EPA-650/2-73-030
October 1973
ENVIRONMENTAL PROTECTION TECHNOLOGY SERIES
ill;
£-^ Jh'XwZ* I»A* jL&'jL**jL&jl4£'lt jjjl'Z£'l'frbQL'^i-L-^-£^'l'^'i'~S.-'. •'.•'.•'.•'•'
-------
EPA-650/2-73-030
INSTRUMENT
TO MONITOR CH4/ CO, AND C02
IN AUTO EXHAUST
by
D. E. BurchandJ. D. Pembrook
Philco-Ford Corp.
Aeronutronic Division
Ford Road
Newport Beach, California 92663
Contract No. 68-02-0587
Program Element No. 1AA010
EPA Project Officers: J. Sigsby and F. Black
Chemistry and Physics Laboratory
National Environmental Research Center
Research Triangle Park, N. C. 27711
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, B.C. 20460
October 1973
-------
This report has been reviewed by the Environmental Protection Agency and
approved for publication. Approval does not signify that the contents
necessarily reflect the views and policies of the Agency, nor does
mention of trade names or commercial products constitute endorsement
or recommendation for use.
11
-------
ABSTRACT
An infrared analyzer employing gas cell correlation techniques and
thermoelectrically cooled photodetectors has been designed and con-
structed to measure the concentrations of methane, carbon monoxide, and
carbon dioxide in automotive exhausts. The dynamic range from the mini-
mum detectable concentration to the maximum concentration that can be
measured accurately for each gas is: methane, 0.3 ppm to 10,000 ppm,
carbon monoxide, 0.5 ppm to 10,000 ppm, and carbon dioxide, 10 ppm to
250,000 ppm. The wide dynamic range is made possible by employing two
different sample cell lengths for each gas. The concentrations of the
three gases are measured simultaneously and independently. Discrimi-
nation against other gases in the automotive exhaust is very good.
This report was submitted in fulfillment of contract number 68-02-0587
by Philco-Ford Corporation, Aeronutronic Division, under the sponsor-
ship of the Environmental Protection Agency. Work was completed as of
August 1973.
iii
-------
CONTENTS
Page
Abstract
List of Figures v
Acknowledgements vi
Sections
I Conclusions and Recommendations 1
II Introduction 3
III Optics 12
IV Electronics 21
V Test Data . 29
VI Glossary of Symbols 36
VII Patent Information 39
iv
-------
FIGURES
No. Page
1 Block diagram of a single gas analyzer 5
2 Rotating gas correlation cells 6
3 Calculated spectral plots of transmittance for a
model spectrum similar to a portion of a CO band 8
4 Calculated spectral plots of transmittance
demonstrating the response of the gas correlation
system to a gas other than the one to be detected 10
5 Schematic diagram of the optics assembly 13
6 Expanded view of slit assembly S2 or S3 18
7 Power supply unit above a single electronics unit 22
8 Schematic illustration of a single channel of the
photodetector electronics within the optics assembly 24
9 CO detection sensitivity 30
10 CH^ detection sensitivity 31
11 C02 detection sensitivity 32
-------
ACKNOWLEDGEMENTS
We wish to give special recognition to the ingenuity of Mr. F. J. Gates,
a member of our research team. He devoted much of his own time to the
development of the instrument and to the testing of its performance.
We appreciate the cooperation and the candor of EPA technical personnel,
particularly Mr. F. M. Black in discussions about the instrument and
its specifications. The EPA personnel at Raleigh-Durham, North Carolina
also furnished us with an analyzed sample of lead-free gasoline for
testing the instrument.
vi
-------
SECTION I
CONCLUSIONS AND RECOMMENDATIONS
This program has been completed successfully. An instrument has been
designed, constructed, and tested. All the major design criteria were
met. The instrument can measure CH4, CO, and C02 simultaneously and
independently.
The measurement range includes gas concentrations that vary approximately
from those of room air to those of raw automotive exhaust. The system
has very good discrimination against other gases that occur in auto-
motive exhausts. Water vapor is the only gas that interferes enough to
consider a correction factor to the measurements of CO, ClU, and COo.
If the exhaust gas sample contained 3% water vapor, the maximum error
resulting from it would be 0.6 ppm (part-per-million) of CO,. 2.3 ppm of CH4,
and 30 ppm of COo. The concentration of water vapor in the sample can
easily be determined, or estimated, to within +0.5%. Thus, the maximum
error, in the measurements due to water vapor can be reduced with a cor-
rection factor to 0.1 ppm for CO, 0.4 ppm for CH^, and 5 ppm for C02.
The large dynamic range of the instrument necessitates the use of two
different sample cell lengths for each gas. Cells of two lengths are
active at all times for the CO and CH^ systems. The operator selects
between the two photodetector outputs from the two cells with a switch.
Electronic automatic ranging could be incorporated into the instrument.
In the 2 system, only one of the two sample cells is in the optical
path at a time. A bi-stable mechanical cell position device allows the
operator to select between and reposition the cells in a few seconds.
Only one photodetector is required for this system.
The spectral passbands of the CO and CH/ channels of this instrument are
defined with a moderately complex grating-filter assembly. This assembly
allowed the passbands to be adjusted and selected for good sensitivities
to CO and City and very good discrimination against other gases. If the
requirements on discrimination were relaxed, the optics assembly could be
simplified by employing other methods of defining the spectral passbands.
-1-
-------
If only one sample cell length were required at a time for the CO and
CH4 systems, then the optical throughput of these systems could be
increased. This would result in an increase in the signal-to-noise
levels of these channels. Further, only one photodetector would be
required for each channel. A single cell device would have a dynamic
range that is an order of magnitude less than that of the present instru-
ment. The C02 system demonstrates one technique for mechanically switch-
ing between two sample cells. Similar techniques could allow switching
between three or more cells of different lengths for extended dynamic
ranges.
The photodetector output signals of each channel are processed independ-
ently by three separate electronics units that are almost identical to
each other. This modular construction technique allowed thorough testing
of the first unit before design approval and construction of the subse-
quent units. If one electronics unit becomes inoperable, either of the
remaining units can be substituted. The total volume occupied by the
electronics could easily be halved by repackaging the three electronic
units into a single cabinet. If this were done, the parts count could
also be reduced by about 25 percent by using a single power supply and
reference amplifier.
-2-
-------
SECTION II
INTRODUCTION
PERFORMANCE
This section presents sufficient detail for a general understanding of
how the exhaust gas analyzer functions. The operational characteristics
are summarized. This is followed by a general description of the optics
and the electronics. The spectral principles of operation are described
in the last portion of this section. Subsequent sections present
detailed descriptions of the instrument, its operation, and its calibra-
tion and performance.
The exhaust gas analyzer provides for the simultaneous measurement of CO,
CH^, and C02- The concentrations of the individual gases are analyzed
with a three-channel gas correlation spectrometer. The output signal is
proportional to the absorber thickness of the sample gas, which is the
product of the concentration of the gas species being measured and the
sample cell length under the fixed conditions of room temperature and
one atmosphere pressure. Each channel employs two sample cells of dif-
ferent lengths to extend the useful measurement range. The CO and the
CH^ systems have common sample cells. Either system may use either or
both cells at any time. All of the photodetectors are thermoelectrically
cooled.
With a 1 second time constant on the most sensitive ranges, the minimum
detectable gas concentrations are approximately 0.3 ppm (parts-per-million)
of CH4, 0.5 ppm of CO, and 10 ppm (0.001%) of C02. The maximum gas
concentrations that can be measured reliably are 10,000 ppm of CH^,
10,000 ppm of CO, and 250,000 ppm (25%) of CO?. The gas correlation sys-
tem offers very good discrimination between the gases being measured and
other gases that occur in automotive exhaust. The system is also insen-
sitive to particulates in the sample, dirty windows, and changes in
source brightness.
-3-
-------
EQUIPMENT DESCRIPTION
All three channels of the instrument are similar. Many of the component
parts are shared. For clarity, the following description is limited to
a single channel. Figure 1 is a block diagram of any channel of the
device. Radiant energy from an infrared source is imaged on a photo-
detector through a chopper, the rotating gas correlation cells, one of
two sample cells, and a narrow bandpass filter. Optical components
between the source and the photodetector were selected and arranged to
give good stability along with a large throughput. The energy is chopped
at 360 Hz (Hertz) by an opaque, 12-bladed, chopper deposited on the entrance
window of the rotating correlation cells. Figure 2 shows an exploded
view of the correlation cells. One of the cells GC2, is filled with a
non-absorbing gas and has a neutral attenuator deposited on the window
of this cell. The other cell, GCl, is filled with gas of the species
being detected. Sufficient gas is placed in this cell to cause equal
attenuation by both cells over the spectral bandpass defined by the fil-
ter in front of the photodetector. The narrow bandpass filter assembly
may be a simple interference filter or a more complex monochromator. The
complexity of this assembly is determined by the requirements of sensi-
tivity to the gas species of interest and the rejection of, or insensi-
tivity to, other gases that may be present in the unknown sample gas.
The sample gas flows continuously through the sample cell to minimize
errors that could be caused by adsorption or desorption of gas on the
sample cell walls.
The gas-filled correlation cells rotate at 1800 rpm, which has the effect
of switching the optical path between the correlation cells at 30 Hz.
When no sample is present in the sample cell, the photodetector output
consists of a 360 Hz carrier signal due to the 360 Hz chopper. When the
sample cell contains gas of the same species as the gas in the correlation
cell, the photodetector output consists of a 360 Hz carrier modulated at
30 Hz. The amount of 30 Hz modulation is proportional to the absorber
thickness of the detected gas present within the sample cell. Thirty Hz
modulation to the carrier occurs only when the sample gas absorbs energy
within the system's spectral bandpass. Further, for modulation to occur,
the absorption must correlate to some extent with the absorption charac-
teristics of the gas within the gas correlation cell. Placing a neutral
attenuator within the sample cell would not cause 30 Hz modulation since
both halves of the correlation cell cycle would be affected equally.
The carrier may exhibit 30 Hz modulation when no sample is present if
one correlation cell window collects more dust than the other or if the
temperature of the correlation cell changes. Since these events are
certain to occur, no attempt was made to achieve perfect balance between
the cells. Instead, an electronic 30 Hz modulator is employed to cancel
any residual modulation that occurs.
-4-
-------
£
Amplifier
i
Phase Shifter
^
Adjustable
Phase Shifter
— »
f
360 Hz Tuned
Amplifier
J
Gain Controlled
Amplifier
4
Modulation Balance
|
AGC Control Amplifier
1
^
Range Switch
4
Calibrate Pot
|
30 Hz Tuned
Signal Amplifier
J
30 Hz Synchronous
j
Adjustable Time
Constant Amplifier
Meter and
Recorder Output
ELECTRONICS UNIT
'Figure 1. Block diagram of a single gas analyzer.
-------
-WINDOW
CHOPPER
ATTENUATOR
-WINDOW
Figure 2. Rotating gas correlation cells. Windows are antireflection overcoated sapphire.
-------
SPECTRAL RESPONSE
The spectral sensitivity of the gas correlation spectrometer is set by
the narrow bandpass filter and by the absorption characteristics of the
gas in the gas correlation cell. The CO detection system is used to
illustrate the operating principles. The use of the carrier at a fre-
quency, fc, of 360 Hz does not affect the spectral sensitivity of the
instrument. If the carrier chopper were removed from the system shown
in figure 1, the detector output would occur at the frequency, fa, of
30 Hz. During one half the cycle GC2 would be in the beam; during the
other half cycle the gas filled correlation cell, GCl, would be in the
beam.
Figure 3 contains several calculated plots of transmittance that can be
used to explain the spectroscopic principles of detection. We have
assumed certain gas sample parameters and have chosen a band model
consisting of two absorption lines with spacings, strengths, and half-
widths typical of the stronger lines of the fundamental CO band. The
half-widths at 1 atmosphere pressure are 0.08 cm"1; strength = 2rr; and
the lines are separated by 4 cm"1. Correlation cell GCl is assumed to
contain 0.2 atm cm of CO at a total pressure of 1 atm. Its transmittance
spectrum over the region of the calculation is shown in the upper portion
of figure 3 and is indicated by Tj_. We note that this sample is essen-
tially opaque (T, = 0) near the centers of the two absorption lines.
We assume^that the bandpass filter passes only the spectral region from
(v - va) = 0 to 4, which corresponds to the region between the centers of
the lines.
The attenuator on GC2 is a netural density filter with constant trans-
mittance, T2, equal to the average transmittance of the CO in correlation
cell GCl over the bandpass of the filter. The third curve in the upper
panel of the figure is a plot of the difference between the other two
transmittance curves. This curve represents the spectral response of
the system. Since T2 is equal to the average of T1; the integral of
the curve (T2 - T^) is 0. Thus, with no sample in the sample cell, no
voltage component Va is produced as the rotating cells move first GCl and
then GC2 into the beam. <
We now consider a sample cell containing enough CO to produce the trans-
mittance spectrum indicated by the center panel of figure 3. This
corresponds to two of the stronger CO lines with an absorber thickness of
0.01 atm cm at a pressure of 1 atmosphere. When GCl is in the beam, the
spectrum after the beam has passed through the sample is given by the
product of the transmittances TI x Tsam. This is indicated by the curve
in the lower panel of figure 3 along with the corresponding curve that
applies when the beam is going through GC2. The resulting detector sig-
nal, Va, is proportional to the difference between the two integrals
-7-
-------
1.00
0.50 H-
w
o
1.00
0.50
'Sam
Y
1.00
0.50
v/ /
I /
1 /
.'T, x T
2 sam
Figure 3
x T
1 sam
.
V
V-V WAVENUMBER (cm"1)
a
Calculated spectral plots of transmittance for a model
spectrum similar to a portion of a CO band. The wave-
number scale is measured from the center of one of two
identical lines. The curves illustrate the spectral
principle of operation of a gas analyzer using one gas
correlation cell filled with the gas species being
measured and an attenuator with a flat response whose
absorptance is equal to the average absorptance of the
gas correlation cell.
-------
J(T1 x Tsam) dv and ,f
-------
1.00
0.50
1.00
H
g
I
0.50
T = T
sam X
1.00f-
0.50
0
-0.10
(T -
x T
sam
-V-v WAVENDMBER (cm"1)
Figure 4. Calculated spectral plots of transmittance
demonstrating the response of the gas corre-
lation system to a gas other than the one to
be detected.
-10-
-------
in the sample. For example, if I , the average transmittance of the
interferring sample is 0.8, the voltage Va produced by the CO is reduced
to 0.8 of its correct value. It follows that changes in source brightness
or detector sensitivity would also change Va accordingly.
We now reconsider the system illustrated in figure 1 with the chopper
operating at a carrier frequency fc. The detector signal, V , at
frequency fc is proportional to J*T2 x T8ain dv when GC2 is in°the beam
and is proportional to JT]_ x Tsa!L dv when GCl is in the beam. The sum
of these two integrals, RTi + T2) Tsam dv is proportional to Vc, aver-
aged over a complete cycle of the pair of rotating cells. Absorption
by an uncorrelated interfering gas reduces Vc by the same fraction that
it reduces Vai which is proportional to J(T2 - Ti)Tsam dv. Thus the
ratio Va/Vc is unchanged by an interfering gas that is uncorrelated with
the CO, or by variations in source brightness or detector sensitivity.
The signal of primary importance is therefore Va/Vc, since it is a mea-
sure of the amount of CO gas and is insensitive to the variations that
produce serious errors in many systems.
An electronic circuit designed to measure Va/Vc is shown in figure 1.
When correlation exists between the absorption spectra of the sample
and the absorption spectra of the gas within the correlation cell, the
carrier is modulated at 30 Hz. The phase of the modulation depends on
whether the correlation is positive or negative. Since a synchronous
demodulator is used in the electronics, the presence of certain sample
gases may result in a negative signal output.
To simplify the measurement of Va/VC; an AGC (automatic gain control) is
employed within the electronics. Both Va and Vc are amplified by the gain
controlled amplifier. The gain of this amplifier is continuously adjusted
by the AGC to hold the average value of Vc constant. The Va component of
Vc is directly proportional to the ratio Va/Vc when Vc is held constant.
The system is calibrated in terms of signal output with the AGC operating
versus the absorber thickness of the gas within the sample cell. The
calibration is performed with a flowing gas sample at one atmosphere
pressure and at room temperature; the absorber thickness is proportional
to the gas concentration for a given sample cell length.
-11-
-------
SECTION III
OPTICS
GENERAL DESCRIPTION
The three-channel gas correlation spectrometer optics assembly was
designed for good stability and large throughput while preserving a
reasonably small package size and weight. The selection of optical
components was limited to off-the-shelf items wherever possible. First
surface mirror optics were used exclusively to eliminate chromatic
aberation. All imagery is performed with spherical mirrors. All of
the windows are sapphire and are antireflection coated to reduce reflec-
tion losses and optical intereference between window surfaces. Apertures
reduce scattered energy; some are overfilled to increase stability.
Components are arranged to minimize alignment errors that could result
from flexure in the baseplate or from differential expansion due to
thermal gradients. Except for the source, all components that can be
removed are either self aligning or do not require alignment. All other
components are bonded in place. The system may either be used horizon-
tally on a bench or mounted vertically within an electronics rack. The
overall dimensions of the basic optics assembly as used on a bench top
are about 39 cm wide by 25 cm high by 100 cm long. Angle brackets
attached to the instrument sides with rubber loaded bolts allow the optics
assembly to be mounted in a standard 19-inch (48.3 cm) electronics rack.
A schematic diagram of the optics assembly is presented in figure 5. The
three-channel system has two major subdivisions; the C02 system and the
CO and Cfy systems. Only the motor, the Nernst source, the reference
pickup, and the bias battery are shared by these two subsystems. Each
of these subsystems is further divided into the subassemblies presented
in figure 1 for a single channel system.
SOURCE ASSEMBLY
A single infrared source is shared by all three channels of the gas
correlation spectrometer. The Nernst glower source has a relatively
high radiant emittance in all three spectral regions. It also has good
-12-
-------
LO
I
M23
Figure 5. Schematic diagram of the optics assembly.
The components are. numbered as a mirror M, window W; thermoelectrically cooled
photodetector D, filter F, plane diffraction grating G, aperture A, slit S,
gas correlation cell GC, or bearing block BB.
-------
stability and long life, and is easy to use. The standard source
mount has been mounted on an assembly that allows transverse adjust-
ment for proper alignment of the source.
The source dissipates about 60 watts of power during normal operation.
The source is surrounded by a double walled box. A fan forces room air
through the box between inner and outer walls to keep the system cool.
Gaskets prevent the cooling air from leaking into the optical path. The
cooling assembly is thermally isolated from the remainder of the optics
unit. The cooling assembly has been black anodized to increase its
emissivity and absorptivity. The wall that separates the source from
the gas correlation cell assembly has been mirrored to reflect heat
back into the source cooling assembly. The power dissipation of the
source is reduced to about 18 watts when the source is operated in the
standby mode. The fan speed is reduced during this mode of operation to
prolong the fan lifetime.
GAS CORRELATION CELL ASSEMBLY
An exploded view of a set of rotating gas correlation cells is presented
in figure 2. The cells are pinned to an axial shaft that is driven at
1800 rpm by a synchronous motor. This has the effect of switching the
optical path between the cells at 30 Hz. Energy from the source is
chopped at 360 Hz with an opaque 12-bladed chopper that is deposited on
the entrance window of the rotating cells. The 360 Hz chopper is on the
source side of the gas correlation cells to minimize scattered energy
within the cells. Cell GC2 contains a non-absorbing gas, N2, and has a
metallic coating of about 70% transmittance deposited on one window.
Cell GC1 contains 5% N2 as well as the two gas species being detected,
32% CH4 and 63% CO at a total pressure of 0.71 atm. The concentrations
and the total pressure were chosen so that the transmittances of the CH4
and the CO over the corresponding spectral band passes matched that of
GC2. The lengths of GC1 and GC2, 1 cm, were selected so that the total
pressure of the mixture in GC1 that balanced the transmittance of GC2
was somewhat less than 1 atm. The construction of the cells is simpli-
fied by keeping the internal pressures below atmospheric pressure. The
windows are pressed against the bond type seal between the window and
the cell body by the pressure gradient. Optimum performance is normally
achieved when the gas pressure in GC1 is equal to or somewhat less than
the pressure of the sample being studied. However, the effect of changing
the total pressure by as much as a factor of 2 is slight provided the con-
centrations of the CH4 and CO are adjusted to balance the transmittance
of GC2.
A similar pair of correlation cells is used for C0~. Cell GC3 contains
0.66 atm of pure C02 and GC4 contains N2 at approximately the same
pressure. These cells are 2 cm long. One window of GC4 is coated with
an attenuator similar to that on the window of GC2. The gas correlation
14
-------
cells are contained within a dustproof chamber that is sealed with a
flexible foam gasket to prevent an accumulation of dust on the cell
windows.
A single-bladded chopper pinned to the correlation cell shaft interrupts
a beam of light between a light emitting diode and a phototransistor.
The 30 Hz output of the phototransistor is used as a reference source to
allow synchronous demodulation of the 30 Hz detector signals.
The motor shaft is connected to the gas correlation cell shaft with a
flexible coupling. The motor and the bearing blocks BB are pinned to
the baseplate to maintain proper alignment. The motor is mounted within
a chamber in the optics assembly. Room air is drawn through the baser
plate and through the motor by a fan in front of the armature and ex- '
hausted back through the baseplate. This airflow removes the heat dissi-
pated by the motor without circulating room air through the optical path.
The amount of heat dissipated by the motor is reduced by operating it
at 80 V ac. This reduction in operating voltage also reduces the amount
of vibration generated by the synchronous motor without changing its
speed.
C02 SYSTEM
The source is Imaged by spherical mirror M51 through a set of gas corre-
lation cells onto aperture A54, which is mounted on window W51. The
source energy transmitted by A54 is imaged by spherical mirror M55 through
the sample cell and onto the plane of window W53. This energy is collected
by spherical mirror M58 and directed onto photodetector D51 through band-
pass filter F51. The passband of this interference filter is centered at
about 4.405nm (2270 cm'1). The width of the passband between the 1/2
maximum transmittance points is about 0.044nm (22 cm"1). The passband
includes most of the P-branch of the V3 absorption band of the 13C ^02
isotope. Photodetector D51 includes a PbSe element mounted on a single
stage thermoelectric cooler.
Aperture A51, which is mounted on the face of M51, is imaged by mirror
M55 onto the plane of window W52. Aperture sizes, focal lengths, and
dimensions were adjusted such that the image of A51 on W52 is the same
size as the image of A54 on W53. These images are smaller than the
inside diameter of the sample cell. The 4 cm long sample cell has
the same inside diameter as the 20 cm cell. The image of A51 on W52 is
re-imaged by M58 onto the photodetector.
Both window Wl and mirror M52 are mounted in the wall of the correlation
cell chamber near the source assembly. This wall has also been relieved
to allow for thermal expansion without affecting the adjustment of M52.
Further, M52 is shielded from direct source radiation by a wall of the
source assembly. Apertures A52 and A53 minimize scattered energy, and
-15-
-------
aperture A54 is overfilled to increase the stability. The limitine
apertures of this system are A51 and A54. A unit ing
It is expected that the 4 cm long sample cell will be used for samples
sr»
s:
-
cell. The flow is directed through the shorter cell first to allow
observance of abrupt changes in gas concentration before mixing can
dowT, ^ 10n8^ Cel1' The S3S fl°W is directed near tne"ell win-
dows to prevent volumes of stagnant gas at the ends of the sample cells.
CO and CH^ SYSTEMS
General
The
the
?4 detection systems are discussed together since most of
P nd °tlCa
te c™
T ZP^f P fnd °PtlCal ray P3ths are C0iranon to b°th systems.
The specific spectral passbands employed are controlled by the grating
fSS aST y "u C0rabination wi<* an interference filter near each
photodetector. These passbands were selected and adjusted to give a
low8* thj°?**ut' 8°od sensitivity to the gas being detected, S a very
*
,
°ther 8aSeS that mi§ht be in th sample gas. The
ered at about 3.315^, (3017 cm-1)
the 1/2 ™*^ tra^ittance points
Q-branch of * , passband is almost centered on the
Q branch of the 3.3[m absorption band of Cfy. The passband of the CO
S (2171 «-1)* w^h a wJdth of about
-
l2 oTh f ^ ? PaSSband lncludeS abs°rption lines Rl through
R12 of the fundamental absorption band of CO. "«ugn
As in the C02 system, two sample cell lengths are employed to extend the
useful range of sensitivities of the CO and CH4 systems. Both smple
iS.TuaCtiVe 3t aU times' The existinS s^ of electronics Sn be
switched between the outputs of the photodetector preamplifiers wi?h
switches on the electronics units. With this optical syste^ It would
6 °d ^
e wou
coTo!^6 °ifd? ^ ^lectr°nic a^omatic range switching devidto
control the cell length in use. Alternately, the outputs of the photo-
detectors used with both cell lengths can be'monitored s imul taneoSs 1 y
for flowing gas samples which have rapid and very large fluctuations in
the gas concentration. About 75 percent of the radiant energy trans-
mitted by the grating- filter assembly is directed through thf 100 «
sample cell. The remaining 25 percent is directed through the 10 cm
sample cell. This division of energy is based on the sifnal-to-noise
requirements of the system. uu^e
-16-
-------
Source Optics and Grating Filter Assembly
The spectral passbands of both the CO and the CH, systems are set by the
grating-filter assembly. This grating monochromater forms an undispersed
image of the entrance slit, Si, on the exit slit, S4, of the unit.
The source is imaged by mirror Ml through a pair of gas correlation cells
onto slit SI. The energy transmitted by SI is collimated by spherical
mirror Mil and directed to the plane diffraction grating G. The grating
is overfilled and acts as a limiting aperture. Part of the energy that
is dispersed by the grating is collected by spherical mirrors M12 and M13.
These mirrors form many dispersed monochromatic images of SI. Radiant
energy near 4.6 pm is focused onto slit assembly S2 by mirror M12. Simi-
larly, energy near 3 |_un is focused onto slit assembly S3 by mirror M13.
Figure 6 shows an expanded view of slit assemblies S2 and S3. The
energy is reflected vertically from the bottom mirror through the slit,
and reflected from the top mirror back to the corresponding mirror M12
or M13. The dispersed energy that is transmitted by S2 and S3 is
de-dispersed upon the second pass through the grating G. Mirror Mil
forms an undispersed image of the entrance slit SI onto the exit slit S4
through flat mirror M14. Slits SI, S2 and S3, and S4 are conjugate to
each other. The spectral widths of the two passbands are set by the
widths of S2 and S3. The center wavelengths of the passbands are set
by the angular locations of S2 and S3. The slopes of each side of the
passbands are set by the widths of the images of either SI or S4 on S2
and S3. Slit SI was made slightly wider than S4 to simplify alignment,
thus the slopes are set by the image of S4, which lies within the bounds
of SI.
The central ray of the energy transmitted by S4 is below the central
ray transmitted by SI. These rays are parallel to each other and to
the optics assembly baseplate. The central rays converge from Mil to
the center of the grating where they cross. They are divergent from
the grating to slit assemblies S2 and S3. The limiting apertures of the
grating-filter assembly are slits S2, S3, S4, and the grating G. Aper-
tures Al, All, A12, and A13 reduce the scattered energy.
The grating-filter assembly transmitts undispersed radiant energy at
multiple orders of the desired spectral passbands. Interference filters
are used to limit each passband to a single order of the grating. A
pair of these narrow bandpass filters are located near the photodetectors
of each channel. This restricts the amount of ambient radiation reaching
the photodetectors. One of these filters is also used as a dichroic
mirror to separate the 4.6 ^m radiant energy from the 3 u.m radiant energy.
The components in the grating-filter assembly are enclosed within a
dust-proof chamber. The base of this chamber serves as the mounting
-17-
-------
oo
MIRROR
SLIT
MIRROR
Figure 6. Expanded view of slit assembly S2 or S3. Many monochromatic
images of the entrance slit SI are formed on slits S2 and S3.
-------
platform for these components. This chamber is, in turn, mounted on
three spring loaded points against the optics assembly baseplate. A
ball and spring-loaded-post maintain correct alignment between the
grating-filter assembly and the remainder of the optics assembly. This
mounting technique isolates the grating-filter assembly from stresses
that may be applied to the optics assembly baseplate. This isolation
assures stable wavelength passbands for the CO and the CH^ systems.
Sample Cell and Photodetector Assemblies
The optical path length between the short sample cell entrance and
exit windows W21 and W32 is 10.0 cm. The radiation that exits through
the lower 25 percent of slit S4 is intercepted by flat mirror M31 and
directed to spherical mirror M33. An image of mirror Mil is formed on
the entrance window W21 of the sample cell. An image of slit S4 is
formed on photodetectors D31 and D32. Interference filter F31 reflects
3 urn energy and transmits radiant energy near 4.6 pm. Interference
filters F31 and F32 are bandpass filters that select a single order of
the energy transmitted by the grating-filter assembly. They also reduce
the amount of scattered energy reaching the photodetectors.
The optical path length of the double-passed sample cell is 100.0 cm
The energy that exits through the upper 75 percent of slit S4 is collected
by spherical mirror M21. The top edge of mirror M31 defines the lower
boundary of slit S4 as viewed from mirror M21. Both M31 and S4 are mounted
on window assembly W12 to maintain stable slit dimensions. An image of
mirror Mil is formed on window W21 by spherical mirror M21. Similarly,
an image of S4 is formed on mirror M23 by mirror M21. Spherical mirror
M23 forms another image of Mil beside the image of Mil on W21. The
image of S4 on M23 is re-imaged on photodetectors D21 and D22 by spheri-
cal mirrors M24 and M25, respectively. Interference filters F21 and
F22 serve the same functions as filters F31 and F32. Photodetectors
D21, D22, D31, and D32 include PbSe elements which are mounted on two-
stage thermoelectric coolers.
The sample cell windows and end mirrors are easily removed for cleaning.
The windows are bonded to flanges. The end mirror is covered by a
metal cap. Rubber "0" ring seals with metal-to-metal contact between
the cell and the end cap or window flanges provide a gas tight unit.
Similar construction is used for the windows on the C(>2 sample cells.
Mirror M23 is self-aligned by spring loading it against three points
that contact the front spherical surface. The sample cell temperature
may differ from the baseplate temperature as a result of the sample
temperature. The sample cell is held rigidly to the optics baseplate
near W21. The other end of the cell is allowed to mo/e along the cell
axis. This allows for stress-free differential thermal expansion of
the cell. Mirror M23 is mounted to the sample cell. Since the cell is
mounted to the baseplate near W21, which is near the center of curva-
ture of M23, the alignment is insensitive to small movements of the
sample cell relative to the baseplate.
-19-
-------
Apertures A21, A22, and A23 reduce the amount of scattered energy that
can reach the photodetectors. In addition, A21 protects the surfaces of
M21 and M22 during the removal and reinstallation of window W21.
The sample gas coming from the C02 cells flows into and through the
10 cm cell and then into the 100 cm cell. It is exhausted from the
system after flowing lengthwise through the 100 cm cell. The sample is
directed through the shorter cell first to allow the observance of
abrupt changes in gas concentration before any mixing can occur in the
longer cell. Although not shown in the figure, a small portion of the
sample gas flows into the cell near W21. This prevents the possible
formation of a stagnant pocket or eddy of gas near W21. Similarly,
although most of the sample is exhausted from the cell just in front
of M23, a small flow of gas is exhausted from behind M23. This prevents
the small volume between M23 and the cell end cap from containing
stagnant gas that could contaminate samples added later.
-20-
-------
SECTION IV
ELECTRONICS
GENERAL DESCRIPTION
The electronics for the three-channel gas correlation spectrometer were
designed for good stability and low noise while preserving a reasonably
small package size and weight. The photodetector is the limiting noise
source in each channel. Component selection was limited to off-the-
shelf items. The system was designed to operate within a laboratory at
room temperature. To minimize costs, the electronics for each of the
three channels are as much alike as possible.
A block diagram of the electronics for a single channel is presented in
figure 1. The photodetector output consists of a 360 Hz carrier modula-
ted at 30 Hz. The ratio of the modulation amplitude to the carrier
amplitude is directly proportional to the amount of gas within the
sample cell. Phase locked synchronous demodulation of the 30 Hz signal
offers the best system signal-to-noise. The 30 Hz reference required
for synchronous demodulation of the 30 Hz signal is also employed to
synchronously modulate the carrier as necessary to cancel any residual
30 Hz modulation on the carrier when no sample is present.
The three electronics units and the power supply unit are mounted in a
standard 19-inch (48.3 cm) electronics rack. The front panel heights
are 8.75 inch (22.2 cm) for each electronics unit and 7 inches (17.8 cm)
for the power supply unit. Line power is furnished through a single
junction box within the electronics rack to prevent possible ground
loops between systems. The power supply unit is mounted in the top of
the electronics rack to prevent the heat dissipated from this unit from
unnecessarily heating the other units. The front panels of the power
supply unit and one of the electronics units are presented in figure 7.
The signal and power cables that connect the optics assembly to the
electronics unit are about 3.5 meters long.
-21-
-------
'
sim *•'«"'
PHILCO
/
0
—— GAIN '•' I— MODULATION
SIG I SIQS HBF SIGNAL CASHIER
Figure 7. Power supply unit above a single electronics unit.
-22-
-------
OPTICS UNIT
A schematic illustration of the electronics contained within the optics
assembly that are used by a single signal channel of the system is pre-
sented in figure 8. The gain of each preamplifier is set to effect a
carrier signal amplitude that is about 1 V peak-to-peak. Five photo -
detectors and preamplifiers are used for the three channel system.
A 45 V dc battery pack supplies the bias voltage to the photoconductive
PbSe detectors. Jacks allow monitoring of the battery voltage in both
the load and the no-load conditions. The total current drain from the
bias battery by the photodetectors is about 40 u-A when the photodetec-
tors are thermoelectrically cooled. This should allow a battery life
in excess of six months. A low-pass filter with a 1.1 second time con-
stant in each channel prevents a possible crosstalk between channels
as a result of coupling through the bias voltage supply. This filter
also reduces any noise generated by the bias supply. A diode in series
with the battery protects the electrolytic filter capacitors from
damage that could occur if the battery polarity is reversed. A
capacitor across the diode lowers the impedance of the protection cir-
cuit. The batteries are contained within a fluid tight chamber to
prevent possible contamination of the optics system with battery
electroyte.
The 30 Hz reference pickup shown in figure 1 consists of a light
emitting diode (LED) and a phototransistor used in the common emitter
mode. The beam of light between the LED and the phototransistor is
chopped at 30 Hz by a single-blade chopper mounted on the correlation-
cell shaft. The output impedance is about 490 .n. and the amplitude of
the reference signal is about 400 mv peak-to-peak.
The reference system is supplied with -15 V dc from a power supply in an
electronics unit. The photodetector preamplifiers are supplied with +15
and -15 V dc from the same power supply. Separate cables connect the
reference and the preamplifier systems to the electronic unit power
supply. Further isolation between systems is accomplished with resis-
tance-capacitance filters that are installed next to the reference
pickup and the preamplifiers.
POWER SUPPLY UNIT
The Nernst source, source heater, and the correlation cell drive motor
receive their 60 Hz power from the power supply unit. The source
current is controlled at the power supply unit and is displayed on a
front panel meter. The amount of current that flows through the source
is limited by either one, two, or three ballast tubes that are switched
-23-
-------
TO OTHER
CHANNELS
DETECTOR
BIAS
NEDA #709
45 VDC BATTERIES
BATTERY
TEST
JACKS
TO ELECTRONICS
UNIT
PHOTODETECTOR ON '
THERMOELECTRIC
COOLER I
PREAMPLIFIER |
Figure 8. Schematic illustration of a single channel of the
photodetector electronics within the optics assembly.
-24-
-------
in parallel between the source and the 115 V ac power line. The fan
that cools the Nernst source is controlled by this same switch. The
fan is operated at 115 V ac except in the standby (STBY) mode when its
voltage is reduced to about 70 volts. This approximately halves the
fan speed. The source heater is supplied from a 60 V ac transformer in
the power supply unit when the spring loaded source start switch is held
on. In normal operation, the synchronous 1800 rpm motor is powered by
80 V ac supplied by a transformer in the power supply. The motor voltage
is 115 V ac when the spring-loaded motor switch is held in the start
position. The thermoelectric coolers for the photodetectors are powered
from a 5 V dc power supply mounted in the power supply unit. A 3 n
power resistor in series with each thermoelectric cooler converts the
fixed-voltage supply to a fixed-current supply. Each of the five coolers
draws about 1.8 amperes.
ELECTRONICS UNIT
General
Each of the three gas channels employs an electronics unit to process
the detector signals. A functional block diagram of a single unit is
presented in figure 1. The electronic unit is subdivided into a 30 Hz
reference section, 360 Hz carrier section, and a 30 Hz signal section
Most of the electronics for these sections are contained on plug-in
circuit cards with one card for each section. The circuit cards are
interchangeable between the three electronics units. The differences
among the three units are in the front panel controls. These differences
reflect the use of a single photodetector for the C02 system and a pair
of photodetectors for the CH4 and CO systems. Switching between photo-
detector preamplifiers is accomplished with a toggle switch on the front
panel of the electronics unit. Figure 7 illustrates the front panel
controls for a unit for use with either the CO or the CH4 systems.
Reference Section
The reference section of the unit is used to amplify, phase shift, and
transform the signal from the reference pick-up into a form compatible
with the remainder of the electronics unit. The input reference signals
are amplified to a set level by the first stage of the reference section.
The gain of this amplifier is adjusted by the operator and is not critical.
The reference signals are then further amplified by a tuned amplifier
with a Q of 5. This amplifier is followed by two adjustable phase shifters
connected in series. The second phase shifter thus tracks any adjustment
made in the first one. The phase shifters are used to obtain the proper
phase between the reference and the photodetector signal. The phase can
be adjusted easily without test equipment and should remain stable. The
output stages of the phase shifters transform the sinusoidal reference
signals into large amplitude square wave reference signals that drive
the gates of field effect transistor (FET) switches in the modulation
balance assembly and in the 30 Hz synchronous demodulator.
-25-
-------
Carrier Section
The 360 Hz carrier signal section amplifies and normalizes the signals
to a constant average carrier level. This section is also employed to
balance out erroneous signals that can result from an imbalance in the
transmissivities of the correlation cells.
The input signal from the photodetector preamplifier is amplified with
a bandpass amplifier. The Q is about 3.5 and the center frequency is
360 Hz. The operator sets the gain control potentiometer of this ampli-
fier to effect an indicated signal level that is near midscale on the
front panel meter when the sample cell is filled with clean air. The
adjustment is not critical since the automatic gain control (AGC) has a
dynamic range of more than 5 to 1. The gain controlled stage of the AGC
follows the tuned amplifier.
The modulation balance assembly follows the gain controlled amplifier.
This assembly consists of an adjustable voltage divider and an FET
switch that shunts part of the carrier signal to ground when the switch
is closed. The FET switch is activated at 30 Hz by the reference sys-
tem. This results in a square wave modulation at 30 Hz of the 360 Hz
carrier. The voltage divider is adjustable to allow the FET to shunt
up to 10 percent of the carrier signal to ground when the switch is closed.
The adjustable system includes two potentiometers mounted on coaxial
shafts. One potentiometer is for fine adjustments, the other for coarse
adjustments. The adjustment resolution of the system is +2 x 10"5 of the
carrier amplitude. A front panel mounted switch allows the modulation
balance system to be turned on and off. This three position switch also
allows the inserted modulation to be either in phase or 180 degrees out
of phase with the 30 Hz signal modulation.
The 360 Hz carrier demodulator follows the modulation balance assembly.
The demodulation is performed by a precision full-wave rectification
circuit.
The average demodulator output voltage is compared with a fixed ref-
erence voltage by the AGC control amplifier. Any error between the two
voltages is coupled back to the gain controlled amplifier with a feed-
back loop. The closed loop response time constant is one second.
Shorter time constants would remove an excessive amount of the 30 Hz
signal modulation from the carrier. Longer time constants would
increase the fractional error in the 30 Hz signal modulation that
could occur due to large abrupt changes in transmissivity of the sample
gas. The output of the carrier demodulator can be monitored on the
panel meter and can be recorded from the carrier output jack. The AGC
can be turned off with a switch on the front panel. The gain of the
gain controlled amplifier is then set by a potentiometer that is mounted
-26-
-------
beside the switch. With the AGC off, the 360 Hz system output is a
measure of the sample gas transmissivity over the spectral bandpass of
the system.
When the manual gain control is set to effect the same average carrier
level as is maintained by the AGC, then the sensitivity of the 30 Hz
system to small concentrations of sample gas is the same as when the
AGC is operating. With this gain setting, the operator can block the
radiation between the source and the photodetector to observe the
character and the relative amplitude of the noise generated by the
photodetector and the electronics systems.
Signal Section
The 30 Hz section of the unit amplifies and isolates the 30 Hz signal
modulation. This is performed with a phase locked synchronous ampli-
fier and demodulator tuned to 30 Hz. The demodulated output of the
carrier serves as the input to the 30 Hz signal section. Much of the
720 Hz ripple that results from the 360 Hz demodulation process is
removed by a three-stage low-pass filter.
The range switch is a stepped attenuator that sets the signal output
range. The signal gains of the attenuator are XI.0, X0.3, XO.l,
• • •, X0.0001, and off. The off position is used in setting the zero
output level of the 30 Hz system. A single curve of the sensitivity
of the system to different amounts of sample gas can be drawn by
normalizing the output signal Va/Vc by the attenuator gain g. The
calibration potentiometer is a variable gain control that follows
the range switch. A calibration constant Ca is proportional to this
gain. It is adjusted by the operator such that the system response
curve of V* = (Va/Vc)/(g Ca) versus gas concentration is in agreement
with the indicated output for the calibration sample gas that is
employed.
The calibration potentiometer is incorporated into the feedback loop
of a high gain bandpass amplifier with a Q of 8. The signal is further
amplified and conditioned before being demodulated. A notch filter in
front of the demodulator reduces any 60 Hz modulation that might be
present on the carrier. The synchronous demodulator only demodulates
the 30 Hz portion of the signal that is in phase with the 30 Hz
reference. Any residual 60 Hz signal is passed by the synchronous,
phase-locked, demodulator as ripple and does not affect the smoothed
signal output. An amplifier with an adjustable time constant smooths
the demodulated signal before it is presented on the front panel meter
or at the signal output jacks. The time constants available are 0, 0.1,
0.3, 1.0, and 3.0 seconds. In the 0 position the response time is
limited to about 0.08 second by the combination of low-pass filters,
notch filters, and bandpass amplifiers.
-27-
-------
A meter mounted on the front panel of the electronics unit allows the
operator to monitor the reference level, the input signal level, the
carrier level, and the output signal level. The reference and input
signal levels are adjusted to effect a mid-scale indication on the
meter.
Identically labeled input and output jacks are provided on both the front
and back of the electronics units. The jacks are wired in parallel. A
10 K n resistor is in series with each pair of output jacks. The outputs
can be grounded without affecting the operation of the system. The full
scale signal output is -1.0 volt. The normal carrier output is about
-1.5 volts. Shunt resistors across the output jacks will reduce these
voltages to match the ranges of low level recorders. For example, a 10 Q
shunt resistor will act as a voltage divider with the 10 K fi resistor to
reduce the full scale signal output from -1 volt to -1 millivolt.
For clarity, the above description was limited to a single channel of
electronics. Since two different sample cell lengths are employed for
each gas species, two separate modulation balance assemblies are
included in each electronics unit. The CO and the CHA detection systems
also employ two different photodetectors and preamplifiers for each gas
species. Separate adjustable gain controls were provided for each
signal input. Toggle switches are used to switch from the signal inputs
and modulation balance assemblies used with the short sample cell to
the corresponding ones for the long sample cells. This construction
allows the operator to switch rapidly between the different sample cells
without changing the settings of the signal input gains or the modulation
balance assemblies.
-28-
-------
SECTION V
TEST DATA
The sensitivity of the gas correlation spectrometer to various gases
has been measured. Plots of the normalized output, y* , versus the
sample concentration are presented in figures 9, 10, and 11 for CO, CH4,
and C02> respectively. For a given gas concentration, the system out-
put is a function of the setting of the calibration potentiometer and
the range switch. Both of these controls adjust the gain of the 30 Hz
system. The calibration controls were set to effect a full scale out-
put of about -1 volt for sample gas concentrations of 100 ppm of CO,
50 ppm of CH^, and 1% of 0)2 when the samples were in the long sample
cells. With the AGC operating, the output of the system Va/Vc was
then normalized by the gain setting, g, of the range switch used for
each measurement and plotted against the concentration of the sample
gases. The ordinate of the curve V* is equal to (Va/Vc)/(g Ca) , where
Ca is a calibration constant that is proportional to the gain of the
calibration control.
The sample gases were prepared by diluting measured amounts of CO,
and C02 with N2« Some of the samples contained both CO and City with
the N2 dilutant. The sample containers were evacuated before use to
prevent contamination of the sample. The samples were mixed with a
mechanical stirring device inside the sample tank. The sample gas flow
rate through sample cells was regulated at 0.2 liters/minute while V*
was being recorded. This flow rate is fast enough to prevent adsorption
or desorption of gas from the walls of the sample cell and the gas
handling system from affecting the concentration of the sample gas. A
slow flow rate also minimizes the amount of sample gas required for each
data point. A much faster flow rate was used to flush an old sample
out of the sample cell with a new sample to reduce the waiting time
between measurements. The gas flow rate is limited by a needle valve in
the input gas line. The flow exhausts from the sample cell to the atmos-
phere, thus the pressure within the sample cell is slightly over one
atmosphere.
-29-
-------
100
o
I M HW- -I-I J_IJJ_U
100 CM SAMPLE CELL
10 CM SAMPLE CELL
100 1000
CO SAMPLE CONCENTRATION
Figure 9. GO detection sensitivity.
ppm
-------
1000 p 1 I I I Mill 1 | | I llll[ 1 | I I Hill 1 I I I Mill 1—| II MM.
100
*>
S
NJ
0.1
0.01
100 CM SAMPLE CELL
0 CM SAMPLE CELL
I n. i I I I Min _ | | | | |||| - 1 I I I Mil - 1 I I I III
ill
0.1
10
100
1000 ppm 104
CH4 SAMPLE CONCENTRATION
Figure 10. CH^ detection sensitivity.
-31-
-------
1000
20 CM SAMPLE CELL
I I I Mill l'l|
0.1
1000
1
104
10 7. 100
105 ppm 106
COo SAMPLE CONCENTRATION
Figure 11. CQ* detection sensitivity.
-32-
-------
Two sensitivity curves are shown for each gas, one curve for each cell
length. Both curves have the same shape but are displaced along the
abcissa by the ratio of the cell lengths. The displacement results from
the output being directly proportional to the product of the gas con-
centration and the sample cell length. This product is the absorber
thickness at a given temperature and pressure. The useful measurement
range is shown by the solid line portion of each curve. The lowest con-
centration indicated is the noise-equivalent-concentration (NEC) of the
channel. The NEC is defined as the concentration that produces a signal
equal to the peak-to-peak noise level. The upper limit of each measure-
ment range is set by the concentration where the slope of the logrithmic
curve is about 0.5. This correspondends to concentrations for which v*
is proportional to the square root of the concentration. Larger gas con-
centrations can be measured, but the accuracy decreases.
The ability of this system to measure small amounts of one gas in the
presence of large amounts of other gas species is one of the main
advantages of the gas correlation system. With the exception of water
vapor, the system is insensitive to other gases in automotive exhausts.
For typical exhaust gas concentrations, the error produced by the inter-
fering gases was either less than the NEC or less than 1% of the amount
of gas being measured. The discrimination ratio is given by the ratio
of the amount of interfering gas to the error signal produced by the
interfering gas. The system exhibits a discrimination ratio of 50,000
against H^O by the CO system. This means that it takes a sample con-
taining 50,000 parts of 1^0 to produce the same signal as a sample
containing 1 part of CO. The CE^ system has a discrimination ratio of
13,000 against H20. A discrimination ratio with a negative sign means
that the interfering gas produces a signal of opposite polarity to the
signals produced by the gas species normally being detected. Since the
error signal may not be directly proportional to the concentration of
the interfering gas, discrimination ratios are measured with concentra-
tions that are similar to those expected in automotive exhaust samples.
Large concentrations of interfering gas may produce an unrealistically
large discrimination ratio.
The discrimination ratios for several individual hydrocarbon species
were measured by placing a pure sample of the interfering gas in a 0.1 cm
long sample cell. This flow-through sample cell was located near the
exit slit of the grating-filter assembly in the beam of radiation that
traverses the 100 cm sample cell. The 100 cm sample cell was filled with
N£« A sample with a concentration of 1000 ppm in the 100 cm cell will
produce the same output as the pure sample in the 0.1 cm cell. The use
of pure samples simplifies the gas handling procedure. The hydrocarbon
discrimination ratio is based on the required number of carbon atoms in
the interfering gas to produce the same signal as a single carbon atom
in a methane sample.
-33-
-------
A hydrocarbon mixture was prepared by vaporizing 31.2 microliters of
Indolene Clear 91 octane gasoline into a 47.3 liter tank containing N0
at J pressure of 1.51 atmospheres. This lead free gasoline was analyled
by EPA^on 11 May 1973. At one atmosphere pressure, the mixed sample con-
tains 486 ppmc in N,,. This mixture in the 100 cm sample cell of the CH,
analyzer produced an indicated output of -0.4 ppm of CIU. This corres-
ponds to a discrimination ratio of -1200. Adding 50 ppmc of ethylene and
50 ppmc of acetylene to this Indolene Clear mixture results in a hydrocarbon
mixture that is representative of typical automotive exhaust minus the
methane content. Increasing the carbon content of the Indolene Clear
sample from 486 ppmc to 586 ppmc by the addition of ethylene and acetylene
to the dilutant gas does not change the indicated output of the system
Thus, the discrimination ratio against typical exhaust minus methane is
about -1450. This hydrocarbon mixture did not produce a detectable out-
put on the CC-2 or the CO channels of the system.
Table 1 presents the noise levels of the system. Each channel is
detector noise limited. The peak-to-peak noise for a 1 second time
constant is expressed in terms of the NEC. The noise levels are approxi-
mately inversely proportional to the square root of the time contants.
The NEC values for the short sample cells are larger than the NEC values
for the long sample cells because the absorber thickness for a given gas
concentration is less by the ratio of the cell lengths. Further, the
optical throughput of the 10 cm sample cell is only about 1/3 of the
throughput for the 100 cm sample cell and the photodetectors with highest
detectivity were used with the long cells.
evewM exhibits a slow drift in the zero signal output
control ad tr ,* P^iodically corrected by the modulation balance
control adjustment. The drift rate is rapid during the first few
Adrift °rrfi0n after tUrninS on bot* ^e source and the motor.
The drift rate decreases to a very small value after about 2 hours
of warm-up operation. The zero-drift was less than the equivalent of
1 ppm of gas in the 100 cm sample cell for the CO and the CH/, systems
over a 2-1/2 hour period after a warm-up period of equal time. The
average drift rate after this period was thus less than 0.4 ppm/hour.
-34-
-------
Table 1. SYSTEM NOISE LEVELS
i
10
Gas
CO
CO
CH,
CH4
Cell Length
(cm)
100
10
100
10
Time
Constant
(sec)
1
1
1
1
NEC (Noise-Equivalent
Concentration)
(peak-to-peak)
0.5 ppm
35 ppm
0.3 ppm
30 ppm
Norma 1
Operating Range
1 to 1,000 ppm
100 to 10,000 ppm
0.5 to 1,000 ppm
100 to 10,000 ppm
20 1 10 ppm (0.001%) 0.002 to 5%
4 l 50 ppm (0.005%) 0.01 to 25%
-------
SECTION VI
GLOSSARY OF SYMBOLS
A# - an aperture
ac - alternating current
AGC - automatic gain control
BB - a bearing block
Cfl " calibration constant proportional to gain of calibration control
cm - centimeter
(im - micrometer
cm - reciprocal centimeter, a wavenumber unit
CH^ - methane
CO - carbon monoxide
C02 - carbon dioxide
D# - a photodetector
dc - direct current
F# - a filter
f - 30 Hz correlation cell frequency
f - 360 Hz carrier frequency
FET - field effect transistor
-36-
-------
g - relative gain of 30 Hz signal attenuator
G - a plane diffraction grating
GCl - gas correlation cell containing CO and
GC2 - gas correlation cell containing N2 and a neutral attenuator
GC3 - gas correlation cell containing COo
GC^ - gas correlation cell containing N2 and a neutral attenuator
Hz - Hertz
Kn • kilo ohms
0 - ohms
LED - light emitting diode
M# - a mirror
mV - millivolt
N, - nitrogen
NEC - noise equivalent-concentration: the gas concentration that will
produce the same output signal as the peak-to-peak noise level
FbSe - lead selenide
ppm - parts per million
ppmc - parts per million of carbon
Q - measure of bandpass of a tuned circuit
rpro - revolutions per minute
S# - a slit
Tj_ - transmittance of GCl
T2 - transmittance of GC2
T - T when sample gas is CO
co sam
T - transmittance of sample gas
sam
-37-
-------
TY " Tsam when samPle is s°me interfering gas
V - Volt
V - normalized output (Va/Vc)/(g C )
Va - 30 Hz signal amplitude
Vc - average 360 Hz carrier amplitude
W# - a window
% - percent
V - wavenumber
va - wavenumber at the center of an absorption line
Av - bandpass expressed in wavenumbers
ABBREVIATED LABELS ON CONTROLS
AGC - automatic gain control
CAL - calibrate
FUNCT - function
MAN - manual gain, the AGC is turned off
REF - reference
SIG - signal
STBY - standby
T.C. - time constant
0 - phase
-38-
-------
SECTION VII
PATENT INFORMATION
Follwoing are two (2) inventions which have been declared as subject
inventions:
Name of Inventor
Dr. Darrell E. Burch,
Francis J. Gates,
David A. Gryvnak, and
John D. Pembrook
Dr. Darrell E. Burch,
Francis J. Gates,
David A. Gryvnak, and
John D. Pembrook
Title of Invention
Gas Analyzer,
Employing a Grating
and a Gas Selector
Cell
Gas Analyzer
Patent Application
Serial No. and/or
Contractor Disclosure No.
33-059
(A-73-07)
31-154-A
(A-71-29)
Philco-Ford is currently requesting that the Government waive title and
retain only a non-exclusive royalty-free, irrevocable license in order
that the Contractor may file for these inventions and pursue additional
in-house work to further exploit their application in the public interest.
-39-
-------
BIBLIOGRAPHIC DATA
SHEET
1. Report No.
EPA-650/2-73-030
3. Recipient's Acceaaion No.
Instrument to Monitor CH^, CO, and C02 in Auto Exhaust.
J. Report Date
October 1973
7. Aixhord)
D. R. Burch and .T. n
•• Performing Organisation Kept.
No.
9. Performing Organization N>me and Address
Philco-Ford Corporation
Aeronutronic Division
Ford Road
Newport Beach, California 92663
10. Project/Task/Voilc Unit No
II. Contract/Grant No.
68-02-0587
12. Sponsoring Organization Name and Addreis
EPA, Chemistry and Physics Laboratory,
National Environmental Research Center
Research Triangle Park, North Carolina 27711
13. Type of Report it Period
Covered
Final Report
U.
IS. Supplementary Notes
16. Abstracts
An infrared analyzer employing gas cell correlation techniques and
thennoelectrically-cooled photodetectors has been designed and con-
structed to measure the concentrations of methane, carbon monoxide, and
carbon dioxide in automotive exhausts. The dynamic range from the mini-
mum detectable concentration to the maximum concentration that can be
measured accurately for each gas is: methane, 0.3 ppm to 10,000 ppm,
carbon monoxide, 0.5 ppm to 10,000 ppm, and carbon dioxide, 10 ppm to
250,000 ppm. The wide dynamic range is made possible by employing two
different sample cell lengths for each gas. The concentrations of the
three gases are measured simultaneously and independently. Discrimi-
nation against other gases in the automotive exhaust is very good.
17. Key Words and Document Analysis. )7o. Descriptors
Air pollution
Measuring instruments
Infrared spectrophotometers
Exhaust gases
Methane
Carbon dioxide
Carbon monoxide
7b. Identifiers/Open-Ended Terms
Air pollutant measuring equipment
Gas correlation spectrophotometer
7e. COSATJ Field/Group 13B, 7B, 7C
I. Availability Statement
FONM NTIS-19 IREV. S-721
19. Security Class (This
Report)
UNCL.
2v> security
'
UNCL
ASSIFIED
21. No. of Pages
46
-40-
U1COMM-DC UII2-P71
------- |