 |
Description  |
|
|
BACKGROUND OF THE INVENTION
This invention relates to a high precision laser interferometer. Laser
interferometers are used to measure the movement of precision positioning
tables, for example.
Generally speaking, an interferometer is an instrument that measures
distances and very small changes in distances by means of the interference
of two beams of light. In a laser interferometer, the light source is a
laser tube providing a strong coherent monochromatic light beam. In the
well known Michelson interferometer, a source of monochromatic light is
split (by a beam splitter) into a reference beam and a measurement beam.
The reference beam is directed along a fixed path (reference arm) to a
light sensor (photodetector). The measurement beam is directed to follow a
path that varies with the distance being measured (measurement arm). The
two beams are recombined at the beam splitter so as to interfere. The
resulting interference is monitored at the sensor. When there is a phase
difference between the two recombining beams due to the difference in
distances traveled by the beams in the reference and measurement arms, the
beams interfere with each other reducing the intensity of the measured
signal. The output of the sensor will reach a positive or negative peak
each time the distance traveled by the measurement beam increases or
decreases by one half wavelength of the monochromatic light being used.
The polarity changes are counted to provide a measure of the change in
distance.
In the well known double frequency Doppler shift interferometer, the
reference beam and measurement beam have frequencies that are slightly
different. The changing distance traveled by the measurement beam causes
the frequency of the measurement beam at the sensor to change (the Doppler
shift). The difference between the source frequency and the Doppler
shifted frequency is integrated over time to provide the change in
distance itself.
The need for a low cost laser interferometer which is stable over a long
period of time in industrial environments is known. Stability in a laser
interferometer means that over a long period of time the measurement of a
standard distance does not deviate more than an allowed fraction. The
search for yet greater and greater stability has led to instruments of
greater and greater complexity.
The lack of stability in a Michelson interferometer may be related to the
lack of pointing stability in the laser beam and in the lack of mechanical
stability of the remainder of the system due to temperature changes in the
structure holding the measurement and reference arms. Fortunately, the
stability of the frequency and power outputs of the laser itself is not a
limiting factor. Extremely stable single-frequency lasers are now
available at a reasonable cost. See, for example, U.S. Pat. No. 4,819,246.
The stability of laser wavelengths are measured in parts per million per
year, say 0.02 to 0.1 ppm/year.
Mechanical instability in the Michelson interferometer structure has the
effect of changing the effective lengths of the beam paths as well as
changing the degree of overlap and the angles of convergence of the
reference and measurement beams at the sensor. Imperfect overlap of the
reference and measurement beams at the sensor results in only one or the
other of the beams striking a portion of the sensor. Without interference,
this portion of the sensor simply produces a DC component in the sensor
output. Mechanical instability can change the degree of overlap and
therefore the DC component of the sensor output. Since the AC (sinusoidal)
component of the interference signal is squared in a single level
detector, any drift in the DC component of the signal will change the duty
cycle of the squared signal and a sufficient instability will result in
loss of the AC component altogether.
As a practical matter, the two beams cannot be made to arrive at the sensor
perfectly aligned. Hence, a fringe pattern strikes the sensor where the
two beams overlap. The fringe pattern may be considered the result of two
collimated beams approaching the sensors at different angles or two
divergent beams approaching the sensor along spaced axes or both. If the
spacing of the fringe pattern relative to the size of the sensor is
insufficient, more than one fringe may strike the sensor at all times.
Since the output of the sensor is related to the sum of the intensity of
the light over the entire surface of the sensor, the output signal is
related to the total number of fringes and factions thereof that strike
the sensor. The AC component of the output signal of the sensor can be
lost altogether when more than a single fringe strikes the sensor.
Decreasing the angle of the convergence of the two beams and/or their
spacing, if the beams are divergent, the fringes become more widely
spaced. Unfortunately, as will be explained in detail, this actually
results in increased instability.
Elaborated systems have been devised to overcome the above-noted problems.
For example, the double frequency Doppler shift interferometers measure
frequency shifts between a measurement and reference beam thus making them
insensitive to fringe contrast and light source intensity. See, for
example, U.S. Pat. Nos. 3,788,746 and 3,714,607. Unfortunately, Doppler
shift laser interferometers can lose count at speeds easily attained by
modern x-y position stages.
Another system based upon the Michelson interferometer uses a single
frequency and a complex detection system detecting three different
interference signals, one in quadrature with the other two. The three
signals are combined in a way to cancel the effects of thermal drift on
the DC level of the interference signal. See U.S. Pat. No. 4,360,271.
With an interferometer, two signals in quadrature (90.degree. out-of-phase)
must be detected to enable discrimination of the direction of change of
the measured distance. Typically, the reference and measurement beams are
both split and recombined as two different interference patterns at
sensors spaced one quarter of a fringe so that the output signals at each
sensor are in quadrature.
SUMMARY OF THE INVENTION
It is an object of this invention to provide a simple single-frequency
laser interferometer having superior stability.
It is an object, according to this invention, to provide a practical
single-frequency laser interferometer in which fringes are counted as the
interference pattern crosses a detector.
It is yet a further object, according to this invention, to provide a
method and apparatus for deriving quadrature signals from successive
fringes in a single interference patter crossing two closely spaced
sensors.
Briefly, according to this invention, a stable laser interferometer
comprises a single-frequency Michelson type interferometer. The light
source comprises a power (frequency) stabilized single-frequency laser.
Preferably, the laser is a helium-neon laser with a wavelength of 632.8 nM
and a beam diameter less than 1 mm. As a practical matter, a beam diameter
of about 0.5 mm is used and in those embodiments where the beam is
collimated, the beam is expanded to have a diameter of about 2.5 mm.
Adjustable means are provided for directing the measurement and reference
beams to converge on a measurement plane at a large effective convergent
angle. By large effective convergent angle is meant an angle large enough
or a spacing large enough in the case of divergent beams, or both to
provide stability given the particular mechanical system and laser
wavelength. The less stable the mechanical system, the larger the
effective convergent angle must be. The larger the wavelength, the larger
the effective convergent angle must be. For a precision mechanical system
constructed with aluminum parts and a helium-neon laser, the effective
convergent angle should be at least 0.1 mRad and preferably 0.2 mRad. The
effective convergent angle cannot be made arbitrarily large, however,
since as the angle is increased, the signal-to-noise ratio reduces and the
fringes may become so close together that detectors small enough to detect
successive fringes may not be available.
Two or four closely spaced sensors mounted in the same plane are positioned
at or about the measurement (interference) plane. If two sensors are used,
they are arranged side-by-side. If four sensors are needed, they are
arranged in the four quadrants defined by two perpendicular axes.
According to one embodiment, the sensors are mounted to be translated in
two directions parallel to the measurement plane; and adjustable means are
provided to rotate both sensors relative to an axis parallel to the
measurement plane (detector tilt). Preferably, the sensors have areas of
less that about 1.3 mm square mounted in substantially the same plane less
than about 2 mm or center. More preferably, two side-by-side sensors are
spaced approximately 1.5 mm on center and are about 1 mm square. Most
preferably, quadrant-type sensors have four sensitive areas
1.45.times.1.45 mm and are positioned with edges 0.1 mm apart. It is this
separation distance (0.1 mm specifically) which determines the actual
interference angle to be used.
Preferably, the reference and measurement beams are directed by a beam
splitter at right angles to each other. The reference beam is redirected
to the beam splitter by a retro prism mounted a fixed distance relative to
the beam splitter. The measurement beam is redirected to the beam splitter
by a retro prism mounted on the table or other device, the linear movement
of which is being measured.
According to one embodiment in which the laser beam is not collimated, at
least one of the retro prisms, preferably the retro prism which redirects
the reference beam, is mounted for rotation about an axis perpendicular to
the incoming beam and preferably is mounted for two degrees of translation
along the axis perpendicular to the incoming beam. Retro prisms return the
outgoing beam parallel to the incoming beam. Rotation of the retro prism
will change the spacing between the incoming and outgoing beam somewhat.
This feature may be used to adjust the effective convergent angle between
the reference and measurement beams at the sensor. By a combination of
adjustments of the effective convergent angle and the angle between the
measurement plane and the sensor plane (detector tilt), stability can be
achieved and the output signals of the two sensors can be brought into
quadrature.
In the case of a laser beam which is not collimated, the change in the
length of the measurement arm beam results in a slight change in the
effective convergent angle. This can result in a slight deviation from the
quadrature condition. The effect of this variation can be further
minimized by careful selection of certain parameters. Specifically, the
beam diameter should be chosen so that the Rayleigh divergence threshold
for the measurement beam arm falls approximately midway between the outer
limits of travel of the measurement retro prism. The length of the
reference arm should be adjusted to be substantially equal to the length
of the Rayleigh divergence threshold.
According to another embodiment, both retro prisms are fixed. The laser
beam is collimated and expanded. The beam splitter is arranged for
rotation and the sensor is arranged so that the sensor plane is
substantially perpendicular to the path of the interfering beams. The
convergent angle is adjusted by rotating the beam splitter about its axis
perpendicular to the beam paths. The embodiment has the advantage of
providing stability up to 40 inches of travel and requiring only a single
adjustment to bring about the desired convergent angle.
BRIEF DESCRIPTION OF THE DRAWINGS
Further features and other objects and advantages will become apparent from
the following detailed description made with reference to the drawings in
which:
FIG. 1 is a schematic diagram of a single-frequency interferometer without
a collimator which is useful for explaining one embodiment of this
invention;
FIG. 2a illustrates the development of an interference pattern with the
interferometer according to FIG. 1 on the basis of convergent plane waves;
FIG. 2b illustrates the development of an interference pattern with the
interferometer according to FIG. 1 on the basis of offset diverging beams;
FIG. 3 illustrates the fringe spatial function with change of convergent
angle for the interferometer of FIG. 1;
FIG. 4 illustrates the fringe spatial sensitivity to convergent angle of
the interferometer of FIG. 1;
FIGS. 5a and 5b illustrate a light sensor suitable for the practice of this
invention;
FIG. 6 illustrates the partial overlap of the converged measurement and
reference beams upon the light sensor and the formation of a single fringe
pattern in the overlap;
FIG. 7 schematically illustrates successive fringe detection with the light
sensor of FIGS. 5a and 5b according to this invention and the adjustment
of angle b (detector tilt) to achieve signals in quadrature;
FIG. 8 schematically illustrates a circuit for detecting the output of the
light sensors and applying same to a conventional up/down counter for a
PID control loop;
FIG. 9a is a section view of the socket for receiving the reference retro
prism;
FIG. 9b is a section view of a lock nut for securing the retro prism in the
socket of FIG. 9a;
FIG. 9c is a front view of a first frame to which the socket is secured by
springs;
FIG. 9d is a top view of the first frame;
FIG. 10 is a back view of the socket and first frame mounted within the
second frame;
FIG. 11 is a front view of a frame for mounting the dual sensor for
adjustment;
FIG. 12 is a broken away side view of the self-contained single-frequency
interferometer unit with a collimator exclusive of the measurement retro
reflector according to the teaching of this invention;
FIG. 13 is an end view of the self-contained unit shown in FIG. 12;
FIG. 14 is a section view through a portion of the self-contained unit
along the lines XIV--XIV in FIG. 13;
FIG. 15a and 15b comprise diagrams illustrating the quad light sensor
circuit for the interferometer according to FIG. 12; and
FIGS. 16a and 16b illustrate a quad light sensor suitable for the practice
of this invention.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
Referring to FIG. 1, there is shown schematically a single-frequency
(Michelson) laser interferometer without a collimator. For applicant's
interferometer, the single-frequency output of the laser 10 is polarized,
frequency and power stabilized. It is not collimated for this embodiment
of my invention in which divergence of the laser beam is relied upon to
establish the effective convergent angle. The beam is directed to a beam
splitter 11 where one half of the beam is reflected to form a reference
beam 12 and the other half is transmitted to form a measurement beam 13.
The reference beam is directed to a fixed retro prism 14 and the
measurement beam is directed to a moving retro prism 15 at one end of the
distance to be measured. In practice, the retro prisms 14 and 15 should be
as perfect as possible. Theoretically, a retro prism has the ability to
redirect the outgoing beam parallel to the incoming beam even when the
retro prism is rotated due to wobble of the base, for example, the
positioning table to which it is mounted. With even the most precision
mechanical system, a certain amount of wobble is inevitable.
The interference projections of the reference and measurement beams overlap
at the detector 18 located at the measurement plane. In this embodiment of
the invention not requiring a collimator, the divergence of the
measurement beam results in the enlargement ("blooming") of the
measurement beam projection upon the measurement plane as the distance
being measured increases. The degree of overlap between the two beams (and
the projections thereof) at the detector is somewhat varied. However, it
has been found that with up to 12 inches of travel of the measurement
retro prism, the blooming of the measurement beam does not adversely
effect the functioning of the interferometer. As will be explained
hereafter, two sensors upon the detector are positioned to remain within
the overlap of the projections of the two beams.
The beams reflected from the retro prisms are recombined at the beam
splitter. They form two sets of interfering beams (one transmitted set 16
and one reflected set 17) having identical information. According to this
invention, a pair of sensors form a detector 18 and are placed in the path
of either set of the interfering beams, usually the reflected beam set
since more space is available for mounting sensors at this location.
Since the measurement and reference beams are both diverging and are offset
from each other or are not exactly parallel after combination, a fringe
pattern is produced on any plane that is substantially perpendicular to
the interfering beams. The detector 18 is mounted in a plane that is
substantially perpendicular to the interfering beams selected for
detection.
It is a feature of retro prisms, which have three orthogonal reflecting
surfaces, that rotation or translation of the prism relative to the
incoming beam will change the spacing of the incoming and outgoing beams.
This feature can be used to space the reference and measurement beams from
each other. With this adjustment, the fringe pattern itself can be
adjusted.
Preferably, according to this invention, the fringe pattern is adjusted or
rotated so that the fringes of the pattern are more or less perpendicular
to a plane defined by the reference and measurement beam paths. The reason
for this orientation of the fringe pattern relates to the typical
application of the interferometer according to this invention in which it
is used to track a positioning table which moves back and forth in the
direction of the measurement beam as it emerges from the beam splitter,
which table has a surface that is generally parallel to the plane defined
by the reference and measurement beam paths (herein the "beam plane").
Typically, positioning tables are most stable against rotation about an
axis perpendicular to the surface of the table and least stable against
rotation about an axis lying within or parallel to that surface. The
fringe pattern orientation, as preferably described, is least effected by
the most likely rotation (an undesirable rotation to be sure).
The correct orientation of the fringe pattern is achieved as follows: the
polarization of the laser beam is rotated so that it is in a plane
parallel to the beam plane and the surface of the positioning table. The
reference retro prism is then translated so that the projections of the
reference and measurement beams at the measurement plane form an overlap
that is approximately an ellipse (somewhat boat shaped) with its long axis
parallel to said plane. With this adjustment, the fringes will be more or
less perpendicular to the long axis of the overlap and will move parallel
to that axis. Of course, the sensor pair is mounted so that a fringe
successively crosses one and then the other sensor.
The development of the interference pattern will now be explained with
reference to FIG. 2a for two nondiverging waves intersecting at an angle
.phi.. Dashed line A is parallel to the reference beam and is known as the
interference axis. Dashed line B is parallel to the measurement beam and
the motion of the stage carrying the measurement retro prism. The
convergent (interference) angle is labeled .phi.. The crests of two plane
waves with wavelengths .lambda. intersecting at the sensor plane SP result
in a fringe pattern. The intensity of the fringe pattern is sinusoidal
with a period .DELTA.x as shown in the top portion of FIG. 2a.
FIG. 2b illustrates the interference of two offset diverging beams wherein
the wave fronts are spherical. It should be apparent that an effective
convergent angle .phi. results in interference much the same as explained
with reference to FIG. 2a.
The intensity of the resulting fringe pattern caused by the interference of
two plane waves can be described by the relationship:
I(x,.DELTA.L)=I[1+cos(2.pi..phi.X/.lambda.+4 .pi..DELTA.L/.lambda.)]
(equation 1)
Note that the argument of the intensity function contains both a spatial
description of the fringe (function of x) and a description of how the
period shifts (propagates) as the path difference in the axes of the
interferometer is changed (function of .DELTA.L). Setting the spatial
component of the intensity function equal to 2.pi. yields the fringe
period as a function of .phi.:
.DELTA.x=.lambda./.phi. (equation 2)
The term .phi. is the interference angle between the plane waves as they
interfere (see FIG. 2a). This angle defines the fringe pattern period
.DELTA.x. The smaller the angle, the greater the distance between fringe
maxima or minima. When .phi.=0, complete addition/cancellation occurs.
This is the effect often strived for but never achieved in
single-frequency interferometers not according to this invention.
The propagation component defines the resolution of the interference
pattern: that is, the function between path difference (.DELTA.L) and how
the fringes shift or propagate with respect to a stationary reference next
to the fringe pattern. Setting this argument component equal to 2.pi.
yields the resolution of the interferometer:
.DELTA.L=.lambda./2 (equation 3)
Thus, for each half wavelength of arm movement, a complete fringe cycle
(maxima intensity to maxima intensity or minima intensity to minima
intensity) will occur. For an interference pattern defined by the
condition of .phi.=0, the maxima and minima is simply a transition from
complete dark to an intensity equal to the sum of the beams no matter
which direction (increasing or decreasing) the path length is changed. An
interesting condition occurs as the interference angle .phi. is adjusted
from the zero value. The effect is shown in FIG. 2a. A fringe pattern is
developed in which the fringe maxima and minima are spaced as defined by
equation 2. However, the direction of movement can be noted by the
direction in which the fringe propagates. As the phase fronts of one beam
(say the measurement beam) move due to the movement of one mirror, their
intersection with the phase fronts of the reference beam (stationary
mirror), which defines the interference pattern, will propagate in a
particular direction. By counting the fringes and noting their direction
of movement, distance and direction can be determined. Therefore, all the
information required for position interferometry is contained in a single
interference pattern.
An unwanted fringe movement can occur if the interference angle .phi. is
also made to change. Note that the spatial propagation will cause the
fringe pattern to expand from the reference point determined by the
intersection of the beam axes. Thus, developing quadrature signals from
fringes spaced by multiples of the fringe period .DELTA.x, allows
significant sensitivity to errors caused by the drift of .phi.. It is a
feature, according to this invention, to develop the quadrature signals
from a single fringe.
The cost of single-frequency interferometry (fringe counting) is
considerably less expensive than two frequency (Doppler) counterparts.
However, since the single-frequency instruments rely on direct
cancellation at the optical frequency, they suffer from susceptibility to
fringe pattern drift. This drift susceptibility usually effects the
interferometers ability to accurately detect direction more so than its
ability to count fringes. As mentioned before, quadrature signals (signals
that are 90.degree. out-of-phase) are required to differentiate direction.
Therefore, the stability of the fringe pattern is necessary for
application to position feedback.
In order to understand this sensitivity, note that a 1 mm fringe period
(.DELTA.x) translates from equation 2 to 0.6 mRad of beam intersection.
Typical warm-up drift of most helium-neon lasers are on the order of 0.2
mRad and possess pointing stabilities of 0.03 mRad/.degree. C. Thus, if
one is attempting to generate quadrature signals from a pinhole detection
method, the phase of these signals could change dramatically over changes
in temperature. Other methods of quadrature detection such as the use of
lossy beam splitters would also be susceptible to beam misalignment.
The most serious problem related to application of single-frequency,
fringe-counting techniques is fringe drift due to beam alignment changes.
These changes can be the result of temperature changes, which result in
misalignments of the optics or angular drift of the laser beam, as well as
mechanical misalignment over the measurement range caused by cosine or
Abbe' errors. In quadrature detection methods, such errors can effect the
ability to detect direction as well as distance moved.
As will be demonstrated by the following analysis, the DC components of any
fringe detection method are theoretically constant and dependent only on
the intensity of the fringe pattern. The intensity of an interference
pattern will remain relatively constant with the application of any
single-frequency stabilized source. The component of the fringe which can
be extremely unstable, depending on the operating conditions, is the
spatial component (.DELTA.x) described by equation 2 and physically
depicted by FIG. 2a.
FIG. 3 is a plot of equation 2 with respect to the interference angle
.phi.. FIG. 4 is a plot of the first derivative of equation 2 with respect
to .phi.. This plot can be considered as a "Figure of Merit" for
fringe-counting interferometers because it describes fringe stability for
a particular interference angle and operating wavelength.
Note that operation at interference angles less than 0.1 mRad yields
extreme sensitivities of spatial fringe propagation for changes in
interference angles (on the order of 7 meter/mRad). It is obvious that
operation at complete wavefront cancellation is nearly impossible. Even
operation at relatively small angles, (0.02 to 0.1 mRad) and the
consequently large distance required between quadrature detectors, results
in a system highly susceptible to drift. From FIG. 4, interference angles
around 0.2 mRad appear preferable. From FIG. 3, a 0.2 mRad interference
angle would produce a fringe separation of about 3 mm.
The problem still remains to develop a detection scheme that can develop
quadrature signals from successive fringes at an operating angle of around
0.2 mRad. Developing these signals from successive fringes has two
benefits: 1) Detection could be accomplished at the intersection of the
beam axes, and thus minimize the effect of spatial propagation due to
slight changes in interference angle .phi.; 2) Quadrature detection can be
accomplished from a single fringe for improved simplicity and stability.
A dual PIN detector manufactured by Hamamatsu was located which fulfills
the operating conditions just described. The dual detector is housed in a
single TO-39 case. Its physical configuration is shown in FIGS. 5a and 5b.
Two 1 mm.times.1 mm.times.0.35 mm spaced detectors commoned at the anodes
comprise the dual detection scheme.
FIGS. 6 and 7 depict the dual elements as they might be in relation to the
intensity profile of a fringe pattern. To understand how quadrature
signals can be derived from such a detector requires the following
analysis.
The current signals generated from detectors 1 and 2 under a fringe
pattern, such as that shown in FIG. 7, can be derived by integrating the
intensity function of the fringe over the active area of the detectors.
Current generated from each detector is then the average intensity times
the detector area times the responsivity of the detector (typically 0.4
A/W at 632.8 nM) Evaluating equation 1.2 (as a function of x only) over
the limits shown in FIG. 7 yields the detector currents.
##EQU1##
where: A=2 .pi..phi.X/.lambda..
Note that the limits of integration take into account detector tilt (b),
which will be shown later to provide a useful degree of freedom in
establishing 90.degree. phase shift between the two signals.
Evaluation of equations 4 and 5 yields:
I.sub.1 =RL.sub.d I(w.sub.d cos(b)+(sin(A(w.sub.d cos(b)+x))-sinA(x))/A)
(equation 6)
I.sub.2 =RL.sub.d I(w.sub.d cos(b)+(sin(A((2w.sub.d +s.sub.d)cos(b)+x))
-sin(A((w.sub.d +s.sub.d)cos(b)+x)))/A) (equation 7)
Analysis of equations 6 and 7 shows that the detector parameters s.sub.d is
to be most critical in controlling the phase relationship between I.sub.1
and I.sub.2. Parameters w.sub.d and L.sub.d only effect the average value
and depth of modulation of the signals. Note that equations 6 and 7 both
contain the DC terms:
RIL.sub.d w.sub.d cos(b)
which are independent of .phi.. This is the case with all pinhole detection
systems. In practice, this term has been found to stay constant within
+/-10% over a 12 inch travel. The equations above describe the dual
detection system illuminated entirely by the fringe pattern. In practice,
this is difficult to achieve with the size of the detector used unless
detection is accomplished at a considerable distance from the pattern, or
the pattern is expanded with a lens system. It has been found that
detection close to the fringe source without the use of expanding optics
produces adequate results.
The interference pattern developed across the dual detectors is
accomplished by overlapping the reference and measurement beams. (See FIG.
6.) This is required since the Gaussian beam fronts become spherical at
distances from the beam waist. This waist, or point of planar wavefront,
occurs at the collimating surface of the laser output mirror. As the beams
propagate, the beam diameters increase and the phase fronts increase in
degree of radius. The resulting interference pattern at the intersection
of the beams is therefore somewhat curvilinear. Due to the integrating
nature of the detection technique, this less than ideal pattern proves to
have little effect on detection performance.
The successive fringe detection technique therefore provides a considerably
smaller and less complex alternative to other systems, which goes to great
lengths in cancelling the DC component. All that is required in the
successive fringe technique disclosed herein is a simple biasing technique
to offset the DC components.
The phase and magnitude of the detection signals can be controlled by
adjustments to the beam interference angle .phi. and detector tilt angle
b. In practice, adjustments to both are required. It was found that
adjustments to .phi. were nicely accomplished by adjusting the tilt of the
reference mirror retro prism 14. Rotation of the sensor 18 will vary the
detector tilt angle b. Note from equations 6 and 7 that changes to the
detector tilt angle b have nearly the same effect as actual changes to the
separation parameter s.sub.d which is a part of the detector and cannot be
changed.
Once the proper signals are available from the dual detectors, they can
easily be processed for direct input to a PID loop using the circuit shown
in FIG. 8. Note that this circuit utilizes comparators directly coupled to
the photodetectors and does not require the complex bandwidth limited
amplifiers as taught by the other techniques. The circuit shown, when
properly designed, is capable of bandwidths in the 10's of megahertz. This
would translate to linear speeds greater than 3 m/sec for the
interferometer described.
With times four multiplication, the basic resolution of the interferometer
is less than 0.1 .mu.m. Further reductions in resolution can be
accomplished by an A/D conversion and subsequent processing of the SIN or
COS signals present at the inputs to the comparators.
The dual detector can also be mechanized by the close proximity of two
small diameter fiber optics. This also may provide two advantages: First,
the detection electronics could be located in a remote location free of
electrical interference and also provide for reduced package size.
Secondly, (and most importantly) the dual detectors could be mechanized in
much smaller dimensions so as to take advantage of operating the
interferometer with larger and consequently more stable interference
angles.
As already explained, it is necessary to adjust the position and
orientation of a retro prism. Any number of mounting schemes are possible.
Consider that the retro prism is a regular pyramid with the angles at the
apex of the three sides equal to 90.degree.. The sides' faces comprise
reflecting surfaces. A beam entering the prism through the base will, in
theory, be redirected back out of the base parallel to the incoming beam
irrespective of the orientation or position of the altitude of the prism.
In fact, no retro prism is perfect.
Referring now to FIGS. 9a to 9d and FIG. 10, a suitable mounting means for
the reference retro prism comprises a socket 41 into which the | | |
