Note: Descriptions are shown in the official language in which they were submitted.
CA 02413098 2002-11-27
TEMPERATURE INSENSITIVE FIBER-OPTIC
TORQUE AND STRAIN SENSOR
FIELD OF THE INVENTION
The present invention relates to torque and strain sensor, and more
particularly this invention relates to a temperature insensitive fiber-optic
torque
and strain sensor.
BACKGROUND OF THE INVENTION
Fiber-optic strain sensors have been developed using a wide variety of
approaches, including fiber Bragg grating sensors [A.D. Kersey, M.A. Davis,
H.J.
Patrick, M. LeBlanc, K.P. Koo, C.G. Askins, M.A. Putnam, and E.J. Friebele,
"Fiber Grating Sensors," J. Lightwave Technol. 15(8), 1442-1463 (1997)],
interferometric (Mach-Zehnder and Michelson) sensors [K.A. Murphy, W.V.
Miller, T.A. Tran, A.M. Vengsarkar, and R.O. Claus, "Miniaturized fiber-optic
Michelson-type interferometric sensors," Appl. Opt. 30(34), 5063 5067 (1991
)],
white-light interferometers [S.C. Kaddu, S.F. Collins, and D.J. Booth,
"Multiplexed intrinsic optical fibre Fabry-Perot temperature and strain
sensors
addressed using white-light interferometry," Meas. Sci. Technol. 10, 416-420
(1999)], intrinsic polarimetric sensors based on polarization-maintaining (PM)
fiber [W.J. Bock and W. Urbanczyk, "Temperature desensitization of a fiber-
optic
pressure sensor by simultaneous measurement of pressure and temperature,"
Appl. Opt. 37(18), 3897-3901 (1998)], extrinsic polarimetric sensors [C.S.
Sun, L.
1
CA 02413098 2002-11-27
Wang, Y. Wang, and J. Lin, "Design of a high-sensitivity photoelastic optical
fiber
pressure sensor: a differential approach," IEEE Photon. Technol. Lett. 9(7)
976-
978 (1997)], and extrinsic Fabry-Perot sensors [K.A. Murphy, M.F. Gunther,
R.O.
Claus, T.A. Tran, and M.S. Miller, "Optical fiber sensors for measurement of
strain and acoustic waves," Smart Sensing, Processing, and Instrumentation,
Proc. SPIE Vol. 1918, 110-120 (1993)]. To varying degrees, all of these sensor
types are plagued with the problem of cross-sensitivity to temperature. For
example, polarimetric sensors that employ PM fiber exhibit a thermal apparent
strain sensitivity on the order of 50 ps/°C [T. Valis, D. Hogg, R. M.
Measures,
"Thermal apparent-strain sensitivity of surface adhered, fiber-optic strain
gauges," Appl. Opt. 31(34), 7178-7179 (1992)] and for other sensor types 10
N~/°C is typical [W. Jin, W.C. Michie, G. Thursby, M. Konstantaki,
and B.
Culshaw, "Simultaneous measurement of strain and temperature: error analysis,"
Opt. Eng. 36(2), 598-608 (1997)].
United States Patent No. 5,723,794 issued to Discenzo is directed to a
photoelastic torque sensor that uses a photoelastic polymer detector in
conjunction with a photoelastic image sensor (CCD camera) and a neural
network. The photoelastic polymer sheet is bonded to the component being
monitored. The CCD camera receives the phase shifted light signal from the
photoelastic sheet and generates a electrical signals indicative of the phase
shift
produced in the beam reflected from the sensor sheet. The neural network then
calculates the torque based on these signals.
United States Patent No. 4,668,086 issued to Redner discloses a method
2
CA 02413098 2002-11-27
and device for measuring stress and strain in a thin film by passing a multi-
wavelength beam through the thin film and then splitting the transmitted
signal
into different spatially separated wavelength beams. The intensities of the
beams
at each wavelength are analysed to produce a measure of the strain in the
film.
United States Patent No. 4,123,158 issued to Reytblatt discloses a
photoelastic strain gauge comprising a photoelastic polymer sheet coated on
the
opposing planar faces with reflective coatings. This produces a waveguide-like
structure so that light is multiply reflected along the polymer sheet before
it exits
and this acts to produce an amplification of the visual patterns reflective of
the
strain produced in the photoelastic sheet from the underlying object.
United States Patent No. 5,864,393 issued to Maris discloses an optical
method for measuring strain in thin films that involves pumping the thin film
with
optical pump pulses and at different time delays applying optical probe pulses
and detecting variations in the transient response to the probe pulses arising
in
part due to the propagation of a strain pulse in the film.
United States Patent No. 5,817,945 issued to Morris et al. discloses a
method of sensing strain using a photoluminescent polymer coating. The method
is predicated on the relative changes or competition between radiative and non-
radiative decay mechanisms of excited photoluminescent probe molecules in the
coating in the presence and absence of strain in the coating. Regions of the
coating under greater strain due to strain in the underlying substrate show up
as
brighter areas in the processed images.
United States Patent Nos. 5,693,889 and 5,728,944 issued to Nadolink
3
CA 02413098 2002-11-27
disclose a method of measuring surface stress and uses a wafer of single
crystal
silicon which must be embedded in the material being monitored so the silicon
surface is even with the substrate surface. Fringe patterns in the light
reflected
off the silicon surface are indicative of the stress present at the surface.
United States Patent No. 4,939,368 issued to Brown discloses an optical
strain gauge comprising a diffraction grating applied to a surface and a light
from
a source having at least two frequencies is reflected off the surface and the
phase differences between the beams at the two wavelengths is related to the
strain in the surface.
United States Patent No. 4,912,355 issued to Noel et al. is directed to a
superlattice strain gauge using piezoelectric superlattice deposited onto the
substrate being monitored. Strain in the underlying substrate will add
internal
strain present in the superlattice which significantly changes the optical
properties among the different superlattice layers and these changes are
monitored by the light probe.
United States Patent No. 4,347,748 issued to Pierson discloses a torque
transducer for measuring torque on a rotating shaft. The device is based on
attaching optically flat mirrors to the shaft and reflecting a laser beam off
each of
the flats. The relative phase displacements of the beams is indicative of the
torque on the shaft.
United States Patent No. 5,298,964 issued to Nelson et al. discloses an
optical stress sensing system that is based on directing three separate
polarized
light beams along three different optical axes in a single photoelastic
sensing
4
CA 02413098 2002-11-27
element. The applied stresses in the three directions are determined
independently, and through the use of sum-difference techniques applied to the
output signals, the results can be made insensitive to fluctuations in light
source
intensity and to losses in the optical fibers that deliver the light.
United States Patent No. 4,777,358 issued to Nelson discloses an optical
differential strain gauge in which a light beam traverses two photoelastic
elements in series. The two are secured on opposite faces of the test specimen
so that the specimen transfers tensile strain to one and shear strain to the
other.
A fiber optic polarization rotator is inserted in the optical path between the
two
elements so that the system measures the difference between the transferred
tensile and shear strains and environmental effects common to the two elements
cancel.
United States Patent No. 4,556,791 issued to Spillman discloses a stress
sensor in which a light beam passes sequentially through a voltage controlled
wave plate and a photoelastic element that is bonded to a test specimen. The
optical powers in the two polarizations at ~45 degrees to the applied stress
axis
are detected and the resulting voltages applied to a difference amplifier. The
amplifier output is fed back to the wave plate to null out the net phase
retardation. The feedback signal is used as measure of applied stress.
A great deal of research has been devoted to developing schemes to
compensate for temperature dependence or to perform simultaneous
measurements of both strain and temperature [8, J.D. Jones, "Review of fibre
sensor techniques for temperature-strain discrimination," in 12'"
International
5
CA 02413098 2002-11-27
Conference on Optical Fiber Sensors, Vol. 16, 36-39, OSA Technical Digest
Series (Optical Society of America, Washington, D.C., 1997)]. Some very good
results have been demonstrated, however reduced cross-sensitivity to
temperature often comes at the cost of adding complexity to the strain sensing
system.
It would be very advantageous to provide a combination strain and/or
torque sensor having a reduced sensitivity to temperature.
SUMMARY OF THE INVENTION
It is an object of the present invention to provide a strain/torque sensor
that has a low temperature sensitivity.
The present invention provides a simple design for a temperature-
insensitive extrinsic polarimetric strain sensor. The sensing element is a
thin
sheet of photoelastic material that is bonded to the test object. It is
illuminated
with linearly polarized light with the polarization direction at 45 degrees
relative to
the strain-induced fast and slow axes in the photoelastic material. The sensor
measures the difference between the strains along these two orthogonal
directions. The reduced sensitivity of the sensor to temperature results from
the
fact that the illumination is perpendicular to the surface of the test object.
All
polarization components that are parallel to the surface will experience
identical
refractive index changes due to thermal effects. Consequently, a measurement
of the difference in strains along two directions in the surface plane is
expected
to be insensitive to temperature.
6
CA 02413098 2002-11-27
In one aspect of the invention there is provided a method of measuring
strain in a workpiece that is insensitive to ambient temperature fluctuations,
comprising the steps of:
illuminating a strain-sensitive material with a beam of selectively polarized
light, the beam of selectively polarized light being substantially
perpendicular to
the surface of the workpiece, the strain-sensitive material being in physical
contact with the surface of the workpiece;
measuring an intensity of at least one polarization component of the beam
of selectively polarized light making at least one pass through the strain-
sensitive
material, the at least one polarization component being tilted with respect to
strain-induced fast and slow orthogonal axes in the strain-sensitive material
so
that it has a projection along both strain-induced fast and slow orthogonal
axes;
and
calculating from said intensity a difference between strains along the
strain-induced fast and slow orthogonal axes in the strain-sensitive material,
the
difference being substantially independent of ambient temperature
fluctuations.
In another aspect of the invention there is provided a method of
measuring strain in a workpiece that is insensitive to ambient temperature
fluctuations, comprising the steps of:
illuminating a strain-sensitive material with a beam of selectively polarized
light, the beam of selectively polarized light being substantially
perpendicular to
the surface of the workpiece, the strain-sensitive material being in physical
contact with the surface of the workpiece;
7
CA 02413098 2002-11-27
measuring a first intensity of a first polarization component and a second
intensity of a second polarization component of the beam of selectively
polarized
light making at least one pass through the strain-sensitive material, the
first and
second polarization components being orthogonal to each other and tilted with
respect to strain-induced fast and slow orthogonal axes in the strain-
sensitive
material parallel to the surface of the workpiece so that each of said first
and
second polarization components has a projection along both strain-induced fast
and slow orthogonal axes; and
calculating from the first and second intensities a difference between
strains along the strain-induced fast and slow orthogonal axes in the strain-
sensitive material, the difference being substantially independent of ambient
temperature fluctuations.
The present invention also provides a temperature insensitive strain
sensor that is insensitive to ambient temperature fluctuations, comprising:
a strain-sensitive material adapted to be physically contacted to a surface
of a workpiece;
light source means and light beam polarization means for illuminating said
strain-sensitive material with a beam of selectively polarized light, said
beam of
selectively polarized light being substantially perpendicular to the surface
of the
workpiece; and
detection means for measuring an intensity of at least one of a first
polarization component and a second polarization component in the beam of
selectively polarized light making at least one pass through said strain-
sensitive
8
CA 02413098 2002-11-27
material, the at least one of a first and second polarization components being
tilted with respect to strain-induced fast and slow orthogonal axes in the
strain-
sensitive material parallel to the surface of the workpiece so that at least
one of
the at least first and second polarization components has a projection along
both
strain-induced fast and slow orthogonal axes; and
processing means connected to said detection means for calculating from
the intensity of the at least one of a first polarization component and a
second
polarization component a difference between strains along the strain-induced
fast
and slow orthogonal axes in said strain-sensitive material, the difference
between
strains being substantially independent of ambient temperature fluctuations.
BRIEF DESCRIPTION OF THE DRAWINGS
The fiber optic torque and strain sensor constructed in accordance with the
present invention will now be described, by way of example only, reference
being
had to the accompanying drawings, in which;
Figure 1 a is a schematic drawing of a first embodiment of a strain and
torque sensor constructed in accordance with the present invention;
Figure 1 b is a schematic drawing of an embodiment of a strain and torque
sensor similar to the sensor of Figure 1 a;
Figure 2a is a schematic drawing of another embodiment of a strain and
torque sensor using reflected light;
Figure 2b is a schematic drawing of another embodiment of a strain and
9
CA 02413098 2002-11-27
torque sensor using transmitted light;
Figure 3 is a schematic drawing of an embodiment of a torque and strain
sensor employing a polarization maintaining optical fiber;
Figure 4 is a schematic drawing of an embodiment of the sensor using two
optical fibers;
Figure 5a is a side view of a test rig for testing the strain and torque
sensor;
Figure 5b is a top view of the test rig of Figure 5a;
Figure 6 is a plot of measured sensor signal S as a function of applied
strain measured in accordance with the present invention;
Figure 7 is a plot of optical strain sensor reading as a function of the mass
suspended from a cantilevered bar under test with measurements being carried
out in a temperature controlled chamber at three different temperature
settings;
and
Figure 8 is a plot of optical strain sensor reading as a function of ambient
temperature for a constant value of applied strain.
DETAILED DESCRIPTION OF THE INVENTION
Broadly speaking, the present invention for measuring strain/torque uses a
transparent strain-sensitive material bonded to the mechanical test object. A
beam of selectively polarized light is directed into the strain-sensitive
material
along a direction that is substantially perpendicular or normal to the surface
of the
test object. Mechanical strain in the test object will be transferred into the
strain-
CA 02413098 2002-11-27
sensitive material, and the components of strain that are parallel to the
surtace
will influence the optical path length inside the transparent material due to
the
photoelastic effect (a strain-induced change in refractive index). If the
strain is not
identical in all directions that lie in a plane that is parallel to the
surface, then the
polarization state of the light will change as it passes through the strain-
sensitive
material.
More particularly, the method of measuring strain in a workpiece that is
insensitive to ambient temperature fluctuations comprises illuminating a
strain-
sensitive material with a beam of selectively polarized light with the beam of
selectively polarized light being substantially perpendicular to the surface
of the
workpiece, the strain-sensitive material being in physical contact with the
surface
of the workpiece. The intensity of a first polarization component and the
intensity
of a second polarization component of the beam of selectively polarized light
making at least one pass through the strain-sensitive material is measured
where
the first and second polarization components are orthogonal to each other and
are tilted with respect to the strain-induced fast and slow orthogonal axes in
the
strain-sensitive material parallel to the surface of the workpiece so that
each of
the first and second polarization components has a projection along both
strain-
induced fast and slow orthogonal axes. There is then calculated from these two
intensities a difference between strains along the strain-induced fast and
slow
orthogonal axes in the strain-sensitive material, the difference being
substantially
independent of temperature fluctuations.
The selectively polarized input light beam may be circularly, elliptically or
11
CA 02413098 2002-11-27
linearly polarized, as long as it contains both a first component of
polarization that
is parallel to the strain-induced fast axis in the strain sensitive material
and a
second component of polarization that is parallel to the strain-induced slow
axis.
The light that is detected after making at least one pass through the strain
sensitive material must also be of a polarization that contains both a first
component of polarization that is parallel to the strain-induced fast axis in
the
strain sensitive material and a second component of polarization that is
parallel to
the strain-induced slow axis. The effect of strain is to change the intensity
of the
detected light due to changes in the relative phase of said two polarization
components. When two orthogonal output polarizations are detected, their
strain-
induced intensity changes are equal and opposite, so that recording the ratio
of
the difference in these two intensities to their sum produces a signal that is
independent of the input light intensity.
Preferred embodiments of the present method devices for a temperature-
insensitive extrinsic polarimetric strain sensor will now be discussed. The
sensing
element is a strain-sensitive material, preferably a thin sheet of
photoelastic
material that is bonded to the test object. The method involves illuminating
the
strain-sensitive material with selectively polarized light which may be either
circularly, elliptically or linearly polarized (preferably a laser beam or
light emitting
diode) with the light having approximately equal polarization components along
the strain-induced fast and slow axes in the strain-sensitive material. This
would
be the case for circularly polarized light or linearly polarized light with a
polarization direction at 45 degrees relative to the strain-induced fast and
slow
12
CA 02413098 2002-11-27
axes in the strain-sensitive material. It will be understood that elliptically
polarized
light or linearly polarized light with the polarization direction at angles
deviating
from 45 degrees could be used with similar results.
The illumination is substantially perpendicular to the surface of the test
object, meaning that the illumination may be perpendicular for best results
but
those skilled in the art will appreciate that the direction of illumination
may also
deviate from perpendicular and still give efficacious strain differences. The
intensity of light having made at least one pass through the strain-sensitive
material is measured from which the difference between the strains along these
two orthogonal directions is calculated. The reduced sensitivity of the sensor
to
temperature results from the fact that the illumination is perpendicular to
the
surface of the test object. Thus, all polarization components that are
parallel to
the surface will experience nearly, if not, identical refractive index changes
due to
thermal effects. Consequently, a measurement of the difference in strains
along
two directions in the surface plane is expected to be insensitive to
temperature.
The fiber-optic sensor is extrinsic, meaning that optical fibers are used for
delivery and collection of light, but the actual sensing element employs bulk
optical components. Disclosed herein are several different embodiments of a
strain and torque sensor including one-fiber and two-fiber embodiments that
employ standard optical fibers and a third embodiment that employs a single
polarization maintaining input/output fiber.
Figure 1 illustrates a torque and strain sensor head for the single-fiber
embodiment shown generally at 10. Sensor 10 includes a multimode optical fiber
13
CA 02413098 2002-11-27
12 for emitting and collecting light. Light emerging from optical fiber 12
through a
capillary termination 14 is collimated by a graded index (GRIN) lens 16 that
terminates the fiber 12. The light is reflected at by a 90° prism 18
and is then
passed through a linear polarizer 20 that sets the polarization direction at
45
degrees relative to the direction of maximum applied strain. The light then
passes
through the photoelastic sheet material 22 that has been bonded to the surface
of
the test structure 24 under examination. The transmitted light is
retroreflected by
a metal coating on the photoelastic material's far side, and retraces its path
through the polarizer and GRIN lens and back into the fiber. The 90°
prism 18
shown in Figure 1 is not essential to the operation of sensor 10 but it allows
the
fiber 12 to be positioned parallel to the test surface, and thereby reduces
the
overall height of the sensor. The GRIN lens 16, prism 18, polarizer 20 and
photoelastic material 22 are bonded together with optically transparent epoxy
to
form a single assembly. This assembly is then bonded to the surface of the
test
structure 24.
Not wishing to have the present invention limited or bound by any
particular theory or hypothesis, the polarizer 20 is oriented so that the
field of the
lightwave, as it enters the photoelastic material, may be written as
E- E2 (X+yl (1)
where x and y specify the directions that are parallel and perpendicular to
the
known direction of the applied strain that is to be measured. The
retroreflected
wave, as it enters that polarizer, is described by
14
CA 02413098 2002-11-27
E= ~" e'~%[x+e~c~5.--mx~y]
where cpx and cpy are the total optical phase shift accumulated in the double
pass
by the x and y polarization components. The phase difference, cpy - cpX, is
given by
~y - ~~ - 4nCEt (~r - ~K )
where CE is the strain-optic coefficient of the photoelastic material, t is
its
thickness, A is the wavelength of the light, and cY - sX is the difference in
the
strains along the y and x directions. The attenuation of the lightwave as it
retraces its path through the polarizer varies sinusoidally with the phase
difference so that the intensity of the return signal is given by
I = ~ [ 1 + cos(~Y - ~~ )]. (4)
Therefore the relative intensity of the retroreflected light may be used as a
measure of the strain difference.
Equation 4 shows that in order to determine the strain difference from a
measurement of the retroreflected intensity, I, the input intensity, lo,
should be
known. to can be determined by varying the applied strain so that I sweeps
through its maximum value, where I = lo. However, without continual monitoring
of the input intensity, the sensor design shown in Figure 1 may be subject to
errors associated with drifting of the source power.
CA 02413098 2002-11-27
The two-fiber sensor 40 illustrated in Figure 2 removes this difficulty. The
polarizer 20 (Figure 1 ) is replaced by a polarizing beamsplitter 42 in sensor
40
(Figure 2). Thus the cross-polarized (or perpendicular) component of the
retroreflected lightwave can be collected by a second GRIN lens 16'//fiber 12'
combination rather than being simply absorbed as occurs in the polarizer 20
(Figure 1 ). While the intensity returned in the input/output fiber is given
by
Equation 4 (just as in the single-fiber device 20 of Figure 1 ), the intensity
returned
in the perpendicular polarization component is given by
I' =h~I-COS(~y -~x)~. (5)
From the measured values of I and I', one can generate an output signal, S,
that
is insensitive to fluctuations in the input power.
_ .
I + I ~ COS~I~i, - ~x
In some applications it may be highly desirable to have only one fiber
running between the sensing element and the control unit that contains the
light
source and detector(s). Figure 3 is a more detailed illustration of a sensor
50 that
permits simultaneous monitoring of the parallel- and perpendicular-polarized
components (I and I') using single input/output fiber 12. Sensor 50 includes
lengths of polarization maintaining (PM) fiber 52 connected by PM connectors
54,
one fiber optic 2x2 coupler 57 and one fiber-optic polarization splitter 56 in
the
16
CA 02413098 2002-11-27
control unit. The input to the final section of PM fiber 52 must be polarized
parallel
to one of the fiber's principal axes so that the output is also linearly
polarized, and
oriented at 45 degrees relative to the direction of maximum applied strain.
The
combination of the fiber-optic polarization splitter 56 (which spatially
separates the
two linear polarization components in both the input and reflected beams) and
the PM fiber performs the same function provided by the polarizing
beamsplitter
42 in the two-fiber sensor 40 (Figure 2). The power of the light source 58 is
measured using a power monitor 60. No additional polarizing element is
required
between the GRIN lens 16' and the photoelastic material 22 in sensor 50. The
two
detectors 62 and 64 are used to measure the reflected light components
according to Equations 4 and 5.
While measuring both parallel and perpendicular components of the
intensity of light is preferred to eliminate the effect of source
fluctuations, it will be
appreciated that only one component needs to be measured, and when only one
component is measured it may be either the component parallel to the selected
input polarization direction or the component perpendicular to the selected
input
polarization direction as long as the polarization component is tilted with
respect
to strain-induced fast and slow orthogonal axes in the strain-sensitive
material so
that it has a projection along both strain-induced fast and slow orthogonal
axes.
The component that is actually measured will depend on the geometry of the
apparatus, the embodiment shown in Figure 1 being configured so that the
parallel component is measured.
Regardless of whether the light is circularly, linearly or elliptically
polarized,
17
CA 02413098 2002-11-27
signal fading will occur for those values of the strain difference, cy - sX,
for which
cos(cpy - cpX) _ ~1 (i.e. for which dS/d~ = 0). To achieve maximum
sensitivity, the
sensor should be operated in a region of linear response, corresponding to the
quadrature points where cos(cpy - cpX) = 0. In the absence of any built-in
birefringence (such as might result from the epoxy curing process), signal
fading
for the designs described above, for the case of linearly polarized input
light, will
occur under conditions of zero strain difference, where ~y - ~X = 0, and the
sensor
will be incapable of measuring low values of strain. Referring to Figure 1 b,
this
problem can be corrected by inserting a birefringent wave plate 21 into the
sensor
head assembly to shift the quadrature point to match the region of strain
values
that are to be measured. The wave plate 21 should be epoxied between the
polarizing element and the photoelastic material. If it is oriented with one
of its
principal axes parallel to the direction of maximum applied strain, it will
add a
fixed phase delay to the strain-induced phase delay between the x and y
polarization components (Equation 3). If, for example, the additional phase
delay
is ~/2 (for the sum of both the forward and reverse pass though the wave
plate),
then the quadrature point will match the zero strain condition.
For designs that make use of circularly or elliptically polarized input light,
the signal fading points will not coincide with zero strain difference.
Nevertheless
it may still be advantageous to insert a wave plate 21 to shift the quadrature
point.
Any of the configurations illustrated in Figures 1 a, 1 b, 2 and 3 may be
modified so as to direct the light beam through the photoelastic element
without a
retroreflecting coating on its far side. Referring to Figure 2b, this would
require a
18
CA 02413098 2002-11-27
small hole 180 to be drilled in the test piece 24 to open up an optical path,
and
the output light would be collected on the side of the test piece 24 opposite
to the
sensing element 22. The nonreflecting equivalent of Figure 1 employs a second
polarizer and GRIN lens/fiber combination (not shown) on the output side. In
the
nonreflecting equivalent of Figure 2a, shown in Figure 2b, a polarizing
beamsplitter 42 and two output fibers 184 and 186 are needed on the output
side,
but a single input fiber 12 and polarizer 20 are sufficient on the side with
the
sensing element 22.
Figure 4 illustrates a more detailed view of a two-fiber sensor shown
generally at 70. The input/output fiber 72 and the crossed-polarized output
fiber
74 are standard 50/125 graded index multimode fibers. Both of these fibers 72
and 74 are terminated in glass capillary tubes 76, 76' respectively with a
polished
end to facilitate bonding to a 1.8 mm, 0.29 pitch GRIN lens 78 and 78' during
assembly of the fibers with the polarizing beam splitter 80 forming part of
sensor
head 81. The optical source 82 is either a high power superluminescent diode
(SLD) operating at a wavelength of ~ = 850 nm or a lower power, 850 nm light
emitting diode (LED). The input light is coupled into one arm 84 of a fused 3-
dB
fiber coupler 86, made using 50/125 multimode fiber, which is attached to the
input/output fiber 72 by means of a fusion splice or an FC fiber-optic
connector
88. Detector 90, which is connected to the other arm 92 of coupler 86, and
detector 94, which is connected to the output end of the crossed-polarized
output
fiber 74, are standard silicon photodiodes. The polarizing beamsplitter 80 is
5 mm
on a side and is coated for operation at l~ = 850 nm. The strain-sensitive
element
19
CA 02413098 2002-11-27
is a piece (several mm square) of 1 mm thick PS-1 photoelastic sheet 22
(Intertechnology Inc. part # PS-1 C Clear) that has had a reflective aluminum
coating evaporated on one side thereof.
In all the embodiments of the sensor disclosed herein the output of the
detectors may be input into a computer controller or processor for processing
the
intensity data and outputting the strain and/or torque. The output signals
from the
photodetectors are proportional to I and I' in equations 4 and 5. The computer
processing involves extracting the strain difference from these signals using
equations 3 and 6.
It will be understood that in the embodiments of the apparatus that do not
use polarization maintaining fibers, multimode optical fibers are preferred
since
they work considerably better than single mode fibers but single mode optical
fiber may still be used. In the embodiments using polarization maintaining
fibers,
single mode optical fibers are used.
Tests of the efficacy of the sensor disclosed herein were conducted. The
sensor head 81 shown in Figure 4 was first assembled by epoxying the input
GRIN lens 78 and 78' and the photoelastic sheet 22 to the polarizing
beamsplitter
96 with UV-curing optical epoxy. Then the glass capillary that terminates the
input
fiber was epoxied to the GRIN lens 78 after active alignment so as to maximize
the reflected intensity reaching detector 90. The second GRIN lens 78' and the
fiber 74 leading to detector 94 were then aligned and epoxied in the same way,
and finally the entire assembly is attached to the test piece using an instant
epoxy
that is commonly used to bond electrical strain gauges. Referring to Figures
5a
CA 02413098 2002-11-27
and 5b the test piece was an aluminum bar 110 with dimensions 305 x 19 x 9.5
mm that is held rigidly at one end with the end portion bolted to a solid
block 112
and deflection being achieved by means of a machine screw 114 at the other end
so as to induce a longitudinal strain. The optical sensor head 81 was oriented
so
that the sides of the polarizing beamsplitter cube 80 and the input
polarization
form a 45 degree angle with the length of the cantilevered bar. For comparison
purposes, an electrical strain gauge 120 was mounted on the bar in close
proximity to the optical sensor head 80.
The intensities I and I' in Equation 6 are taken to be the photocurrents
measured at detectors 90 and 94, respectively, each one being normalized to
the
maximum signal detected as the strain is swept through its full range. Figure
6
shows the resulting signal, S, as a function of the electrical strain gauge
reading
for a series of applied strain values between zero and -2400 pc. The response
is
sinusoidal with a phase that varies with applied strain at a rate of 0.00142
rad/ps.
S was measured with a relative uncertainty of t 2.5 x 10-4, as limited by
detector
noise and the quantization error in the 12-bit analog-to-digital converters
used to
record I and I'. This uncertainty translates into a strain sensitivity of ~0.6
pe in the
linear portion of the S vs. strain curve (approximately -1750 to -550 N~). The
error
bars in Figure 6 correspond to averaging 20 readings at each strain setting,
which
improves the strain sensitivity to ~0.3 N~.
The rate of change of S with strain varies linearly with the thickness of the
photoelastic sheet that is used as the sensing element. Increasing this
thickness
will enhance the sensitivity to strain, but it also reduces the period of the
21
CA 02413098 2002-11-27
sinusoidal oscillations in S. Thus there is an inherent design tradeoff
between
sensitivity and operating range. The limitations on the operating range can be
removed using various methods to avoid signal fading and remove the
ambiguities of the sinusoidal response [see R.D. Turner, D.G. Laurin, and R.M.
Measures, "Localized Dual-wavelength Fiber-optic Polarimeter for the
Measurement of Structural Strain and Orientation," Appl. Opt. 31(16), 2994-
3003
(1992)], but none of these methods was employed in the present studies.
The temperature sensitivity of the optical strain sensor was measured by
placing the entire test assembly in a thermally insulated enclosure that could
be
heated or cooled. In order to remove any complications that might result from
thermal expansion of the machine screw that was previously used to apply the
strain, the cantilevered bar was subjected to strain by hanging a series of
fixed
weights from its free end. Figure 7 shows the optical strain sensor reading as
a
function of suspended weight for three different ambient temperatures. The
maximum observed deviation corresponds to a temperature dependence of 0.6
N~/°C. The temperature sensitivity was also measured by holding the
strain value
constant while the temperature was varied over a range of approximately 30
degrees. The results of this measurement are shown in Figure 8. The thermal
apparent strain sensitivity was observed to be 0.50 ~0.02 Nc/"C. These low
levels
of temperature sensitivity were only observed after the epoxies had been
thermally cured so that the inventors believe that strain asymmetry in the
epoxies
is a large contributor to the residual temperature dependence and that it will
be
possible to reduce it further.
22
CA 02413098 2002-11-27
The inventors have developed a simple polarimetric fiber-optic strain
sensor with a very low sensitivity to temperature. It measures the difference
between the strains along two orthogonal directions in a thin sheet of
photoelastic
material therefore allowing it to be configured as a torque sensor. The
reduced
temperature sensitivity depends upon a geometry in which the illumination is
normal to the surface of the test object, so that the large thermal apparent
strains
along the two orthogonal measurement directions disappear from the difference
measurement. The present sensors have exhibited a strain sensitivity of less
than
0.3 Ns over a 1200 Ne wide operating range, and a thermal apparent strain
sensitivity of less than 0.5 N~/°C.
As stated above, a very significant result obtained with the present sensor
is the reduced sensitivity of the sensor to temperature fluctuations which
results
from the fact that the illumination is perpendicular to the surface of the
test object
so that all polarization components that are parallel to the surface will
experience
nearly identical refractive index changes due to thermal effects thereby
cancelling
temperature induced fluctuations. As noted, while illumination at exactly 90
degrees to the surface will give the best result in terms of temperature
insensitivity, good results may be obtained if illumination at angles a few
degrees
off perpendicular is used.
This sensor can be used to measure a strain that is known to be
unidirectional. It can also be used as a torque sensor, if it is oriented so
that it
senses the difference in strains along the two directions that experience
maximum torque-induced tension and compression. For example, on the surface
23
CA 02413098 2002-11-27
of a drive shaft these two directions are ~ 45 degrees with respect to the
axis.
For a disk that is subjected to a torque about an axis that is perpendicular
to the
surface, these two directions are ~ 45 degrees with respect to the radial
direction
on the disk's surface. There are many applications for torque sensing in the
robotics, automotive and aerospace industries that would benefit from the
inherent advantages that fiber-optic sensors have over their electrical or
mechanical counterparts.
The present sensor may be made very small and it requires no electronics
or electrical wiring at the point of measurement. The power supply, laser
light
source and signal processing electronics are remotely located and connected to
the sensor head by lightweight optical fibers. In a single element, the
present
sensor can automatically detect a difference between two orthogonal strains,
as
is required for the determination of torque. In contrast, a torque sensor
based on
conventional strain gauges requires two separate unidirectional strain sensors
mounted at right angles.
As used herein, the terms "comprises", "comprising", "including" and
"includes" are to be construed as being inclusive and open ended, and not
exclusive. Specifically, when used in this specification including claims, the
terms
"comprises" and "comprising" and variations thereof mean the specified
features,
steps or components are included. These terms are not to be interpreted to
exclude the presence of other features, steps or components.
The foregoing description of the preferred embodiments of the invention
has been presented to illustrate the principles of the invention and not to
limit the
24
CA 02413098 2002-11-27
invention to the particular embodiment illustrated. It is intended that the
scope of
the invention be defined by all of the embodiments encompassed within the
following claims and their equivalents.
