Field of the Invention The present invention pertains to spectrophotometric instrumentation utilizing light to measure spectral absorbance characteristics of a sample material, and in particular to Fourier transform analysis of infrared absorbance characteristics utilizing an interferometer and a laser to obtain spectral data.

Background of the Invention The Fourier transform infrared (FT-IR) spectrophotometer consists of two basic parts: (1) an optical system which includes an interferometer through which an infrared light beam is directed before passing the beam through a sample, and (2) a dedicated computer which is used to analyze the spectral information contained in the light issuing from the sample. . The advantage in improved performance of the FT-IR spectrophotometer results from the use of the interferometer, rather than a grating or prism, to obtain variance in wavelength of the infrared beam applied to the sample to measure spectral characteristics. An interferometer permits measurement of the entire spectral profile of the sample, increasing accuracy of analysis, in a fraction of the time previously required.

The operation of a Michelson interferometer as applied to FT-IR spectrophotometry to vary and analyzing infrared light wavelength passing through a sample is well known. The interferometer consists of two perpendicularly arranged optical paths, each having a reflector or mirror positioned at its end to reflect light traversing the path. The mirror of one path is fixed. The mirror of the other is longitudinally movable

to increase or decrease the length of the light path. An infrared light beam entering the interferometer is optically split into two components by a beam splitter so that a separate component of the beam will traverse each optical path. After reflection of each light beam component and redirection along its respective path, the components are recombined through the beam splitter to constructively and destructively interfere. The reconstructed beam is directed through a sample and focused onto a photodetector for measurement of intensity and intensity variance of the range of frequencies in the issuing beam.

The intensity characteristic of any selected frequency of the reconstructed light beam depends in part on the difference in length of the optical paths over which the beam components travel. Generally, when the movable mirror is axially moved or scanned at a constant velocity, the intensity of an emerging light beam will modulate in a regular sinusoidal manner for any selected wavelength of light passing through the interferometer.

A typical infrared light beam emerging from the interferometer is a complex mixture of modulated frequencies due to its polychromatic nature. After the infrared light beam has passed through a sample material, it can be detected to determine specific wavelengths of light which have been absorbed by the sample. This is accomplished by measuring change in the regular sinusoidal intensity pattern expected when the light beam leaves the interferometer. The measurement of the differences in the characteristics of sinusoidal patterns for each light wavelength composing the emerging beam indicates those wavelengths of light which are absorbed by the sample. Infrared light absorbance characteristics measured provide spectral data from which the matter comprising the sample can be determined.

The output signal of the detector which measures the intensity modulation of the emerging light beam can be recorded at very precise intervals during scanning of the movable mirror, to produce a plot known as an interferogram. The interferogram is a record of data points indicating the output signal produced by the infrared photodetector as a function of the difference in length of the optical paths traversed by the components of the infrared beam passing through the inter- ferometer. Successive scans of the sample are obtained and co-added to obtain an average interferogram having improved signal-to-noise characteristics. The average interferogram provides information and data relating to the spectral characteristics of the sample material. After mathematical preparation, a Fourier transform calculation is performed on the inter erogram to obtain a spectral fingerprint of the sample composition. The results are compared with known reference data to determine the composition of the sample.

Most Fourier transform techniques require averaging of a large number of interferograms in order to obtain accurate results. As many as 32 to 50 scans of the movable mirror during which measurements are taken may be averaged. It is important that an interferogram be precisely reproducible in order to maintain accuracy in its averaging with other co-related interferograms. Since an interferogram is generated as a function of mirror position, more accuracy in the interferogram and the applied Fourier transformation will be obtained if more accuracy is obtained in the determination of mirror position at the time when the data points are measured to define the interferogram.

To accomplish accuracy and reproducibility in generating an interferogram, both the timing of the data

point measurement (sampling control) and mirror position change must be very precisely controlled. In other words, the exact position of the mirror along a scan must be determinable when the data point is measured.

Modern systems accomplish sampling control and mirror velocity control and/or mirror position measurement by passing a laser beam through the interferometer, concurrently with the infrared light beam. The laser beam is used to directly measure the movement and/or position of the movable mirror to accurately determine change in path length of the interferometer. Since the laser beam undergoes the same optical splitting by the beam splitter and traverses the same change in optical path as the infrared light beam, the recombined laser beam exhibits a .measurable monochromatic light wavelength displaying an intensity interference pattern containing information about the scan velocity of the movable mirror. The intensity interference pattern may also serve to indicate position of the mirror during a scan and to determine and correlate the collection of data points at uniform intervals of mirror displacement.

In a conventional system, when the movable mirror is moving at a constant velocity, a Doppler shift in light wave frequency is generated in the component of the laser beam traversing the changing length optical path. When the Doppler shifted component is recombined with the component traversing the fixed length path, a modulated frequency beam exhibiting a measurable amplitude modulation or beat signal is produced, yielding a varying intensity or fringe pattern which may be analyzed to determine the mirror position and/or velocity. The beat signal is useful because the

frequency of the laser beam produced by most lasers is much too high for measurement by common detectors .

Conventional systems generally drive the movable mirror at a velocity which provides a 5 KHz amplitude modulation or beat signal frequency in the exiting beam. The beat signal frequency is equal to the magnitude of the Doppler shift in frequency because it equals the difference in frequency between the recombined light beam components after traversing their respective optical paths. At increased mirror velocities, the beat signal frequency will increase providing increased signal resolution while at slower mirror velocities the beat signal frequency will decrease to a point at which it is not distinguishable. Precision with this technique can be maintained to approximately one cycle in 5,000 to provide very accurate velocity and position information.

In a conventional system, however, the movable mirror must be in motion to obtain a Doppler shift in frequency of the light beam traversing its path. Thus motion of the movable mirror is necessary to obtain a measurable beat signal frequency in the recombined light beam. To explain, when the movable mirror is stationary, component light beams traveling along adjacent paths of the interfereometer are recombined to form an identical frequency light beam since no Doppler frequency shift has been introduced in either component. The emerging recombined beam exhibits no intensity modulation and no beat signal. Thus when the mirror is not moving, there is no information contained in the emerging laser beam which can be used to determine mirror position or mirror velocity. This occurs at every instance that the movable mirror is stopped, such as when it reaches the end of its scan and stops to proceed in the other direction.

Furthermore, in a conventional interferome ric system the Doppler shift in frequency generated by a scanning mirror produces the same intensity modulation effect in the recombined light beam independent of the direction of mirror travel. For instance, a 5 KHz amplitude modulation or beat frequency can be obtained for travel of the mirror in either a forward or backward direction. Thus it is impossible to determine the direction of mirror travel from the emerging laser beam, even though the difference between optical paths may be increasing or decreasing. This shortcoming generally requires additional circuitry to obtain an indication of direction of mirror travel so that the exact position of the mirror may be determined at any given time.

Moreover, in conventional systems the amplitude modulation or beat frequency of the emerging light beam becomes very difficult to measure as the velocity of a scanning mirror becomes very slow. For instance, for a 0.3 centimeter per second scan velocity, a beat frequency of 5 KHz is generated in the emerging light beam. However, if the mirror is driven at a scan velocity of 0.03 centimeters per second, the beat frequency is reduced to .5 KHz, which becomes very difficult to measure. Thus, as the scan velocity is decreased, the modulation frequency in the recombined light beam is decreased to a level which is difficult to measure with modern electronic detectors, reducing accuracy and resolution.

An FT-IR spectrophotometer has limited resolution in sample identication determined by its ability to produce and reproduce accurate interferograms. The only dynamic part fundamental to the optical system is the movable mirror of the interferometer. This part greatly determines the

accuracy with which a spectrophotometer can generate interferograms. The accuracy with which the spectrophotometer can analyze a sample is directly related to the accuracy and reproducibility of the interferogram and thus the ability of the instrument to control and determine the velocity and position of the movable mirror.

The conventional use of a laser reference to control and determine the velocity and position of the movable mirror continues to suffer limited precision in determining sample test and mirror scan control. Improvements in the precision with which mirror position can be measured and mirror velocity controlled will necessarily produce significant improvement in the accuracy with which an infrared spectrophotometer can analyze a sample substance.

To accomplish accuracy and reproducibility in generating an interferogram, the mirror bounding the fixed length path of the interferometer must be maintained in optimal alignment with the beam splitter to redirect a light beam to meet with the light beam reflecting from the moving mirror. This is most often accomplished by providing a biaxial adjustment for the fixed mirror so that it may be adjusted about two perpendicular axes to bring the image of its surface into absolute parallel alignment with he reflecting surface of the moving mirror.

It is extremely important that the image of the adjustable mirror be unchangingly positioned in relation to the reflecting surface of the moving mirror. Parallelism must be maintained within one wavelength of the shortest wavelength of light being measured to generate the interferogram. Failure to provide precise

alignment results in decreased resolution in the interferogram produced, reduction in signal-to-noise performance and phase error introduction in the multi- component heterodyned beam leaving the interferometer. Each of these defects substantially reduces the accuracy with which an interferogram can be reproduced and the precision with which the interferogram can be analyzed to determine the sample composition. The accuracy with which a spectrophotometer measures intensity modulation can be no better than the precision limitations of its components. The molecular geometries cannot be accurately determined if the frequency modulations in the light beam are not reproduced and measurable precisely.

Prio art designs for mirror adjustment have used massive structures and extremely fine mechanical adjustments to obtain accurate alignment. Temperature compensation has also been considered to reduce thermal distortion error. It has been found, however, that static alignment cannot assure accurate mirror alignment - throughout a scan of the movable mirror and warmup periods of the instrument. Wobble and support inaccuracies of the movable mirror continually introduce alignment error. Dynamic mirror misalignment results in aperiodic errors in the light intensity measurements of the emerging light beam, which cause unpredictable and accuracy-reducing gliches in the interferogram basesd on these measurements.

Modern systems accomplish mirror alignment automatically by passing a reference light beam, such as a laser beam, through the interferometer with the usable measuring light. The reference beam is used, among other functions, to directly measure misalignment between the fixed and movable mirrors.

Laser light beams have been used most effectively. Since a laser beam undergoes the same disection and traversal of the described optical paths in the interferometer as does infrared light, the recombined laser beam exhibits a measurable interference pattern of monochromatic wavelength containing information indicative of mirror alignment. A measurement of phase difference of wavelength across the width of the laser beam is used to determine the difference in the length of path traversed by one portion of the beam, relative to others. Nonsynchronized phase measurements are indicative of unequal path lengths, indicating mirror misalignment.

In a conventional system, when the movable mirror is moving at a constant velocity, a doppler shift is generated in the component of the laser beam traversing the changing optical path. When the doppler shifted beam is recombined with the component traversing the fixed length path, a modulated frequency beam exhibiting measurable beat frequency is produced. The recombined beam yields a series of varying intensity or fringe patterns which may be analyzed across the cross section of the beam to determine mirror alignment. Conventional systems generally drive the moving mirror at a velocity which produces a 5 Khz modulation, i.e., doppler shift, in the exiting beam. At faster mirror velocities, the modulation will increase providing increased resolution for alignment measurement, while at slower velocities the modulation will decrease.

Precision with this technique can be maintained to approximately one cycle in 5,000.

In a conventional system, however, a movable mirror must be scanning to obtain a doppler shift in the light beam traversing its path, and thus a measurable

modulation signal. When the movable mirror is stationary, the light beams traveling along adjacent paths of the interferometer are combined to form an identical frequency light beam without modulation. Thus when the mirror is not moving, there is no information obtained in the recombined beam which can be used to determine mirror alignment. This occurs at each time the movable mirror reaches the end of its scan and stops before turning around to proceed in the other direction. With prior art auto-alignment systems mirror alignment is lost at the ends of mirror ^" scan.

Furthermore, with conventional systems modulation of the recombined beam becomes very difficult to measure as the velocity of a mirror scan becomes very slow. For instance, for a 0.3 centimeter per second scan velocity, a modulation frequency of 5 Khz is obtained in the recombined light beam. However, if the mirror is driven at a scan velocity of 0.03 centimeters per second, the modulation frequency is reduced to .5 Khz. Thus, as the scan velocity is decreased, the modulation frequency in the recombined light beam decreased to a level which is difficult to measure with modern electronic detectors, providing no control of alignment.

A spectrophotometer has limited resolution for frequency measurement determined by its limited ability to produce an interferogram. The optical system is fundamental in determining the accuracy with which a spectrophotometer can measure spectral data. The accuracy with which the spectrophotometer can analyze a sample is directly related to the ability of the instrument to produce an accurate measure of intensity of the emerging infrared beam. This requires proper and continuous alignment of the fixed and movable mirrors.

Conventional use of a laser reference to obtain mirror alignment continues to suffer limited precision and control ability. Improvements in the precision with which mirror alignment can be measured and controlled will necessarily produce significant improvement in he accuracy which an FT-IR spectrophotometer can analyze a sample substance.

Summary of the Invention The present invention comprises improved mirror alignment control for dynamically aligning the image of the mirror bounding the fixed length optical path in the interferometer, with the reflecting surface of the movable mirror bounding the second optical path. The invention utilizes a laser which generates a laser beam having two component frequencies which when heterodyned produce a constant beat frequency. A closed loop servo control provides constant alignment control in response to a comparison of the phase difference exhibited across the cross section of the laser beam leaving the interferometer. The dynamic mirror alignment control obtains precise stable mirror control through analysis of frequency modulation in the light beam emerging from the interferometer.

A laser beam having two components of slightly differing frequency is obtained by applying a magnetic field, to a helium-neon gas laser. This phenomenon is well known and described as the Zeeman effect. The laser beam having differing frequency components is directed through the interferometer. Each component of the beam is combined with its opposing component after traversing the optical paths of the interferometer. The recombined heterodyned laser beam exhibits a continual intensity modulatin having a beat frequency equal to the difference frequency of the component, plus any doppler-affected

change in frequency caused when the movable mirror is scanning. The continually displayed frequency modulation of the emerging light beam provides continuous information which can be used to indicate the precision of alignment between the fixed and movable mirrors.

A detector array is provided to obtain phase detection at various point throughout the cross section of the laser beam leaving the interferometer. By measuring and comparing the phase of intensity avariation across the cross-section of the light beam it can be determined whether alignment of the mirrors is correct, since misalignment results in a longer or shorter optical path over which the light beam must travel as compared to another, thus effecting the cross-sectional phasing of a wave point.

A triad of piezoelectric elements which longitudinally respond to applied voltage are used to mount the fixed mirror in the interferometr. Control signals obtained through phase comparison of the cross- section of the emerging laser beam are applied to the piezoelements to obtain correction of mirror alignment. Control signals are generated by an electronic servo system to pivot the fixed mirror for adjustment about a central point of the reflecting surface of the mirror so that adjustment of the mirror does not in itself interduce phase change in the light beam. Removal of induced phase change substantially increases occurring in mirror alignment control.

Due to.the advantageous use of the two- frequency laser which yields a continually modulated beam emerging from the interferometer, the alignment of the mirrors can be continually controlled, even when the mirror is stationary. Furthermore, due to the continuous

information signal provided by the system, it is unnecessary to perform frequent alignment calibrations since the mirror alignment can be accurately controlled continually.

Description of the Drawings Figure 1 is a schematic drawing of an interferometric portion of a Fourier transform infrared spectrophotrometer and the tilt servo control circuit which comprises the present invention;

Figure 2 is a schematic representation of the interferometric portion of the spectrophotometer depicting the polarization relationship of the individual component of the two-frequency laser as the laser beam passes through the inteferometer;

Figure 3 is a schematic of the electrical circuit of a first most of the tilt servo control.

Best Mode of the Invention

The interferometric portion of the Fourier transform infrared (FT-IR) spectrophotometer is described with reference to Figure 1. A Michelson interferometer is depicted which comprises a beam splitter 10 positioned to distribute a portion of an incident light beam along each of two perpendicular optical paths 11 and 13. The beam splitter 10 receives a laser beam 16 from a magnetically influenced laser 18, and an infrared light beam, shown bounded by lines 20, generated by an infrared light source 22. Generally, the infrared beam 20 is reflected and collimated by a non-planar mirror 24 for entry into the interferometer, while the laser beam 16 is directly applied to the beam splitter 10 through an opening 26 centrally located in the mirror 24.

The beam splitter 10 reflects a first portion of each beam 16 and 20, along the first fixed length optical path 11, which is bounded by an adjustable fixed mirror 12. The ^" light beams 16 and 20 are reflected by the mirror 12 to return along the optical path 11 to the beam splitter 10. A second portion of each of the light beams 16 and 20 is passed through the beam splitter 10 along the second optical path 13 which is bounded by a movable mirror 14. The movable mirror 14 is longitudinally movable with respect to the optical path 13, to change the length of the optical path within the selected scan range, indicated by arrow 15.

The second portions of each of the light beams 16 and 20 are reflected from the movable mirror to return along optical path 13 to the beam splitter 10, where they are recombined with the first portions of the light beams 16 and 20 returning along the first optical path 11. The recombined portions of the laser beam 16 form a heterodyne beam 30. The light beam 30 contains information of the alignment of the fixed mirror 12 relative to the movable mirror 14, through intensity modulation caused by interference phenomena. The recombined portions of infrared beam 20 form a heterodyned beam 32 which has each individual frequency modulated at a characteristic rate to provide a range of modulated frequencies of infrared light which can be applied to the sample material under analysis.

The recombined laser and infrared beams, 30 and

32, respectively, are directed along an exit path 33 of the interferometer in which a reflector 34 similar to reflector 24, is positioned. The reflector 34 receives the collimated infrared heterodyne beam 32 and reflects and focuses the beam on a sample chamber 36. The infrared beam 32 passes through the sample chamber 36 and

reflects from a third mirror 38 to focus on an infrared photodetector 40. The photodetector 40 receives the amplitude modulated infrared beam which is modified by the sample material through which it passes. The photodetector 40 produces an electrical information signal proportional to the modified modulation of the beam which is used to generate an interferogram.

The modulated laser beam 30 passes from the interferometer through an opening 42 in the mirror 34. Beam 30 is directed to a detector 44. Preferably, the detector 44 comprises an array of photodetectors for measuring the intensity of the modulated laser beam 30 at selected points in its cross section. Electrical signals 45 produced by detector 44 are used to control mirror alignment. The signals 45 are applied to a tilt servo control 50 to produce control signals 53.

The control signals 53 are applied to a plurality of piezoelectric elements 101, 102 and 103 which mount the adjustable mirror 12 to the interferometer framework. The piezoelectric elements can be varied in length through application of an electrical signal to control alignment of the mirrors 12 and 14.

The He-Ne laser 10 is magnetically influenced to produce a laser beam with two different frequency components, each having opposing circular polarization. The differing frequencies and polarizations are used to obtain a continuous flow of information in the heterodyne beam 30 leaving the interferometer. Referring to Figure 2, the laser beam 16 having two component frequencies is passed through a quarter waveplate 15 before entering the interferometer. The quarter waveplate 15 converts each of the circularly polarized components into a linearly polarized component. One linearly polarized component

exists in a plane parallel with the drawing, as shown by the bars 17 and has a frequency f^. The other linear component exists in a plane perpendicular to the drawing, as shown by the dots 19 and has a frequency f « This is due to the opposing nature of the circular polarization the incident components exhibit. Thus, the light beam directed into the interferometer consists of two components each having an individual frequency and polarization making them clearly distinguishable from one another.

The first portion 21 of the laser beam 16 reflected along the fixed length optical path 11 passes through a second quarter waveplate 23, reflects from the adjustable fixed mirror 12, and again passes through the quartr waveplate 23 in returning to the beam splitter 10. Passing the first portion 21 of the beam 16 twice through the quarter waveplate 23 acts to rotate the polarization of each component of the beam through a 90° angle about the axis of the beam. Thus, the first component with a parallel polarization plane upon entering the fixed optical path 11 shown by bars 17 returns to the beam splitter 10 with a perpendicular polarization plane as shown by dots 17'. Similarly, the second component having a perpendicular polarization plane entering the fixed length optical path 11 returns to the beam splitter 10 with a parallel polarization plane, shown by bars 19* .

The second portion 25 of the laser beam which passes through the beam splitter 10 and along optical path 13 is reflected from the moving mirror 14 without change in polarization. Each of the components of the second portion 25 of the laser beam may, however, be changed in frequency by a value Δf. This is caused by a

Doppler effect produced in the beam of movement of the movable mirror 14.

Since only like polarized light beams will combine, upon returning to the beam splitter 10, the component of the laser beam having frequency fi which has traversed the first optical path 11, and which has been rotated in polarization by 90°, will recombine with the component of the laser beam having a frequency f ₂ ± Δf which has traversed he second optical path without change in polarization. A resulting recombined wave of one polarization 27 will thus exhibit a frequency of f]_ (f ₂ ± Δf). The resulting recombined wave of the other polarization 29 will exhibit a frequency of (f^ ± Δf) - f ₂ . The recombined waves 27 and 29 are then directed from the interferometer through a polarizer plate 31 which filters out one of the two polarization plan.es of the beam. Thus, the detector 44 will receive a light beam having linear polarization in only one plane, and having a frequency which is continually intensity modulated by the combination of the differing frequency components of the laser beam, one of which may have a Doppler shift in frequency Δf introduced.

It should be noted that a Dopplar shift Δf is introduced to the frequency of each of the components only when the movable mirror 14 is moving. When the movable mirror 14 is held stationary, no Doppler effect is generated. Thus, when the mirror 14 is stationary, the beam component traversing the first optical path 11 will recombine in the beam splitter 10 with the opposing component traversing the second optical path 13 to yield a heterodyne beam which exhibits an intensity modulation or beat frequency exactly equal to the difference between the component frequencies, i.e., f^ - f ₂ . Thus, the detector will continually receive a light beam having a

measurable intensity modulation from which phase relationship across the beam may be detected to determine whether the mirrors are in alignment. Thus due to the continual modulation or beat frequency exhibited by the exiting light beam, an information signal will be produced which permits control of mirror alignment throughout the range of scan and in forward and rearward directions of scan.

The mirror alignment control system is described with reference returned to Figure 1. The laser beam detector 44 comprises an array of photodetectors which measure the intensity of the laser beam 30 leaving the interferometer. The detector array 44 consists of at least three detectors positioned at three different locations in the beam cross section. The signals produced by the photodetectors provide an indication of the intensity difference across the cross section of the beam, which is characteristic of the phase difference across the cross section of the beam. A difference in phase across the beam is directly related to alignment of the moving mirror 14 with the projected image of the adjustable fixed mirror 12. This is reasoned in that a coherent beam entering the interferometer will be changed in phase if the beam traverses unequal length paths, from one side of its width to the other. Misalignment is generally the result of non-parallel or angular relationship between the surface of another. The angular relationship will cause a difference in path length across the width of the laser beam 16, resulting in a phase change across the laser beam cross section after it has traversed the interferometer path. If the mirrors are out of alignment by as much as a fraction of the wavelength of the laser beam, a phase change will be introduced across the beam. This phase change is measurable by determining the difference of intensity

across the beam, and can be used as a sensed parameter in parallel alignment of the mirrors. Since a light wavelength is used as the measuring tool for determining mirror alignment, accuracy of alignment can be maintained to a very high degree.

The adjustable mirror 12 is mounted to the base of the interferometer (not shown) by three electrostrictive elements, such as piezoelectric transducers are responsive to a voltage signal to change in length. The piezoelectric transducers are positioned between the mirror and the interferometer base in an arrangement similar to the arrangement of the detector array 44, so that each piezo transducer is relatable to a corresponding photodetector.

Preferably, piezo transducers 101, 102, and 103 and the photodetectors of detector array 44 are positioned in a circular manner equidistant from the center of the mirror and the center of the laser beam 30, respectively, and equidistant from one another. In this relationship a piezo transducer will be proportionally responsive to an individual detector detecting the laser beam 30, due to a similar positioning within the beam path.

A signal produced by an individual detector can be applied to a corresponding piezo transducer to increase or decrease the length of the path to which the transducer is responsive. With all the piezo transducers cooperating, an equal path across the laser beam width can be continually obtained, indicating alignment of the moving mirror.

The piezo transducers as described are available from the Vernitron Co., as Part No. 16-8031-5H.

The tilt servo control circuit which processes the signals produced by the detector array 44 to generate correction signals 104, 105 and 106 is presented in Figure 3. The tilt servo control 50 receives three electrical signals 80, 81 and 82 produced by the photodetectors of detector array 44 in response to intensity fluctuations exhibited by the heterodyned laser beam 30. Each of the electrical signal 80, 81 and 82 indicates the relative phase of the respective portion of the heterodyne beam 30 leaving the laser indicating the relative differences in optical path length for light of those portions of the beam. The difference in path length measures the relative alignment of the related portions of the interferometer.

The signals 80 through 82 are applied to a relative phase error detection circuit 90. The relative phase error circuit 90 compares the phase relationship of the detector signals 80 and 81 with detector signal 82 to generate two phase error signals 92 nd 94. The phase error signals 92 and 94 are applied to integrators 96 and 98, respectively. The integrators 96 and 98 integrate signals 92 and 94, respectively, by means commonly known by those skilled in the art to obtain a summation over time of the phase error between the detector signals indicated by the relative phase error detection circuit 90. Integration of error signals 92 and 94 enables much more accurate control of small phase errors detected between the signals 80-82.

The output of integrators 96 and 98 are applied to a mirror anti-translation circuit 100. The mirror anti-translation circuit 100 receives the two integrated relative phase error signals 92 and 94 and produces a signal for each of the piezoelectric elements 101, 102 and 103. The produced signals will be used to control

the elongation of each of the piezoelements 101-103 to tilt the mirror by an amount specified about the center of its reflecting surface to control optical path length through the interferometer. Each of the piezoelectric elements 101, 102 and 103 is controlled and varied in length. If only two of the three piezoelectric elements were controlled using the third as a pivot for the mirror, there would be a translation movement of the ^' mirror when tilt adjustment was performed. The translational movement would vary the length of the optical path to the interferometer thus introducing error through optical path change. Advantageous use of the three piezoelectric elements 101, 102 and 103 to obtain tilt adjustment of the mirror about the center of its reflecting surface substantially removes any translational effect through mirror adjustment and thus removes error caused through changes in optical path length.

The control signals for the piezoelectric elements 101, 102 and 103 are applied to high voltage drivers to step up the low voltage signals produced by the mirror anti-translation circuit 100. High voltage drivers 104, 105 and 106 are electrical elements commonly known to those skilled in the art and simply function to increase the voltage level of the signals obtained respectively from the mirror anti-translation circuit 100.

A tilt status signal 115 may be generated by a level comparator 114 through comparison of the piezoelectric drive signals 110, 111 and 112 to known voltage signals indicative of each end of the end of the piezoelectric element range. For instance, a low voltage signal may be used to indicate a piezoelectric element has expanded to approximately the end of its range and a

high voltage level signal may be utilized to determine that the piezoelectric element has contracted to approximately the end of its range. The tilt status signal generated through comparison of drive signals 110 through 112 can thus be used to indicate whether or not the tilt servo control is able to perform mirror alignment within the system's physical range of adjustment. For instance, inoperative status indicated by tilt status signal 115 would indicate that alignment of the mirrors within the interferometer is not being maintained or that a gross misalignment has occurred which the servo electronics is unable to correct. The tilt status signal and the electronics utilized to obtain it are not considered part of the present invention.

A tilt enable/disable signal 116 may be provided by a main computer controlling the spectrophotometer. The enable/disable signal 116 allows the computer to enable and to disenable operation of the tilt servo control 50. When the enable/disable signal 116 is applied to the tilt servo 50, the servo 50 generates a fixed voltage signal which is selected to drive the piezoelectric elements 101, 102 and 103 to the center of range of movement. Additionally, phase comparators included in the tilt servo control 50 are forced to a known state and the integrators 96 and 98 are driven to a nominal "zero" value. The enable/disable signal 116 may be provided to determine disenablement of the tilt servo control 50 when gross mechanical alignment of the interferometer is to be performed. It may also be used to determine disablement of the interferometer during reset and test procedures to determine proper functioning of the tilt servo.

The relative phase error detection circuit 90 may be understood and described with reference to Figure

4. The error detection circuit 90 functions to produce phase error signals 92 and 94 which are indicative of the phase difference between two of the detector signals 80 and 81 relative to the third detector signal 82. It should be understood that the detector signals can be rearranged and compared relative to one another in any order and a comparison of the signals 80 and 81 with 82 may be revised to provide comparison of signals 81 and 82 with signal 80 or signals 80 and 82 with signal 81, etc.

The detector signals 80, 81 and 82 are applied to zero crossing detectors 120, 121 and 122, respectively. The zero crossing detectors are used to convert the analog or sine wave signals characteristic of the detector signals 83, 82 to a square wave signal. The zero crossing detectors 120 and 121 used to convert detector signals 80 and 81 are inverted in relation to zero crossing detector 122 used to convert detector signal 82 which provides output signals 180° out of phase when the detector signals 80 and 81 are found to be in phase with detector signal 82. This inverted relationship is advantageously used to improve linearity and functioning of the described circuit.

The outputs of zero crossing detectors 120 and

121 are applied to a state forcing network 123. State forcing network 123 is used to establish the initial state of operation of phase comparators 128 and 130. The phase comparators 128 and 130 have a range of -360 to +360° of phase difference between their two inputs, respectively. This permits ambiguity and phase comparison, in that two signals having a difference in phase of 10° may be considered to have a difference in phase of +10° or -350°. This ambiguity is unacceptable for control of mirror alignment since the direction of correction for adjustment would be undeterminable. State

forcing network 123 forces each of the phase comparators 128 and 130 to a known condition. This is done by forcing one input of each phase comparator 128 and 130 to a high voltage condition, while the other input continues to vary. This forces each of the phase comparators 120 and 130 to a known initial state. Initial state control can be accomplished when the tilt servo control is disabled through application of the tilt enable/disable signal 116.

The state forcing network 123 utilized with the present invention comprises an analog switch 124 and an analog switch 126. Analog switch 124 receives the output of zero crossing detector 120. Analog switch 126 receives the output of zero crossing detector 121. Both of the analog switches 124 and 126 receive the tilt enable/disable signal 116. When the tilt enable/disable signal 116 instructs the tilt servo control to disable, the signal 116 also instructs the analog switches 124 and 126 to apply a selected voltage to a first input of each of the phase comparators 128 and 130, respectively. The second input of each of the phase comparators 128 and 130, respectively, receive the output of zero crossing detector 122 and is permitted to vary. This drives each of the phase comparators 128 and 130 to a known initial state from which the phase comparators and the direction of phase correction may be determined.

As discussed, the relative phase error detection circuit 90 comprises two phase detectors 128 and 130. Phase detector 128 receives a first input corresponding to the output of zero crossing detector 120 and thus detector signal 80 and receives a second input corresponding to the output of zero crossing detector 122 and thus detector 82. Phase comparator 130 receives a first input corresponding to the output of zero crossing

detector 121 and thus detector 81, and a second input corresponding to zero crossing detector 122 and thus detector 82. By this design the phase detectors 128 and 130 compare the phase of detector signal 80 with detector signal 82 and detector signal 81 with detector signal 82, respectively. The phase detectors 128 and 130 are common elements widely known to those skilled in the art. The phase detectors described are available from the Motorola Corporation under Part No. MC14046.

Phase comparators 128 and 130 produce output pulses having a pulse width which is indicative of the phase difference between the input signals received. If the input signals are 180° out of phase, the pulse output will have a 50% duty cycle. The pulse would increase if the phase difference is greater than 180° and the pulse width will decrease if the phase difference is less than 180°.

The outputs of phase comparators 128 and 130 are applied to a relative phase adjustment network 132. The relative phase adjustment network 132 permits introduction of a phase correction adjustment to the detected and processed signal received from each of the detectors 80-82 to correct known optical distortions in the interferometer optic system. For instance, dispersion effects cause the laser beam and the infrared beam passing through the interferometer to be directed in slightly different directions from the beam splitter. This inexact directional integrity causes a loss in infrared beam energy due to slight misalignments of the optic system for maximum efficiency for the IR beam, when the system is aligned for use as a laser beam having a characteristic differing light wavelength. Dispersion effects are well known and can be understood through a study of many well known treatises, such as "Fundamental

Optics" by Francis A. Jenkins and Harvey E. White published by the McGraw-Hill Company.

To make up for dispersion effects and other optical inefficiencies it is advantageous to introduce a correction value to the output signal used for controlled mirror alignment. Introduction of this corrected value is accomplished through the relative phase ajust ent network 132. The relative phase adjustment network 132 comprises a pair of potentiometers each receiving an output from phase detector 128 and 130, respectively. The potentiometers form voltage dividers between the pulse signals received from the phase comparators 128 and 130 and ground. Thus, the potentiometers provide a means for selecting a ratio of the control signals produced by phase comparators 128 and 130 in comparison to system ground and relative to one another, to drive the piezoelectric element with a selected correction factor. The relative phase adjustment network 132 provides a means for introducing a selected phase offset into the tilt servo control's ability to obtain perfect parallel alignment of the image of mirror 12 with a movable mirror 14. Simply, phase adjustment network 132 provides a means for providing alignment of the fixed mirror 12 to a selected non-parallel position at which the infrared beam energy is maximized.

The mirror anti-translation circuit 100 and the theory upon which its design is based can be learned from Figures 5-8. Referring first to Figure 5, from the geometry of the piezoelectric element mounting, 101, 102 and 103, it can be shown that movement of the center of the mirror (in and out of the page as shown) is related to movement of each of the piezoelectric elements 101, 102 and 103 according to the following equation:

X _c = 1/3 ( X _χ + X ₂ + X ₃ ) ( 1 )

where X _c equals the position of the center of the mirror X^ equals movement of piezoelectric element 101 X ₂ equals movement of piezoelectric element 102

X3 equals movement of piezoelectric element 103

Since the teaching of this invention and the desired function of mirror alignment is that adjustment should be made to mirror 12 without movement of the center X _Q , X must be held constant in equation (1) as the other variable X _lr X ₂ and X3 are changed. Assuming that the length-to-signal relationship of each of the piezoelectric elements 101, 102 and 103 is a linear function and that each of the piezoelectric elements is identical, in equation (1) can be converted into electrical terms as follows:

K I == 1/3(V _X + V ₂ + V ₃ ) (2)

where K^ is a constant

V- _j _ equals the voltage applied to the piezoelectric element 101

V equals the voltage applied to piezoelectric element 102

V3 equals voltage applied to piezoelectric element 103.

Application of equation (2) shows that if the voltages ^v ι~ ^v 3 ^are equal that' the constant K^ will be equal to the value at which these voltages are equal, i.e., K will equal voltage 1, voltage 2, voltage 3.

Assuming that the final high voltage drivers 104-106 have a transfer function of the form V _Q _ _{κ v} . ₊

K-i, the desired inputs of the mirror anti-translation circuit 100 will be as follows:

i' = (V _χ - K _χ )/K ₂ (3)

V ₂ ^« = (V ₂ - K _χ )/K ₂ (4)

V ₃ ^« = (V ₃ - K _χ )/K ₂ (5)

where V^' is the correction voltage for the first piezo element 101

V ₂ * is the correction voltage for the second piezo elment 102 V3* is the correction voltage for the third piezo element 103.

Applying these equations (3-5) to equation (2), we have:

0 = 1/3(Vi ¹ + V ₂ ' + V ₃ ') (6)

If we have three signals _A , V _B , V _c which indicate the desired rotation to be performed by each of the piezoelectric elements 101, 102 and 103, respectively, about their respective axis of rotation through the center X _Q of the mirror, equations (3), (4) and (5) may be transformed as follows:

V _χ ^« = K _n ' (V _A - 1/2(V _B + V _c )) (7)

V ₂ ^» = K _n ^« (V _B - 1/2 (V _A + V _c )) (8)

V ₃ ^» = K ₃ ^« (V _C = 1/2(V _A + V _C )) (9)

K ₂ and K _n are a constant selectable by the designer to determine the amount of signal gain desired in the control system.

Since the tilt servo control utilizes one of the piezo elements as a reference, only two tilt correction signals need to be derived, the third being defined as a zero factor, i.e., Vς equals zero. Applying this statement to equations (7)-(9) we find

Vl' = K _n (V _A - 1/2 V _B ) ^" (10)

V ₂ ^« = K _n (V _B - 1/2 V _A ) (11)

V3' = (-K _n 2 ⁾ ( _A + V _B ) (12 ⁾

These equations (10)-(12) represent the functions that the anti-translation circuit 100 must function to resolve.

For instance, the general characteristic of the mirror anti-translation circuit 100 can be visualized in Figure 6. Each of the networks 140, 141, and 142 receive the outputs of integrators 96 and 98 as each of their first and second inputs, respectively. Network 140 produces an output signal having a characteristic of equation (10). Network 141 produces an output having the characteristic of equation (11). Network 142 produces an output having a characteristic of equation (12). The respective outputs of networks 140-142 are applied to high voltage drivers 104-106, respectively. The design of circuits to perform the function of generating a signal characterizing each of the equations (10)-(12) is common to those skilled in the art and need not be described, herein in detail to understand the presented invention. The design of each of the individual networks 140, 141 and 142 are not individually considered to be inventive.

An alternative circuit which may be utilized to perform the function of the mirror together with anti- translation circuit 100 is shown in Figure 7. The circuit comprises an inverting summing circuit 146 which receives the outputs of integrators 96 and 98, respectively, and provides an output signal characterized by the equation

V _χ = -1/3(V _A + V _B )

The output of the inverting and summing circuit 146 is applied to summing circuits 148 and 158 and to a dividing gain circuit 152. Summing circuit 148 receives the output of integrator 96 in addition to the output of inverting and summing circuit 146 to generate a signal characterized by the equation

V ₁ ^I » 1/2 V _A + 1/2 V _χ .

Summing circuit 150 receives the output of integrator 98 in addition to the output of inverting and summing circuit 146 to generate a signal characterized by the equation

V ₂ ^« = 1/2 V _B + 1/2 V _χ .

The divide and gain circuit 152 receives the output of inverting and summing circuit 146 and provides a simple divisional modification to this signal, dividing its value by one-half such that it produces an output signal characterized by the equation

V ₃ ' = 1/2 V _x = 1/6(V _A + V _B ).

Each of the constants in the above-described circuit may be changed depending upon each characteristic

desired in the output signals of circuit elements 148, 150 and 152 providing only that they meet the requirements of equations (10)-(12) as determined by electrical circuit design considerations known by those skilled in the art.

Figure 8 discloses the circuit elements can be combined to provide the functions described for the circuit of Figure 7. Each of the networks 146, 148, 150 and 152 is comprised of circuit elements which are generally known in the art. Their selection and network design comprising their interconnection is clearly presented in the drawing without need of description.

Figure 9 depicts an alternative selection for receiving inputs for the inverting and summing circuit 146. The inputs are received from the outputs of summing circuits 148 and 150, respectively.