensi-
tivity tests performed with the cell when it was first assembled indicated
that the anodized surface adsorbed SC>2 from dilute mixtures at such a rapid
rate that it would produce erroneous readings unless the gas flow was excep-
tionally high. In addition, some of the 862 that had been adsorbed previously
would desorb from the walls and contaminate "zero-gases" or mixtures of lower
concentration as they were flushed through the cell. After testing several
different coatings for the anodized aluminim to reduce adsorption while main-
taining good resistance to corrosion, we coated the inside surface with
KEL-F 800. The solid material was dissolved in a solvent and sprayed on the
anodized surface and heated for a total of about 5 hours at different tempera-
tures up to 425°F. Tests indicate that the rate of adsorption of S02 on this
material is slow enough that it can be accounted for by slowly flushing the
sample through the cell.
18
-------
The two thin stainless steel plates on either side of the volume tra-
versed by the monitoring beam were added after the aluminum sides were made
and some preliminary tests were performed. These plates reduce the width of
the gas stream, decreasing the amount of mixing and, in turn, reducing the
gas turnover time. The surfaces of these plates were roughened by sand-
blasting to reduce the specular reflection of light from the surfaces on
which "spill-over" of the monitoring beam is incident. Extra light from the
oversize monitoring beam is incident on these surfaces at a grazing angle,
causing the reflection to be quite efficient. The reflected light is a
possible source of error because it does not travel the same distance in the
sample as does the main beam. In addition, this reflected light makes align-
ment of the multiple-pass optics confusing. No Kel-F coating was put on
these stainless steel plates because it would form a glossy surface and
reduce the effect of sandblasting. No coating on stainless steel is required
because the adsorption of S02 on this metal is negligible.
CELL HEATING
Condensation of water in the sample cell is avoided by heating the cell
to any desired temperature up to approximately 60°C. Electrical heating
tapes, not shown in Figure 5, are bonded to the sides of the cell and provide
the heat. Heating wires attached to the inlet and outlet gas lines provide
additional heat to these parts. A temperature sensor embedded in the side
of the cell near the monitoring beam provides the input signal to a control-
ler that varies the power to the heater as required to maintain a constant
temperature. The temperature to which the cell is regulated is varied by
adjusting a small potentiometer on the temperature controller. When the
instrument was delivered to EPA, the temperature controller was adjusted to
approximately 55°C, which is high enough to avoid condensation under most
operating conditions even if some parts of the cell are a few degrees cooler
than the part near the sensor. A transformer in the heater circuit limits the
maximum voltage across the heaters to 36 volts. During typical operation, the
duty cycle of the heater is between 20% and 40%. Approximately 2 hours are
required to bring the cell from room temperature to 55°C.
Two thermocouples make it convenient to monitor temperatures with a high
input-impedance voltmeter. One of the thermocouples is embedded in the
aluminum side of the cell adjacent to the sensor for the controller. This
makes it possible to determine the control temperature and to monitor the
performance of the controller. The other thermocouple extends through the
side of the cell wall into the gas at a point just downstream from the moni-
toring beam. This thermocouple monitors the gas temperature, which may
fluctuate and differ by a few degrees centigrade from the cell walls. The two
thermocouples are made of iron and constantan wires; the emf produced is
0.052 mv for each °C difference between the sensing junction and the refer-
ence junction of the thermocouple.
19
-------
Sheets of urethane insulation approximately 1.5 cm thick insulate the
main cell body. Four stainless steel posts, each approximately 1.2 cm long
and 1.2 cm in diameter support the cell body and insulate it thermally from
the remainder of the optics assembly. A solid aluminum partition (see
Figure 4) separates the sample cell from the grating assembly and the elec-
tronics. During typical operation, the grating assembly and the electronics
are less than 5°C above room temperature.
OPTICAL COMPONENTS OF SAMPLE CELL
Figure 6 illustrates the optical portions of the sample cell. A side
view of the multiple-pass optical system is shown in the upper panel.
Mirrors Cl, C2, and C3 are nearly flush with the inside surfaces of the end-
plates so that they interfere a minimum amount with the gas flow. An image
of the deuterium arc source is formed beside mirror C2 and just inside
lens L2, which also serves as a window. Spherical mirrors Cl, C2, and C3
have a common radius of curvature that is equal to the distance from mirror
C2 to Cl or C3. Images are formed on mirror C2 as indicated in the lower
panel of Figure 6 when the optics are adjusted for 12 passes. The image
pattern is viewed through the window below mirror C3. The number of passes
is changed by adjusting mirror C3. When adjusting the optics, the tungsten-
iodide bulb is used as an energy source because it emits much more visible
energy than the arc. The centers of lenses LI and L2 are separated by
approximately 30 mm, and the openings in the endplates near these lenses are
approximately 5 mm in the long dimension. Mirrors Cl and C2 are each 16 mm
square. Thus the cross section of the volume of gas being monitored is
approximately 16 mm by 32 mm.
The oblong opening in the endplate near lens L2 is overfilled by the
enlarged image of the deuterium arc source. Therefore, the images on mirror
C2 have the shape of the oblong opening in the endplate. Lens L2 forms an
image of lens LI near mirror Cl. Lens L3 images mirror C3 near spherical
mirror X3 (see Figure 4). Flat mirrors XI and X2 are each tilted approxi-
mately 45° from the vertical. The combination of these two mirrors lowers
the beam and directs it horizontally to mirror X3, which forms an image on
the entrance slit of the grating assembly. The image formed at lens L3 is
rotated 90° by mirrors XI and X3 with the result that the long dimension of
the image at the entrance slit is parallel to the length of the slit. The
entrance slit, 0.3 mm wide by 1.5 mm high, is over-filled in width and
slightly under-filled in height..
20
-------
I END FLUX OF SAMPLE CELL
END PLATE OF SAMPLE CELL
MONITORING BUM
MOUNT FOR MIRROR c2
MIRROR XI
LENS L2
SILICONS RUBBER SEAL
MIRROR ci-
— MIRROR 02
MIRROR 03-
SILICORI ROBBER SEAL
LEHSL3
10
MIRROR C2-
CD CD-
ENTRANCE IMtCE
CD CD
CD CD
CD «
MODNTING POST FOR MIRROR Cl
LEVER ARM FOR MIRROR Cl
LETER ARM FOR MIRROR C3
MOWTISG POST FOR MIRROR C3
WINDCU
-PRISM
EXIT IMftGE
Figure 6. Multiple-pass optical system. The
cross-sectional views illustrate the
manner in which the optical compo-
nents are assembled. The image
pattern on mirror C2 is shown in the
lower panel.
21
-------
SECTION 6
PUMP AND MANIFOLD ASSEMBLIES
The pump assembly and the manifold assembly are designed to provide the
user with a wide variety of sample flow conditions. Sample flow rates can
be varied easily from less than 5 1pm to more than 160 1pm. Flow rates are
adjusted with the manifold assembly and by the choice of one of two pumps.
During normal operation, sample gas is drawn into the system to a pump that
forces it through the manifold system and sample cell. The sample cell
pressure is slightly more than 1 atm. By re-arranging some of the gas-
handling components, the pumps can be made to draw the sample gas through the
sample cell before the gas passes through the pump or the manifold assembly.
This latter procedure may be useful if it is found that the pump or the mani-
fold assembly changes the concentration of S02- All of the following discus-
sion pertains to normal operation during which the sample gas passes through
the pump before it reaches the sample cell.
PUMP ASSEMBLY
The pump assembly with two pumps mounted side-by-side is represented
schematically in Figure 7. Sample flow rates as high as approximately 160 1pm
can be achieved with the high-capacity pump if all of the gas lines are short
and have sufficiently large diameters. Approximately 80 1pm flow rate can be
achieved with the smaller pump.
A three-way electric switch is mounted on the pump assembly so that
either the high-capacity pump or the low-capacity pump can be turned on. The
operator must check several valve positions and settings on both the pump
assembly and the manifold assembly before turning on either pump. Two three-
way ball valves mounted on the plexiglass front of the pump assembly are
used to switch the gas lines. If the high-capacity pump is to be used, these
two three-way valves must be rotated to the high-flow position so the gas can
pass from the pump to the manifold assembly. Valves on the manifold assembly
must also be opened to pass the sample gas. Similarly, the valves must be
rotated to the low-flow position before the low-capacity pump is turned on.
The three-way valves and the electrical switch for the pumps are color-coded.
Connected to the line leaving each pump are a pressure gauge, a pressure
switch and a pressure-relief valve. The pressure gauge is useful in monitor-
ing the output pressure of the pump which must be kept below 10 psig to avoid
damaging the pump. The adjustable pressure switch automatically turns off the
electrical power to the pump when the pressure exceeds approximately 9 psig.
The pressure-relief valve opens at about the same pressure and serves as an
additional safety feature.
22
-------
Compression of the sample gas by the pumps heats the gas, making addi-
tional heating of the outlet lines unnecessary. The amount of heating
increases with increasing pressure in the outlet line. When the high-capacity
pump is operating with 8 psig pressure, the outlet line may reach a tempera-
ture as high as 85°C.
MANIFOLD ASSEMBLY
The manifold assembly, shown schematically in Figure 8, makes it possible
to adjust the rate of flow of the sample gas to the desired level. Low-flow
rates are read from a small flow meter with a full-scale reading of 80 stan-
dard cfh (38 1pm). A larger flow meter with a full-scale reading of 6 cfm
(170 1pm) measures higher flow rates. A 3-way ball valve is used to direct
the gas through the appropriate meter. The drop in pressure from the outlet
of the manifold to the sample cell can be determined by comparing the pres-
sures indicated by the gauges at these two locations.
Both the flow-adjust valve and the bypass valve are used simultaneously
to adjust flow rates. Opening the bypass valve prevents the pressure from
building up to a high level when the flow-adjust valve is adjusted for a low
flow rate. Approximate pump pressures can be determined easily from the
gauge near the inlet to the manifold.
Calibration gas is flushed through the manifold to the sample cell by
opening a toggle valve in the calibration-gas line. A standard fitting on the
end of the calibration-gas line makes it easy to connect it to a container of
gas mixture with a known concentration of S02. The flow-adjust valve is
closed when calibration gas is introduced into the sample cell.
The flow meters, gauges, valves and gas lines on the manifold assembly
are heated to approximately 55°C to avoid condensation of t^O. A 36 -volt
transformer provides electrical power, at a safe voltage, for the heating wire
wrapped around all of the components that are in direct contact with the
sample gas. A contact switch attached to the gas line opens, stopping the
electric current, when the temperature reaches approximately 55°C. A small
light indicates when electrical power is being supplied to the heating wire.
The switch-point can be adjusted to other temperatures. All of the heated
components are packed in insulation and enclosed in a metal box. Handles for
the valves extend through the front panel of the box. Glass windows on the
front panel make it possible to see the scales on the gauges and flow meters.
23
-------
TO MANIFOLD ASSEMBLY
->~
H
3-WAY BALL VALVE
PRESSURE
GAUGE
i
(3 ujn PORE SIZE)
A
IN ~> • FILTER
RELIEF
VALVE
PRESSURE
SWITCH
RELIEF
VALVE
PRESSURE
GAUGE
3-WAY BALL VALVE
ll
PRESSURED!
SWITCH !
/ HIGH
I FLOW
PUMP
/ LOW
FLOW
V PUMP
Figure 7. Flow diagram of pump assembly.
24
-------
TO S02 MONITOR-<~~
/ LOW FLOW METER
HIGH FLOW METER \
CALIBRATION GAS
TOGGLE VALVE
FROM S02 MONITOR
DISCHARGE - < ' <-
FLOW ADJUST VALVE
PUMP PRESSURE
A
PUMP
BYPASS
Figure 8. Flow diagram of manifold assembly.
25
-------
SECTION 7
SPECTROSCOPIC PRINCIPLES OF DETECTION
Figure 9 shows a spectrum of the ultraviolet S02 band that is employed
by the instrument. This band is used rather than the well-known 4 ym band,
which is much weaker and therefore requires longer optical paths in order
to achieve the required sensitivity. It is noted that the part of the band
shown in Figure 9 shows some relatively sharp structure, but little structure
appears on the short wavelength side of the bandj Each of these structural
features that stand out in the region near 3000 A consists of many over-
lapping lines. For samples near 1 atm pressure, these many lines are broad-
ened so that they overlap and there is little remaining structure that is not
resolved in the spectrum shown in Figure 9. At low sample pressures, many of
these individual lines could be resolved with a spectrometer having very high
resolving power.
In order to simplify the problem of handling the sample gas, it is
highly desirable that the sample be maintained near 1 atm pressure. There-
fore, the very fine structure in the spectrum cannot be exploited for de-
tection purposes. It is necessary to make use of the relatively coarse
structure that is observed in Figure 9. Figure 10 shows a portion of the
spectrum near 3000A with the wavelength scale expanded so that the lines can
be seen more clearly. The instrument makes use of this relatively coarse
structure by comparing the transmission through the gas in the 3 intervals
labeled with A to the transmission through the two intervals labeled with B.
The A intervals include very strong absorption by the S02, whereas the B
intervals contain weaker absorption. The two B intervals are the same width
and are each 50% wider than each of the A intervals so that the total width
of the spectral intervals passed by the B intervals is the same as that by
the A intervals.
The average transmissions through intervals A and B are compared by
use of the spinning reticle illustrated in Figure 11. After passing through
the grid, the energy of the selected wavelength intervals is incident on the
spinning reticle. The reticle is bonded around a cylindrical piece of alumi-
num tubing which has 12 equally spaced slots cut along its long axis and is
secured to the shaft of a motor, which rotates at 1800 rpm. The reticle has
6 rows of 3 slits and 6 rows of 2 slits cut in it. The three slits are
aligned directly behind the three grid openings that correspond to slit
intervals A. The two slits are aligned directly behind the grid openings that
pass intervals B. The openings in the reticle are oversized with respect to
the grid openings so that this spinning component will not contribute signif-
icantly to instabilities, noise, or zero drift in the output signal. The
26
-------
t
w
u
M
H
100% Transmittance
2600
2800 3000
WAVELENGTHS (Angstroms)
3200
Figure 9 . Spectrum of S02 from 2600 to 3300 Angstroms. The sample
consisted of 0.1 atm of pure S02 in a 0.4 cm cell.
27
-------
f
ta
I
CO
•n^Bnot
/
100% Transmittance
2950
3000
WAVELENGTH (Angstroms)
3050
Figure 10 . Partial spectrum of S02 band showing spectral slits A and B.
28
-------
to
VO
o
Reference Pick-Up
Motor Shaft >JI /
r°^^^^ P _ 30 Hz Tuned
BHI SJ fc Preamplifier l^llf T*
I }
: IXDlIlL Spinning
• ^QnHB Reticle
yu i I
/ l 1
Grid
i 3^6 Hz Gain Controlled
"] Amplifier Amplifier
|
Adjustable iO Hz 30 Hz Tuned
Time Constant * Synchronous * Signal
Amplifier Demodulator Amplifier
t
r v
Fixed Variable
Output Output
Adjustable
-* Phase
Shifter
Adjustable
Shifter
i r
nouu j. au ion 360 Hz
». caj-ancc w Demodulator
Assembly
AGC
Control
Amplifier
Calibrate Range
Pot Switch
' '
9^,,,.,,. — . ,.._ .„,„,
1 r
Digital
Output
Figure 11 . Block diagram of signal-processing electronics.
-------
energy passing through the openings in the spinning reticle travels radially
to the photocathode of the side-looking photomultiplier, which has an S-5
spectral response. The photomultiplier is mounted with its axis coincident
with the axis of the rotating chopper-alternator. As the reticle spins, the
beam is chopped at 360 Hz and the energy incident on the photomultiplier
alternates at 30 Hz between spectral intervals A and spectral intervals B.
The relative widths of the openings corresponding to spectral intervals A
and B is adjusted so that the same amount of energy is incident on the photo-
multiplier during both halves of the cycle of the spinning reticle as long as
there is no SC>2, or other absorbing gas, in the sample cell. In practice, it
is not possible to maintain exactly this equality between the two intervals,
but the slight imbalance is accounted for electronically by a method
described in Section 8. After some experimentation, it was determined that
the cathode of the photomultiplier has a non-uniform sensitivity and thus
the response of the photomultiplier to photons incident on different regions
of the cathode is not the same. The imbalance created by this non-uniform
sensitivity is also accounted for electronically by the method described in
Section 8.
A 30 Hz component of the detector signal results when S0£ is added to
the sample cell, and the size of this signal depends on the source bright-
ness, amount of dirt on the windows, and on any other parameters that might
change the total amount of energy incident on the slit of the grating assem-
bly. In order to avoid the problem of the detector output signal being
dependent upon these parameters, the instrument is made to measure a quantity
that is proportional to the fractional modulation of the beam, rather than to
the total change in energy striking the detector. This is accomplished by
chopping the beam at 360 Hz. When S02 is present in the sample cell, the
360 Hz carrier signal is modulated at 30 Hz. The carrier signal is main-
tained at a constant level by an automatic gain control circuit. The elec-
tronics described in Section 8 process the detector signal in such a way as
to produce a dc output that is proportional to (Tg - TA)/(TB + TA.) . This
quantity is proportional to the fractional modulation of the beam and is
related directly to the concentration of SC>2 in the sample cell of fixed
pathlength. The changes in source brightness, detector responsivity, and
electronic gain are automatically accounted for so that the "span" calibra-
tion remains constant.
30
-------
SECTION 8
ELECTRONIC PROCESSING OF DETECTOR SIGNAL
Figure 11 is a block diagram of the electronics that process the signal
from the detector and produce a dc voltage that is proportional to the con-
centration of S02- The relationship between the output voltage and the gas
concentration is determined by calibrating the instrument with samples of
known concentration. The electronics consist basically of two demodulators
in series. The first demodulator operates at 360 Hz, the carrier frequency
determined by the chopper. The average dc component in the output from the
first demodulator is proportional to the amount of radiant energy chopped by
the spinning reticle. The 30 Hz component of the signal output of the first
demodulator is proportional to the modulation of the beam at this frequency.
This 30 Hz component then passes through a series of gain controls and
amplifiers to the second demodulator, which is synchronous and receives its
reference signal from a reference pick-up mounted near the rotating grid
chopper. The output of the second demodulator is then proportional to the
30 Hz modulation, and thus to the concentration of S02 in the sample cell.
The amplified component of the detector signal at the carrier frequency,
360 Hz, is defined to be Vc. Correspondingly, the component at the modula-
tion frequency, 30 Hz, is defined as Va. It is desirable for the instrument
output to be proportional to the fractional modulation of the carrier signal.
This fractional modulation is proportional to Va/Vc, which we define as V,
and it is desired that this relationship be independent of source brightness,
detector sensitivity, etc. An automatic gain control (AGC) circuit maintains
Vc at a constant value. If, for example, the signal from the detector
decreases because of dirt on the windows or a decrease in source brightness,
the gain of the gain-controlled amplifier increases to maintain Vc constant.
The increase in gain also increases the amplification of both the 360 Hz
carrier signal and the 30 Hz modulation signal by the same factor. Therefore,
V is kept directly proportional to the output Va and maintains a constant
relationship with the gas concentration. The AGC circuit can account for
changes in the 360 Hz component of the signal by approximately a factor of 4.
The reference pick-up for the 30 Hz synchronous demodulator consists of
a light-emitting diode (LED) and a small phototransistor. The phototransistor
senses the light reflected from the metallic tape wrapped around the outside
of the spinning reticle.
A zero-balance assembly electronically accounts for any imbalance
between the two beams passing through spectral intervals A and spectral
intervals B when there is no absorbing gas in the sample cell. This
31
-------
electronic assembly would not be required if it were practical to obtain a
perfect optical balance between the two sets of slit intervals and if the
cathode of the photomultiplier had a uniform sensitivity. Approximately a
+ 10% variation in the transmittance through either of the sets of intervals
can be accounted for by this electronic zero-balance assembly. The assembly
changes the amplification ahead of the 360 Hz demodulator during the two
halves of the cycle of the spinning reticle. The signal from the 30 Hz
reference pick-up switches the zero-balance assembly between the two different
gains with the proper phase. The difference in the gains between the two
halves is adjustable by the "zero" potentiometer mounted on the front panel
of the instrument.
Five ranges of sensitivity are provided; 5, 10, 20, 50, and 100 ppm.
The gain in each range can be adjusted independently over a factor of
approximately 4. A six-position switch mounted on the front hinged panel of
the optics assembly makes it convenient to change ranges. When the switch is
in the off position, the output of the 360 Hz synchronous demodulator is
shorted to ground. Because of optical saturation, it is necessary to reduce
the number of passes to 2 or 4 passes in order to operate in the 0 - 1000 ppm
range. All ranges must be recalibrated with appropriate span gases for
operation in the 0-1000 ppm range.
A four-position switch mounted on the main electronics card makes it
possible to use the analog panel meter to monitor other quantities in addi-
tion to the signal output. With the switch in one position, the meter
indicates the level of the reference signal; with the switch in another
position, the meter reads the level of the 360 Hz signal from the photo-
multiplier before the automatic gain control circuit; with the switch in
still another position, the meter indicates the signal input level. Thus it
is possible to check to see that both the reference signal and the carrier
level are within the correct operating range.
The digital panel meter directly indicates the 862 concentration in the
sample gas. The relationship between the analog meter indication and the
concentration, of course, depends on the range to which the instrument is set.
The digital panel-meter is connected to an output jack so that the digital
output can be recorded or transmitted over a line to another instrument. The
size of the analog voltage output that corresponds to full-scale reading of
the instrument can be varied continuously from less than 1 mV to 10 V. Thus,
this analog output can be recorded easily with virtually any strip-chart
recorder with a high input impedance.
32
-------
SECTION 9
CALIBRATION AND PERFORMANCE
SPAN CALIBRATION
Each of the five sensitivity ranges (5, 10, 20, 50, and 100 ppm full-
scale) has been calibrated with the sample cell controlled at 55°C. The
calibration depends on the cell temperature because the instrument response
is approximately proportional to the number density of S02 molecules, re-
gardless of temperature. If the S02 concentration in ppm remains constant
and the pressure is one atm, an increase in temperature results in a pro-
portional decrease in the number density of S02 molecules in the monitoring
beam. Thus, the instrument response decreases and the potentiometers used
to vary the gains of the electronic circuits must be adjusted so the meters
will indicate the proper concentration. In most cases, a small correction
to the calibration for all five ranges can be made by adjusting a single
potentiometer that changes the gains for all five ranges by the same factor.
At a fixed temperature, the output signal is essentially proportional
to S02 concentration for concentrations less than 30 ppm. At higher con-
centrations, the response becomes slightly non-linear, necessitating small
corrections. The meters and the voltages on the output jacks are directly
proportional to the output signal; no circuits have been added to account
for the non-linearity. Table 1 contains the information required to correct
for the non-linear response when using the 50 ppm and 100 ppm full-scale
ranges. No correction is required for the 5, 10, and 20 ppm full-scale
ranges.
33
-------
TABLE 1. CORRECTIONS TO CALIBRATION TO
ACCOUNT FOR NON-LINEAR OUTPUT
Meter Reading Correction Corrected Reading
(ppm) (%) (ppm)
50 ppm Full-Scale Range
^30 0 Same as meter
40 +1 40.4
50 +2 51.0
100 ppm Full-Scale Range
§ 30 -4 0.96 x meter reading
40 -3 38.8
50 -2 49.0
60 -1 59.4
70 0 70.0
80 +1 80.8
90 +2 91.8
100 +3 103.
Errors due to the non-linear relationship between SC>2 concentration and
output signal are negligible for the 5, 10 and 20 ppm full-scale ranges, and
for concentrations lower than 30 ppm measured on the 50 ppm full-scale range.
The approximate corrections required for the 50 ppm and 100 ppm full-scale
ranges are given above. The gain for the 100 ppm full-scale range has been
adjusted so that it is nearly correct at 70 ppm. Thus the maximum error due
to non-linearity on the 100 ppm full-scale range is approximately 3% for any
sample that would normally be measured on this range.
The same corrections apply whether the values are read from the analog
panel meter, the digital panel meter, or one of the output jacks.
34
-------
ADSORPTION OF S02 ON CELL WALLS
Sulfur dioxide adsorbs readily to the walls of containers. The adsorp-
tion process may take place for several minutes before the amount adsorbed on
the walls comes to an equilibrium. The equilibrium amount depends on the
concentration of the S02 vapor in contact with the walls. The S02 monitor
senses only the concentration of S02 in the vapor phase; therefore sizeable
errors can be made if adsorption is ignored or not accounted for properly.
In order to minimize the problem due to adsorption, we have coated the inside
surface of the sample cell with Kel-F 800, an inert plastic material that is
noted for low adsorptivity. Nevertheless, some care in sampling is still
required because of adsorption in the gas lines and the small amount that
occurs in the sample cell.
The simplest method of accounting for the adsorption is to flush the
gas sample continuously through the sample cell so that the S02 that adsorbs
on the walls is replenished before the vapor concentration is depleted
significantly. "Zero" gas being used when adjusting the zero-setting should
also be flushed to carry away S02 that comes off the cell wall after having
been adsorbed from previous samples.
A few simple tests were performed to determine the rates of adsorption
and the rate of sample flow required to avoid significant errors in the
measurements. The first test involved introducing a sample of 3.9 ppm S02
concentration into the sample cell after it had been flushed with N2 long
enough that there was essentially no S02 adsorbed on the walls. The 3.9 ppm
mixture was flushed rapidly through the cell for 30 sec, a period long enough
for the S02 vapor concentration in the cell to reach its maximum value. The
gas flow was then stopped abruptly, and the S02 vapor concentration was moni-
tored for a few minutes. At one minute after the flow was stopped, the vapor
concentration had decreased because of adsorption to 92% of its original
value. The concentration continued decreasing and was down to 75% of the
original value after 5 minutes. A similar test was performed with a 76 ppm
mixture; the rate of percentage decrease was essentially the same as for the
leaner mixture.
Another simple test was performed by introducing a 76 ppm mixture into
the "clean" sample cell at different'flow rates. A constant flow rate of 5
liters per minute was adequate to maintain the S02 vapor concentration within
less than 1% of the maximum concentration that would be measured with much
higher flow rates. The 5 1pm flow rate represents the minimum that can be
measured reliably with the low-capacity flowmeter on the manifold assembly.
Lower flow rates may be adequate, but if they are used, checks should be
made for possible errors due to adsorption.
35
-------
NOISE LEVEL AND ZERO-DRIFT
The dominant source of noise is the photoelectron noise generated in
the photomultiplier. An oscilloscope trace of the photomultiplier output
shows no apparent noise during the periods of each cycle when the spinning
reticle blocks the uv energy from the photocathode. The noise level, at
30 Hz frequency, on the photomultiplier output was found to be approximately
proportional to the square-root of the uv energy level on the photocathode;
this relationship is expected when photoelectron noise is dominant. The
output of the electronics that process the photomultiplier signal is pro-
portional to the ratio of the 30 Hz component to the 360 Hz component. The
360 Hz component is proportional to the uv energy level on the detector;
therefore the 30 Hz noise level on the instrument output is inversely pro-
portional to the square-root of the incident uv energy level. It follows
that for a given S02 concentration, the signal-to-noise ratio is proportional
to the square-root of the incident uv energy level.
The noise level on the 5 ppm full-scale range was measured over a
5-minute interval with no S02 in the sample cell, the electronic time
constant at 1 sec, and the deuterium arc source operating at 0.7 amps, the
maximum current obtainable with the power supply. The output signal was
recorded on a strip-chart recorder, and the digital output meter was read
once each 5 sec. The peak-to-peak noise level corresponded to approximately
0.2 ppm of S02, and the rms noise level as determined from the 60 digital
readings was 0.049 ppm. As expected, the noise level decreases when the
electronic time constant is increased.
The zero-setting is quite stable, particularly after the instrument has
been "warmed-up." During one test period of 30 minutes, the zero-setting,
as averaged over a 1-minute interval, drifted less than 0.05 ppm. The
amount of zero-drift in practical use will depend mostly on the temperature
stability of the air surrounding the instrument.
36
-------
SAMPLE TURN-OVER TIME
Table 2 sumarizes the results of a series of tests performed to deter-
mine the time response of the instrument to an abrupt change in the S02
concentration of the gas entering the sample cell. Time resolution, rather
than the actual response time, is the quantity of most interest. In a
practical application, there is a delay between the time of a sudden change
in S02 concentration at the exhaust pipe of an automobile and the time that
the gas reaches the sample cell where the change is sensed. This delay
depends on the gas lines and on the flow rate and will be different for
different systems. This delay can be measured and accounted for so that the
proper time relationship can be determined between the observed concentration
change and any changes in the engine operation.
An important factor in the design of the sample cell and gas-handling
system has been the need to maintain good time resolution. Consider the case
in which the S0£ concentration at the inlet to the filter on the pump assem-
bly changes abruptly from level A to a different level B. After a certain
delay, the instrument reading starts deviating from level A and goes through
a continous change to level B. This period of transition of the instrument
reading is a measure of the time resolution, which is a function of gas flow
rate, mixing in the sample cell, and electronic response time. The very
gradual change in the instrument reading as it approaches level B makes it
impractical to measure the complete transition time. In the laboratory tests,
we measured times required for 70% and 90% of the change. In every case the
period started when there was the first distinct change in the deflection of
the recorder that monitored the output signal.
Test No. 1 was performed to determine the response of the recorder to a
sudden change in its input signal. The recorder signal was started at approx-
imately 80% of full-scale; the input jack was unplugged suddenly, and the re-
corder response was measured. This measurement, of course, does not involve
the main part of the S02 monitor. As in all of the other tests, the recorder
chart speed was 1 inch per second. The estimated error in the measured times
is approximately 0.05 sec for periods less than 1 sec, and somewhat more for
longer periods. The values tabulated represent averages of several indepen-
dent readings.
Test No. 2 provides information on the response of the electronics,
including the photomultiplier, the signal processing electronics and the
recorder. The results of this test, or the previous one, have nothing to
do with the mixing of the gas in the sample cell. A small mask was held so
that it covered a portion of one of the slits of set A in the grid. This
produced an output signal that corresponded to approximately 17 ppm of S02.
While recording the output signal, we abruptly removed the mask. Approxi-
mately 0.35 sec was required for the recorder to make 70% of the change; 0.45
sec was required for a 90% change. These times represent the absolute
minimum that we would expect to measure when changing the S02 concentration
if no mixing of the sample took place. The electronic time constant was 0.1
sec for this test and for all tests involving the main electronics.
37
-------
TABLE 2. RESPONSE TIMES FOR STEP CHANGE IN S02
CONCENTRATION AT VARIOUS SAMPLE FLOW RATES
Test
No.
1
2
3
4
5
6
7
8
9
Change Made
or
Flow Rate in 1pm
Unplugged Recorder
Input Cable
Uncovered Portion
of Grid
160a
160b
130b
85b
38b
19b
9.5b
Time for
90% Change
(Sec)
0.29
0.48
0.74
0.74
0.78
1.18
1.25
1.96
3.68
Time for
70% Change
(Sec)
0.25
0.35
0.45
0.48
0.55
0.73
0.88
1.43
2.52
Electronic time constant is 0.1 sec for all tests.
aSample introduced at filter on pump assembly
Sample introduced at distribution tube of sample cell
38
-------
It is apparent from the results of tests No. 1 and No. 2 that the
recorder adds significantly to the measured response times. Somewhat more
accurate data on the sample turnover time could be obtained with a faster
recorder; however, the lack of a rapid recorder response probably added no
more than 0.1-0.2 sec to the times measured when the gas concentration was
changed.
Test No. 3 was performed by injecting a slow, steady flow of pure S02
into the gas stream just as it entered the filter on the pump assembly. The
injection rate of the S02 was adjusted to produce a concentration in the line
of approximately 17 ppm of S02. While recording the output signal, the flow
of pure S02 was stopped abruptly in such a way that there was no slow diffu-
sion of S02 from the injection needle after the flow was stopped. The 70%
and 90% turnover times of 0.45 and 0.74 sec indicate that mixing in the gas
line and sample cell prevents a gas "front" from moving sharply across the
monitoring beam.
Test No. 4 was performed by the same method as Test No. 3, except that
the pure S02 was injected into the line just ahead of the distribution tube
so that it did not travel through the pump and manifold. The results of
Tests 3 and 4 are essentially in agreement, indicating that most of the
breaking-up of the gas front occurs in the sample cell, not in the pump
or manifold assemblies.
Tests 5 through 9 were identical to Test No. 4, except that the sample
flow rates were varied. As expected, the turnover times decrease with in-
creasing flow rates, but not in the same proportions. There is apparently
less breaking-up of the gas front at low flow rates. For example, in re-
ducing the flow rate by a factor of 9, from 85 1pm to 9.5 1pm, the 90%
turnover time increases by a factor of 3.68/1.18 =3.1. It is apparent that
little is gained by operating at flow rates above 130 1pm.
If a much faster recorder were used, the minimum 90% turnover time would
probably be between 0.6 and 0.7 sec. This period is somewhat higher than
the 0.5 sec design objective.
39
-------
INTERFERENCE BY OTHER EXHAUST GASES
Interference by automotive exhaust gases other than S02 is negligible
at the concentrations these gases normally occur in the exhaust. No measurable
interference was found for the following hydrocarbons tested individually at
Aeronutronic: acetylene, methane, butane, propane, ethylene and ethane. Each
sample consisted of a mixture of a single hydrocarbon gas in N2 at 1 atm total
pressure. The data on hydrocarbon interference and other data obtained at
Aeronutronic are summarized in Table 3 along with still other data obtained at
EPA by EPA scientists.
Many of the interference data were obtained with a small sample cell
temporarily placed in the monitoring beam while the main multiple-pass sample
cell was by-passed. This improved the signal-to-noise ratio because of the
reduction in reflective losses in the multiple-pass optics. The short sample
cells also enabled us to use smaller amounts of gas at higher concentrations
than would be used in the large, multiple-pass cell. The mixtures of higher
concentration are easier to mix accurately, and there is less potential
error due to adsorption on the walls of the sample cell. Table 3 gives the
cell lengths and the gas concentrations along with the equivalent concentra-
tions for the multiple-pass cell.
Ammonia (NHo), nitric oxide (NO), and nitrogen dioxide (M^) each inter-
fere slightly, but the amount of interference produced by the low concentra-
tions in automotive exhaust is negligible. Because of the detector noise and
the very slight interference, the discrimination ratios could not be measured
accurately. Nitric oxide absorbs quite strongly in the spectral band employed
by the instrument, but there is very little correlation between the structures
in the S02 spectrum and the N02 spectrum. Thus, there is very little inter-
ference .
When the 100% C02 test gas first entered the sampling volume, the output
signal indicated an 862 concentration of approximately 0. 2 ppm, but this sig-
nal vanished as the C02 completely displaced the N2 that was originally in
the sample cell. The temporary signal was apparently caused by a non-uniform
distribution of the C02 in the cell. Differences in the indices of refraction
of C02 and ^ probably produce small deflections in the monitoring beam when
the gases are both present in the cell and are not mixed. Absorption by C02
is completely negligible in the spectral interval passed by the instrument.
A small signal was also observed some of the times when an ^0 mixture
first entered the sample cell that was previously filled with N2. The size
of this transient signal was not reproducible, but it never corresponded to
more than 0.2 ppm of S02. As in the cases when C02 was introduced into the
cell, the transient signal disappeared when the cell became completely filled
with either the ^0 mixture, or with N2 when the ^0 mixture was being flushed
out. When the ^0 mixture was in the sample cell, the 360 Hz component of the
detector signal was typically about 1% lower than it was when the cell was
filled with No, The 1% attenuation is more than can be accounted for by ^0
vapor absorption. This attenuation is not understood, but it may be due to
a change in the reflectivity of the mirrors of the multiple-pass optical
system that results from a layer of absorbed water on the reflecting surfaces.
40
-------
The lower part of Table 3 summarizes the results of tests performed
on two complex mixtures simulating automobile exhaust hydrocarbons. These
mixtures did not contain other common exhaust species such as CO, C02,
N0y, or aldehydes such as formaldehyde
The samples were prepared by injecting either gasoline or diesel fuel
into a heated tube while purging with nitrogen, The vaporized fuel samples
were collected in large Tedlar bags and the total hydrocarbon concentration
in ppm determined with a flame ionization detector (F.I.D.) hydrocarbon
analyzer, The overall compositions of the two fuels are given as a footnote
in Table 3.
The aromatic species in each of these samples would be the most likely
to cause interference as they tend to absorb in the U.V. close to the 300nm
region which the 862 monitor operates. The aromatics in the gasoline are
mostly benzene, toluene and other alkyl benzenes. The diesel fuel aromatics
contain alkyl benzenes, alkyl naphthalenes, alkyl anthracenes, alkyl phenanth-
rences and benzothiophenes.
-------
TABLE 3. INTERFERENCE DATA
Test
No
1
2
3
4
5
6
7
8
9 '
10
11
Gas
Acetylene
Methane
Butane
Propane
Ethylene
Ethane
Ammonia
Nitrogen-
Dioxide
Nitric
Oxide
Carbon
Dioxide
Water
Vapor
Concentr-
tion
7o
100
100
100
100
100
100
100
25
40
100
2.5
Path
Length
(cm)
3
3
3
3
3
3
3
0.4
0.4
480
480
Equiv. Con-
centration
(ppm)
6250
6250
6250
6250
6250
6250
6250
208
333
106
25,000
Interfer- Discrimi-
ence Error nation Ratio
(ppm of S02>
N.D.a
N.D.
N.D.
N.D.
N.D.
N.D.
0.02 b
0.32 c
0.01 d
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
330,000:
650:
30,000:
N.D.
N.D.
1
1
1
The following tests on gasoline and diesel fuel were performed at EPA by EPA personnel.
El
E2
E3
E4
Gasoline e
Gasoline
Diesel e
Fuel
Diesel
Fuel
100 ppm
THC
400 ppm
THC
100 ppm
THC
180 ppm
THC
480
480
480
480
100
THC
400
THC
100
THC
180
THC
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
(a) N.D. is the abbreviation for Not Detectable
(b) Absorptance of the ammonia sample was 0.02
(c) Absorptance of the N02 sample was 0.027
(d) Absorptance of the NO sample was 0.04
(e) The overall compositions of the two fuels were:
Gasoline Diesel Fuel
7o Paraffins
7, Olefins
7» Aromatic s
47.3
7.8
44.9
100.0
82.4
0.4
17.2
100.0
42
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
"^REPORT NO. ~" I ~ '
EPA-600/2-79-121
4. TITLE AND SUBTITLE
DEVELOPMENT OF AN S00 MONITOR FOR MOBILE SOURCES
7 AUTHOR(S) *•
jp.E. Burch, P.S. Davila. F.J. Gates, and J.D. Pembrook
9. PERFORMING ORGANIZATION NAME AND ADDRESS ~~^
Ford Aerospace and Communications Corporation
Aeronutronic Division
Ford Road
Newport Beach. California
29663
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory
- RTP, NC
Office of Research and Development
U. S. Environmental Protection Agency
Research Triangle Park. North Carolina 27711
15. SUPPLEMENTARY NOTES
16. ABSTRACT "
An instrument has been designed and built to monitor the
exhaust of mobile sources. A grating assembly desperses
deuterium-arc source and passes five narrow
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
Oiily 1979
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO
U-6392
10. PROGRAM ELEMENT NO.
1AA601B CA-29 (FY-78)
11. CONTRACT/GRANT NO.
68-02-2448
13. TYPE OF REPORT AND PERIOD COVERED
Final 10/76 - 1/78
14. SPONSORING AGENCY CODE
EPA/600/09
concentration of S0» in the
the ultraviolet energy from a
spectral intervals. Three of the intervals
(set A) coincide with strong absorption features in the S02 band near
3000 A; the other
two intervals (set B) coincide with weak absorption. A spinning reticle alternately
transmits energy passing through set A and
put signal proportional to
appropriate electronics.
B to a photomultiplier detector. A dc out-
the concentration of S0? in the sample cell is produced by
A pump and manifold assembly permit operation over a wide range of sample flow rates.
At high flow rates, the 90%
sample turn-over time is approximately 0.7 sec. The sample
cell contains a multiple-pass optical system adjusted to
path length greater than 5
12 passes, giving a sample
m. The rms noise level corresponds to approximately 0.05
ppm of SO. when the electronic time constant is 1 sec.
•
17.
•™— — »— ^ — ^ ________ „ _________«__ _______»__
a- DESCRIPTORS
* Air Pollution
* Motor vehicles
* Exhaust emissions
* Sulfue dioxide
* Monitors
* Development
»?. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
KEY WORDS AND DOCUMENT ANALYSIS ~~
— ___• •! Ill !!•_• ••
•^•^•••••^•^•••••••••^•i^^B^^^HIIII^HMVBIIIIBIMV
b.lDENTIFIERS/OPEN ENDED TERMS
Correlation-spectrometer
19. SECURITY CLASS (This Report 1
UNCLASSIFIED
20. SECURITY CLASS (This page)
UNCLASSIFIED
e. COSATI Field/Group
13B
13F
21B
07B
21. NO OF PAGES
51
22. PRICE
Form 2220-1 (Rev. 4-77) PREVIOUS EDITION is OBSOLETE
43
------- |