WO1983000065A1

WO1983000065A1 - Automatic tracking system for a second derivative spectrometer

Info

Publication number: WO1983000065A1
Application number: PCT/US1981/000867
Authority: WO
Inventors: Inc. Lear Siegler
Original assignee: Price, Gerald, W.; Polhemus, Carl, E.
Priority date: 1981-06-24
Filing date: 1981-06-24
Publication date: 1983-01-06
Also published as: AU7414481A

Abstract

An optical spectrometer (10) for measuring low concentrations of gas constituent, such as SO2, provides automatic tracking of an absorption band of the constituent. A stepper motor (26) rotates an optical frequency modulator (24) in the spectrometer through a preset range to vary the frequency of the light passing through a gas sample (12) about an adjustable center frequency. The ouput of the spectrometer, which varies in amplitude with the change in frequency of the light due to the absorption characteristic of the gas, is filtered by a digital filter (64) and stored as a set of digital values in memory (44) of a digital processor (40). The maximum peak value of a set of peak values corresponding to different selected center frequencies is used to control the stepper motor to control the angle and hence the center frequency of the optical modulator to track the absorption peak of the constituent gas.

Description

AUTOMATIC TRACKING SYSTEM FOR A

SECOND DERIVATIVE SPECTROMETER Field of the Invention

This invention relates to optical spectrometers, and more particularly, to an automatic tracking control for a second derivative type spectrometer. Background of the Invention

In copending United States application Serial No. 06/216,636 assigned to the same assignee as the present application, there is described an optical system for a second derivative type spectrometer. This optical system is designed to provide a light source for passing light through a gas sample containing a constituent gas such as SO2 and measuring the change in light intensity as a function of light frequency. A quantitative measurement of the amount of constituent gas can be made by measuring the change in intensity as a function of frequency of the light passing through the sample. One method of deriving quantitative information from a spectrometer is known as second derivative spectroscopy in which the curvature or rate of change of the slope of the varying light intensity as a function of frequency is measured. Regions of large curvature only occur where narrow band absorption takes place. Certain gas molecules such as SO2 exhibit such narrow band absorption characteristic. Second derivative spectroscopy requires a light source which is modulated in frequency over a frequency band that is centered on the peak frequency of the absorption band being analyzed. Various optical arrangements have heretofore been proposed for modulating the frequency of the light falling on the entrance slit to the spectrometer. See, for example, U.S. Patents 3,565,567 and 3,756,721 which show mechanically reciprocated optical systems for shifting the frequency. One problem with such known systems has been to adjust the center frequency of the frequency modulated light source to maintain the center frequency at the peak of the absorption band being analyzed. Any variations in amplitude and wavelength of the incident light adversely affects the signal-to- noise ratio and linearity of the spectrometer and can be critical in making accurate quantitative measurements of a constituent compound of very low concentration, for example, a concentration of the order of several parts per billion. «

Summary of the Invention

The present invention is concerned with an arrangement for automatically tuning an optical spectrometer to the desired absorption band. This is accomplished, in brief, by providing a frequency modulator having a stepping motor for stepping a thick quartz crystal through a predetermined number of steps starting from an established null position. The change in angle produces a shift in frequency of light falling on an exit slit after passing through the quartz crystal. The successive intensity output levels with each step of the quartz crystal angle are passed through a digital filter synchronized with the stepping of the motor, and converted to digital values by a ratiometric analog-to- digital converter. A digital processor analyzes the digital values to determine the stepping position that corresponds to the peak absorption level. The stepping motor is then adjusted in relation to a null position and stepped by an up-down counter through an equal number of steps on either side of the position corresponding to the absorption peak position of the stepping motor.

Brief Description of the Drawings

For a better understanding of the invention, reference should be made to the accompanying drawings, wherein: FIG. 1 is a graphical representation useful in explaining the operation of a spectrometer;

FIG. 2 is a block diagram of the spectrometer system of the present invention;

FIGS. 3 and 4 are schematic diagrams of the preferred embodiment of the present invention;

FIG. 5 is a flow diagram of the microprocessor program used in connection with the present invention; and

FIG. 6 is a graphical representation useful in explaining the operation of a spectrometer.

Detailed Description

Referring to FIG. 1 in detail, curve A shows the light intensity versus wavelength characteristic of an absorption band for a particular compound, such as Sθ2« The absorption band can be relatively narrow, for example, about 10 to 30 angstroms wide. The location of these absorption, bands in the frequency spectrum is uniquely determined by the physical properties of the particular molecular compound. The width and sharpness of the peak provide a means of identifying the presence of a constituent gas while the magnitude of the peak is a measure of the concentration of the gas .

In second derivative spectroscopy, the wavelength is varied or modulated at some modulating frequency about the peak frequency of the absorption band. Curve B in

FIG. 1 represents a modulating signal which, in one cycle, varies the frequency of the incident light from the center frequency λ Q to a minimum frequency λ g -Δλ to a maximum frequency of λg + Δλ and back to λ _Q (see points 1, 2, 3, 4, and 5 in-yiG.l). By thus modulating the frequency, the output intensity varies in the manner shown by curve C, with points 1, 2, 3, 4, and 5 corresponding to the minimum and maximum intensities resulting from the frequency variation of the input light. It will be seen that the maximum absorption (minimum intensity) occurs at points 1, 3, and 5, corresponding to an input frequency λ Q# which is the center frequency of the absorption band. It will be seen also that the amplitude of the intensity fluctuation represented by curve C is proportional to the magnitude of the dip in the intensity curve produced by the absorption band. Thus the amplitude of curve C can be used to provide a quantitative measure of the amount of the constituent gas which is present. Also, it will be noted that for each cycle of the modulating signal, the intensity modulation curve goes through two complete cycles. Therefore, if the frequency of the incident light is continuously modulated at a given frequency, the intensity fluctuation due to narrow band absorption occurs at twice this frequency. Since the modulation frequency is known, the frequency of the intensity modulation is also known and can be separated out from fluctuations of intensity from other sources (noise) by a pass-band filter. It can be shown that the second derivative voltage signal is proportional or linearly related to the gas concentration. To obtain a signal proportional to gas concentration, the second derivative voltage signal is divided by the total output voltage of the intensity measuring element. However, the accuracy of such measurements requires that the center frequency λ Q be set exactly at the peak frequency of the absorption band. Automatic*tracking of the absorption peak by the modulator is provided by the spectrometer control arrangement shown in the block diagram of FIG. 2. The optical spectrometer, indicated generally at 10, is described in more detail in the above-identified copending application. The spectrometer includes a sample cell 12. A sample gas is drawn from an inlet by a vacuum pump 14. When a solenoid 16 is closed, the outlet of the pump 14 is recirculated through a filter 18 back to the sample cell so that there will be no net flow into the sample cell from the inlet. Light from an ultraviolet light source 20 is focused on an entrance slit 22. The projected image of the entrance slit is modulated by a thick quartz window 24 rotated by a stepping motor 26. The light passing through the quartz window is refracted or transversely displaced by an amount dependent on the position of the stepping motor and the refractive index of the window. The stepping motor oscillates the window through an angle of 13.5° at a frequency of 5.3 hz about a central point which is adjusted to track the peak of the absorption band of the gas constituent being analyzed, all in a manner hereinafter described in more detail.

Light from the entrance slit is spectrally dispersed by a concave holographic grating 28 onto an exit slit 30. The grating is positioned such that light from the entrance slit will result in the desired wavelength at the exit slit. The exit wavelength will change with the apparent displacement of the entrance slit resulting from rotation of the quartz window. The wavelength^' shift at the exit slit is about 1 angstrom per degree of rotation of the* stepping motor 26.

Light from the exit slit 30 is directed in multiple paths through the sample cell 12, exiting the sample cell to a photomultiplier tube 32. The photomultiplier tube 32 provides an output signal which is proportional to the intensity of the incident light. The sample cell 12 is an airtight and light-tight enclosure. Any gas, such as air in which a constituent gas is present, will absorb some light and reduce the intensity at the photomultiplier tube depending on the concentration of the constituent gas within the sample cell.

A span cell 34 containing a known concentration of the constituent gas, e.g., SO2, can be moved in or out of the light path between the sample cell and the photomultiplier tube by a linear actuator 36. The span cell, with its known concentration of the constituent gas, provides a means of calibrating the spectrometer in order to obtain quantitative measurements of the concentration of the particular gas in the sample cell. The photomultiplier tube is driven from a high voltage power supply 38 which can be adjusted by an external signal to correct for any losses in light level that normally occur due to the lamp aging, contamination on the optics, or slight misalignments.

Control of the spectrometer 10 is provided by a microprocessing unit (MPU), indicated generally at 40, which includes a central processor and timer 42 that communicates by a data and address bus with a memory 44 and with timers and power down control 46.^" The memory 44 includes a RAM for storing data and an EPROM or - storing the program including the operating system and diagnostics. Tlje multiprocessor 40 communicates with an I/O circuit 50 through an interface 52. The I/O circuit provides the processor with the ability to monitor and display analog voltages, to monitor discrete events such as contact closures, display results to the user, and provides the processor with digital I/O for monitoring and controlling logic level events. Input analog signals are applied to an analog-to-digital converter 54 for providing digital input information to the processor from various analog signals generated by the spectrometer, such as the output derived from the photomultiplier tube 32. Digital-to-analog converters 56 convert digital output information from the processor to analog signals for controlling strip recorders and the like. A display panel 58 provides input by way of control switches and the like, and provides output by way of a printer, a digital display, and indicator lamps. A digital I/O circuit 60 provides binary coding and decoding for monitoring or providing an on-off signal on a plurality of individual input and output lines.

A power supply board 62 provides control of the spectrometer 10 by the microprocessing unit 40. The output signal from the photomultiplier 32 is passed through a digital comb type band pass filter 64. The output of the filter 64 includes an ac signal corresponding to the second derivative of the light intensity from the sample cell to the analog-to-digital converter 54. It also provides a dc signal corresponding to the average light level on the photomultiplier tube to the converter 54 as a reference level. The digital filter is synchronized with a scanner drive 66 which also controls the stepping of the motor 26. The scanner drive, as hereinafter described in detail, receives a null sensor signal from the stepping motor 26 which signals when the motor is in a zero reference or null position. The scanner drive 66 responds to the null position sensor on the stepping motor 26 and position information from the microprocessing unit in the form of serial output data

(SOD) to position the quartz window 24 at the proper angle and then causes the window to oscillate at a 5.3 hz scan rate over a predetermined angle to frequency modulate the light passing through the exit slit 30. The power supply 62 includes a linear actuator drive 68 for operating the linear actuator 36 to move the span cell 34 into or out of the light beam exiting from the sample cell. The linear actuator drive 68 responds to a span cell signal from the microprocessing unit. The lamp source 20 is driven from a lamp regulator and power supply 70 which can be turned on and off by a signal from the microprocessing unit.

Referring to FIG. 3, the scanner drive for controlling the stepping motor is shown in more detail. The four phases 80, 82, 84 and 86 of the stepping motor are driven in conventional manner by four drivers, indicated at 87. The drivers are controlled to step the motor in response to the four bit words shifted out of a recirculating shift register 88 to advance the stepping motor in a conventional half step mode. By reversing the shift register 88, the sequence in which the four bit words are read out of the shift register is reversed, thereby reversing the direction in which the stepping motor is stepped. The direction in which the shift register 88 is shifted is controlled by a flip-flop 94. Depending on the setting of the flip-flop 94, the shift register 88 will either shift forward, causing the stepping motor to rotate in one direction or to shift in the reverse direction, causing the stepping motor to step in the reverse direction.

To operate the stepping motor in a scanning mode, the flip—flop 94 is controlled by an up-down counter 96 and the flip-flop 98 that controls the direction of the counter 96. When the counter 96 is activated, it initially counts up at a clock rate (CLK-2) double the shift rate (CLK-1) of the shift register. When the counter counts up to 16, it reverses the flip-flop 98, which in turn controls the counter to count back down. When the counter 96 counts back down to zero, it resets the flip-flop 98 and at the same time reverses the flip-flop 94, causing the direction of the shift register 88 to change. 'The steppng counter advances in one direction for 16 steps in one direction, corresponding to one complete up-down cycle of the counter, and then reverses for 16 steps in the opposite direction, causing the quartz window of the modulator to cyclically scan the wavelength of the light over a predetermined range of frequencies. Typically, each step of the stepping motor produces a 1 angstrom shift in frequency of the light incident on the exit slit and corresponds to a rotational step of 0.9° of the stepping motor.

The angular position of the stepping motor about which the scanning takes place is determined by setting the point at which the counter 96 is activated in terms of the angular position of the stepping motor. This is accomplished by providing an adjustable delay between the activating of the counter 96 and the rotation of the stepping motor through a null position. An optical null detector 100 provides an output pulse when the stepping motor rotates through a zero reference or null position. This null pulse, after being synchronized with the clock source by a flip-flop 102, is used to activate a down- counter 104 which has been preset in a manner hereinafter described to a predetermined delay value, corresponding to a predetermined number of steps of the stepping motor. With the enabling of the counter 104, the counter is counted down in synchronism with the shifting of the register 88 and when it counts down to zero, the up-down counter 96 is enabled, thereby initiating the oscillating or scanning action of the stepping motor in which it alternately advances and reverses through 16 steps. The delay interval is controlled by presetting the down counter 104 to the count condition of an up-counter 106. The up-counter 106 is activated at the start of the scan and is eithr counted up to a predetermined value in response to serial pulses from the SOD output of the MPU or is forced to a preset count condition by the setting of a group of switches 108. In normal operation, the counter 106 is initially set by the switches 108 to provide a delay which will give an aproximate centering of the modulator at the desired absorption peak. The automatic tracking system of the present invention modifies the value in the counter 106 by using the SOD output from the microprocessor to set the counter to an adjusted value. The up-down counter 96 counts up 16 counts and then switches the flip—flop 98 causing the counter 96 to count down 16 counts, at which point it again reverses the flip-flop 98 to repeat the up-down cycle. Each time the flip-flop causes the up-down counter to go from a down count condition to an up-count condition, it switches the flip-irlop 94, causing the shift register 88 to change direction thereby reversing the order in which the stepper motor 26 is stepped by the drivers. Since the shift counter is reversed each time the up-down counter 96 goes through a complete cycle of 32 counts and the shift register shifts at half the rate that the counter counts, it will be seen that the stepping motor 26 advances 16 steps and then reverses itself and steps back 16 steps. This results in the scanning action required to modulate the frequency of the light passing through the exit slit 30 of the spectrometer. The setting of the down counter 104 determines the offset from the null position at which the acanning begins. Therefore-, by changing the initial setting of the down counter 104, the center frequency about which the modulation of the -light takes place can be adjusted. During the calibration mode of operation of the spectrometer, the output signal from the photomultiplier tube 32, after being filtered by the filters 64 and converted to digital values by the analog-to-digital converter 54, is analyzed by the microprocessing unit 40 which then outputs -a value to the up counter 106 to adjust the stepping motor 26 so that the scanning action centers on the peak of the absorption curve of the particular gas constituent being analyzed. The filter circuit 64 is shown in detail in FIG. 4. Referring to FIG. 4, the output signal from the photomultiplier tube 32, which has a waveform corresponding to curve C in FIG. 1 when the scanner is centered approximately at the peak frequency of the absorption curve, is applied to the input of a prefilter 110. The prefilter conditions the- signal to fall within the narrow passband of a digital comb filter 112. The prefilter 110 includes two operational amplifier stages 114 and 116 which are direct coupled to retain the dc component of the output of the photomultiplier tube, which corresponds to the average light level. The output of the second stage 116 is connected to the analog-to-digital converter 54 by line 118.

The ac component is separated out by the third operational amplifier stage 120 of the prefilter 110 and applied to the input of the comb filter 112. The comb filter includes a commutating switch 122 and a decommutating switch 124 which are stepped through 8 steps by decoding the three most significant bits of the up-down counter 96. The digital comb filter 112 operates as a narrow passband (2 hz) filter for extracting the second derivative signal from other ac components and . noise which may be present in the output from the photomultiplier tube 32. By synchronizing the commutator and decommutator switches with the scanner drive, the comb filter passes only the second derivative frequency in phase quadrature with twice the scanner drive frequency. The commutator divides the ac signal into 8 time slices. Each time slice is averaged by an RC filter including resistor 126 and capacitors 128 and 130 and to limit the frequency band to 2 hz. Any signal that is not symmetrical with the 8 time slices is averaged out to zero. Thus the only signal retained in each of the time slices is the required second derivative signal.

The output of the digital comb filter 112 is amplified by an operational amplifier stage 134 and applied to an integrating or smoothing operational amplifier stage 136. The output of th^e ^'filter is then applied to the analog- to-digital converter 54. Because of the narrow band filtering, most frequency components resulting from distortion of the second derivative waveform by any offset between the center frequency of the input scan and the peak frequency of the absorption band are removed, leaving only the fundamental component at twice the scan frequency. Any shift of the center of the scan frequency away from the absorption peak results in a reduction in peak amplitude of the output of the filter. The digital- to-analog converter is conventional in that a sample-and- hold circuit samples the instantaneous amplitude of the input signal relative to a reference level, which is the dc component of the photomultiplier tube as received over line 118 from the filter. See FIG. 4. The amplitude of each sample is then converted to a proportional digital count. The MPU receives the digitized output on command and stores the digitized information.

The tracking of the absorption peak scanner drive 66 is controlled by the microprocessing unit using the program shown by the flow diagram shown in FIG. 5. To analyze the output of the filter to determine when a maximum peak amplitude is achieved, indicating that the scan is centered on the absorption peak. At the start (block 140 in FIG. 5) of the routine, the processor sets an initial value in the up-counter 106 in the scanner drive 66. This sets the starting position (block 142) for the scanning action, in the manner described above in connection with FIG. 3. The program then sets a stored initial maximum amplitude value to zero (see block 144 in FIG. 5). A loop count is initially set (at block 146} to 8 since the program is designed to determine which of 8 different peaks* gives ^"the maximum response. Each of the 8 peaks is tested by moving the stepping motor one step at the start of a scan. The peaks are evaluated by stepping the scan center positions in a predetermined order, but the order is not crucial to the process. Once the loop count is set and the loop is entered, the number of scans required to evaluate a peak is set, as at block 148 in FIG. 5. The number of scans preferably is in the order of 200 scanning cycles per peak measurement. The digital representation of an average peak for the 200 scans is then computed (block 150). This averaging operation can best be understood by reference to FIG. 6 which shows the scanning waveform, corresponding to curve A of FIG. 1 and the filter output corresponding to curve B of FIG. 1. FIG. 6 illustrates the phase quadrature relationship between the filter output and the scan of the light frequency. The digital representations of the waveform at the output of the filter are limited by the processor to 8 digitized sampling points equally spaced at 20 degrees and symmetric about the center (zero crossover point) of the scan as shown in FIG. 6. These 8 values define one peak during the 200 scans, a digitized value is accumulated for each of these 8 samples so that at the end of the scanning operation, the peak waveform is represented by 8 values, each value being the sum of the respective sample points for the 200 scans. Thus the 8 digitized values defining the second derivative waveform peak correspond to the average over the 200 scans. Once the 8 average values defining the peak are computed, the program tests (block 152 of FIG. 5) for any scanning error and if there is no scanning error, the program estimates a single peak amplitude value ' (block 154) from the 8 average values. The estimated peak amplitude js determined by a signal-processing technique used to estimate the height of a pulse of known shape and phase in Gaussian noise, a technique frequently referred to as "correlation and integration". This operation reduces the digitized waveform obtained from the digitized average peak to a single digitized value representing the amount of absorption. The algorithm for this operation is given by the expression

where f( θ ) is the received waveform and g( θ ) is the expected shape of the waveform. The present program performs the correlation and integration steps simultaneously by storing the product of the integration constants and the values defining the assumed waveform, and storing these values in a read-only memory. The integration constants are based on an 8-term Newton- Cotes approximation of the closed type, a technique described, for example, in the "Handbook of Mathematical Functions" by Abramowitz and Stegun published by the National Bureau of Standards, 1964, paragraph 25.4.17. The correlation and integration procedure is used not only in the calibration of the instrument for peak tracking, but is also used to measure the percent of the constituent gas present by using Beers law, which provides that change in intensity of light at the absorption frequency is directly related to the concentration of absorbing gas. Because the concentrations are extremely small, the maximum, value at the selected peak must be extracted from the noise. Such a correlation and. integration technique is an effective way of isolating the peak value from the noise.

For tracking an absorption peak, estimated peak amplitude value «is then compared (block 156 in FIG. 5) with the previous estimated peak amplitude, which during the initial loop has been set to zero. If the new peak amplitude is larger than the previously stored maximum amplitude value, the new peak amplitude is stored (block 158) as the maximum amplitude and the current peak position is stored (block 160) . During the initial loop, this would be the initial starting position that was set in the counter 106.

Continuing with the flow diagram as shown in FIG. 5, after the new maximum amplitude is stored or if the comparison indicates the new peak amplitude is less than the maximum previously stored, the program continues by setting (at block 162) the counter 106 in the scanner drive to the next peak position. The loop count is also 12742 :MRS -18- decremented (at block 164) by 1., If all peak positions- have been analyzed by the program, as determined by the program step represented at block 166 in FIG. 5, the peak tracking routine is terminated (block 168). However, if all of the peak positions have not been analyzed, the program loops back (line 170 in FIG. 5) and the next peak position is analyzed. In this way, when all of the 8 peak positions have been analyzed, the position corresponding to the maximum amplitude is established in the counter 106, and all subsequent modulation of the light frequency by the scanner is exactly centered on the absorption peak of the constituent gas being analyzed. The scan position is maintained during subsequent gas sample measurements, but is recomputed whenever the instrument is again calibrated.

OMH ^y

Claims

WHAT IS CLAIMED IS:

1. A spectrometer for measuring the absorption characteristic of a gas, comprising: a light source, means for modulating the frequency of the light source over a predetermined frequency range at a selected modulation frequency, means directing the modulated frequency light through the gas, transducer means for generating an output signal that changes in amplitude in proportion to changes in intensity of the light passing through the gas, band-pass filter means responsive to said output signal from the transducer means for passing components of said output signal at twice the modulation frequency, means adjusting the center frequency of the frequency modulation means through a predetermined number of steps, means determining the average peak amplitude of the output of the band-pass filter at each center frequency step, and means setting the center frequency of the frequency modulating means to a value corresponding to the maximum of said average peak amplitudes.

2. Apparatus of claim 1 wherein the filter means includes a digital comb filter.

3. Apparatus of claim 1 wherein the modulating means includes a stepping motor, and scanning drive means for stepping the motor alternately in both directions a predetermined number of steps to modulate the frequency of the light source.

4. Apparatus of claim 2 wherein the modulating means includes a stepping motor, scanning drive means for stepping the motor alternately in both directions a predetermined number of steps to modulate the frequency of the light source.

5. Apparatus of claim 4 wherein the digital comb filter includes multiplexing means for switching the input successively through a plurality of separate filters and connecting the filters successively to a common output.

6. Apparatus of claim 5 wherein the multiplexing means is synchronized with the stepping of the motor.

7. Apparatus of claim 6 wherein the means for adjusting the center frequency of the frequency modulation means includes a first counter for reversing the direction of the motor after a fixed number of steps in either direction.

8. Apparatus of claim 7 further including means generating a sync pulse when the stepping motor rotates through a null position, and delay means responsive to the sync pulse for actuating said first counter when the motor has stepped a predetermined number of steps after the generation of the sync pulse.

9. Apparatus of claim 8 wherein the delay means includes a- second counter, means for setting the counter to a Selected value, means starting the second counter in response to said sync pulse, means for starting the first counter when the second counter has counted a predetermined number from said selected value.

10. Apparatus of claim 9 wherein the means setting the center frequency of the frequency modulation means includes means for setting the second counter successively to each of a sequence of values, said means for setting the center frequency including means for determining which of said successive settings produces the maximum peak amplitude, and means for resetting the second counter to the count value corresponding the sequential setting producing the maximum peak amplitude.

11. A second derivative spectrometer for measuring the light absorption characteristic of a gas sample comprising: a light source, means directing monochromatic light from the source through the gas sample, means for modulating the frequency of the monochromatic light directed at the source over a predetermined frequency range, means for adjusting said means for modulating the light to shift the center frequency of said predetermined range, said means including means sensing the variation in the intensity of light from the source received through the sample with change in frequency, means determining the frequency at which the light sensing means indicates .a maximum level of absorption, and means responsive to , the frequency determining means for adjusting the center frequency of the modulating means to correspond to the frequency of maximum light absorption.

O PI

12. The method of calibrating a second derivative spectrometer for measuring the concentration of a constituent gas in which light, before passing through the gas sample, is modulated in frequency about a center frequency corresponding to the peak of an absorption band produced by the constituent gas, and the variation in intensity of the light, after passing through the gas sample is detected and measured, comprising the steps of: initially setting the center frequency at a preselected value approximately at.the peak frequency of a known absorption band, modulating the light frequency at a modulation frequency about said center frequency, detecting variations in amplitude of the light intensity having a frequency twice the frequency of the modulation frequency, measuring the peak amplitude of the detected variations relative to the average level of light intensity passing" . through the gas sample, averaging the measured peak amplitudes over a plurality of modulation cycles, shifting the center frequency by a predetermined amount in a succession of steps, comparing the average measured peak amplitudes for each of said steps of the center frequency to determine which center frequency setting produces the maximum peak smplitude, and setting the center frequency of the modulated frequency light to the maximum peak amplitude setting.

O PI

13. The method of claim 12 further including the steps of: ⁴ⁱ converting the detected variations in light intensity to digital representation of samples of the continuously varying light intensity, selecting digital representations of signal samples symmetrical about each of the center frequency cross-over points of the frequency modulated light for measuring the peak amplitudes over a plurality of modulation cycles.

^E

OMPI

-