High-Temperature, High Bandwidth, Fiber-Optic, MEMS
Pressure Sensor Technology for Turbine Engine Component
Luna Innovations, Inc.
Applied Research Associates, Inc.
Fiber-Optic, Sensors, MEMS, High-Temperature, Turbine Engines
Acquiring accurate, transient measurements in harsh environments has always pushed the limits of available
measurement technology. Until recently, the technology to directly measure certain properties in extremely high
temperature environments has not existed. Advancements in optical measurement technology have led to the
development of measurement techniques for pressure, temperature, acceleration, skin friction, etc. using extrinsic
Fabry-Perot interferometry (EFPI). The basic operating principle behind EFPI enables the development of
sensors that can operate in the harsh conditions associated with turbine engines, high-speed combustors, and other
aerospace propulsion applications where the flow environment is dominated by high frequency pressure and
temperature variations caused by combustion instabilities, blade-row interactions, and unsteady aerodynamic
phenomena. Using micromachining technology, these sensors are quite small and therefore ideal for applications
where restricted space or minimal measurement interference is a consideration. In order to help demonstrate the
general functionality of this measurement technology, sensors and signal processing electronics currently under
development by Luna Innovations were used to acquire point measurements during testing of a transonic fan in
the Compressor Research Facility (CRF) at the Turbine Engine Research Center (TERC), WPAFB. Acquiring
pressure measurements at the surface of the casing wall provides data that are useful in understanding the effects
of pressure fluctuations on the operation and lifetime wear of a fan. This measurement technique is useful in both
test rig applications and in operating engines where lifetime wear characterization is important. The
measurements acquired during this test also assisted in the continuing development of this technology for higher
temperature environments by providing proof-of-concept data for sensors based on advanced microfabrication and
Although sensors using current technologies are able to provide measurement range, frequency response and most
other requirements for aerospace applications, these sensors perform well only in benign environments. There are
at least two operating regimes where current technology fails to meet minimum requirements: environments with
either high temperatures and/or high electromagnetic interference. Harsh environmental conditions in aerospace
applications therefore constitute an extreme in at least one of these two conditions.
As engine designers continue to push the limits of current propulsion technology, the need to monitor flow
conditions in harsh environments for performance evaluation and control of new propulsion systems becomes
paramount. Because the flow environment inside an operating propulsion system can be harsh, the demands
placed on sensor and instrumentation technology are considerable (Figure 1). For example, in order to meet
design goals for the Integrated High Performance Turbine Engine Technology (IHPTET) program or the Versatile
Affordable Advanced Turbine Engines (VAATE) initiative, engine designers will need health monitoring and
engine test instrumentation that can withstand temperatures as high as 1300° C (~2400° F) while providing
adequate frequency response and low flight weight. Control methodologies developed to avoid high-cycle
fatigue (HCF), combustion instabilities or compressor surge will require sensors capable of operating in the
engine environment. The combination of high temperature tolerance and high frequency response cannot be met
by currently available technology. As a result, these and other propulsion programs have created a clear need for
accurate, high frequency sensors capable of operating in harsh environments.
Combustor / Turbine
~ 2000° C
1300° C - 1400° C
900° C, Sapphire-based Fiber Optic
600° C, Conventional Fiber Optic
550° C, Piezoelectric
350° C, Electric Resistance
250° C, Piezoresistive
Figure 1. High temperature regimes for sensors and aerospace applications
A heavily instrumented validation engine can have up to 3000 pieces of instrumentation with a typical engine
having about 1000.i During validation testing, the instrumentation is subjected to extreme temperatures and
pressures, high rotational speeds, vibration, and noise levels, and must survive and function in an electrically
noisy and metal rich environment. Today’s validation programs are more aggressive than those past and expose
instrumentation to more frequent and extreme test cycles. Better engine performance with fewer engine parts
requires higher rotor speeds and increased gas path temperatures and pressures. As a result, new and innovative
measurement techniques and sensors are required. One of these desired innovative sensors is a dynamic pressure
transducer, capable of operating at high temperatures without the need for external cooling. An application that
has been touted for such a sensor is for active compressor surge control.
Currently, engine designers must build in a margin to the compressor to protect the engine from surging. This
margin is essentially an over design of the compressor. The total amount of margin is typically on the order of
35%, meaning the engine is carrying 35% more compressor than it needs for steady-state operation. Compressor
designers build in margin for engine acceleration, inlet distortion, clearance changes, engine deterioration, engine-
to-engine variance, and random variation. However, this excess margin results in extra cost, weight, compressor
length and part count. ii Przybylkoiii states that if active compressor surge control could be achieved, it is possible
that a stage of compression would be eliminated from an engine with the following benefits for a typical military
• +5% thrust/weight
• -1.5% thrust specific fuel consumption
• -3.2% acquisition cost
• -1% operating cost
• Surge free operations
• Extension of maintenance intervals
• Increase inlet distortion envelope
The need for accurate high-temperature dynamic pressure measurements is an integral step to the implementation
of active surge control methodologies. Dynamic pressure measurements in the compressor have a direct role in
the detection of instabilities that are often precursors to surge or rotating stall. Typically, pressure measurements
in high temperature environments are made through a sensing port, which separates the transducer from the
extreme temperature. For high frequency applications, it is generally necessary to reduce the distance from the
sensor to the environment to a minimum. In the extreme, flush diaphragm transducers may be used that expose
the transducer sensing element directly to the cavity where the pressure fluctuations are to be measured. Placing
the transducer that close exposes the transducer to high temperatures. Even with temperature compensation, the
accuracy of the absolute pressure component of the measurement is usually compromised. Typical pressure
transducers are limited to about 500° F, while in today’s large gas turbine engines, compressor exit temperatures
are on the order of 1200 to 1400° F, meaning that current pressure transducers are not sufficient for these
conditions. The normal technique used to handle this temperature situation is to either air or water cool the
transducer. The additional complexity, weight, cost, and potential safety issues due to this cooling typically
means that this is unacceptable in revenue service. Although surge/stall events are relatively low frequency, a
high temperature transducer with a high bandwidth would have multiple applications of interest. Cullinane states
that “a transducer with a frequency response of 100 kHz is desirable, providing a multipurpose device for 1)
surge/stall detection, 2) measurement of individual blade pressure pulses, 3) combustor instability, and 4) pressure
distributions on rotating hardware.” i
Fiber-optic sensors are seen by many researchers as the best available technology for acquiring measurements in
these harsh environments. As a result, these sensors have begun to replace conventional electrical sensors in
specific applications. For example, the transducing technique used by fiber-optic sensors does not involve
electrical signals, so they are essentially immune to electromagnetic interference. These sensors have also been
demonstrated to function at higher temperatures than their electrical counterparts. However, the modest increase
in operating temperature fiber-optic sensors afford often do not justify their additional cost. In order for fiber-
optics to compete with other sensor technologies, the increased temperature tolerance these sensors offer must be
significant enough to justify their added expense. Efforts to address this issue are at the forefront of current fiber-
optic sensor research.
A variety of different fiber-optic sensing techniques have been put to practical use in the last two decades,
including intensity-based interrogation and interferometry. One the most versatile techniques for a variety of
fiber-optic sensor applications is extrinsic Fabry-Perot interferometry, or EFPI.
The transducing mechanism used by these sensors is based on the extrinsic Fabry-Perot interferometer (EFPI)iv
developed by Kent Murphy et al. EFPI-based sensors use a distance measurement technique based on the
formation of a low-finesse Fabry-Perot cavity between the polished end face of a fiber and a reflective surface,
shown schematically in Figure 2. Light is passed through the fiber, where a portion of the light is reflected off the
fiber/air interface (R1). The remaining light propagates through the air gap between the fiber and the reflective
surface and is reflected back into the fiber (R2). These two light waves interfere constructively or destructively
based on the path length difference traversed by each. In other words, the interaction between the two light waves
in the Fabry-Perot cavity is modulated by the path length. The resulting light signal then travels back through the
fiber to a detector where the signal is demodulated to produce a distance measurement. Several different
demodulation methods exist to convert the return signal into a distance measurement.
Figure 2. Extrinsic Fabry-Perot interferometer concept
A basic demodulation system using single wavelength interrogation is shown in Figure 3. A laser diode supplies
coherent light to the sensor head and the reflected light is detected at the second leg of the optical fiber coupler.
The output can then be approximated as a low-finesse Fabry-Perot cavity in which the intensity at the detector is,
cos ∆φ ,
= 1 + 2 +
if A1 and A2 are the amplitudes of R1 and R2, and ∆φ is the phase difference between them. The output is then a
sinusoid with a peak-to-peak amplitude and offset that depends on the relative intensities of A1 and A2. A phase
change of 360 degrees in the sensing reflection corresponds to one fringe period. If a source wavelength of 1.3
µm is used, the change in gap for one fringe period is 0.65 µm. The drop in detector intensity is due to the
decrease in coupled power from the sensing reflection as it travels farther away from the single -mode input/output
fiber. By tracking the output signal, minute displacements are determined. The disadvantage of this type of
demodulation system is the non-linear transfer function and directional ambiguity of the sinusoidal output. If gap
changes occur at a peak or valley in the sinusoid as shown in Figure 4 at π, 2π, 3π, ..., they will not be detected
because the slope of the transfer function is zero at those points. The sensitivity of the system correspondingly
decreases at points near multiples of π. If the direction of diaphragm movement changes at a peak or valley, that
information is lost, which causes directional ambiguity in the signal.
One approach to solving these problems is to design the sensor head so that the maximum gap does not exceed the
linear region of the transfer function. The linear region of the sinusoidal transfer function is shown in Figure 4(a).
Confining operation to the linear region places difficult manufacturing constraints on the sensor head by requiring
the initial gap to be positioned at the Q-point of the transfer function curve. In addition to the difficult
manufacturing constraints, the resolution and accuracy are limited when the signal output is confined to the linear
region. To solve the non-linear transfer function and directional ambiguity problems, alternative signal
demodulation approaches can be used such as dual wavelength interrogation and white light interferometry.v
Diaphragm Discplacment (microns)
Figure 3. Basic demodulation system and typical output as the diaphragm moves away from the
90° phase difference
Phase of signal
Figure 4. Single wavelength and Dual wavelength interferometric signal output
Demodulation using dual wavelength interrogation
Dual wavelength interrogation is a slightly more advanced demodulation technique than the single wavelength
approach. This demodulation technique has seen considerable use in a variety of applications. The system uses
advanced digital signal processing and analog filtering to provide unambiguous, accurate, high frequency
measurements. Two laser diodes at slightly different wavelengths are used to measure a relative gap change in an
optical sensor. By choosing the wavelengths of the two light sources correctly, the modulated return signals are
90° out of phase, or in quadrature, as shown in Figure 4 (b). This technique allows very rapid changes in the gap
to be tracked and calculated, and thus provides a method for demodulating signals with frequencies up to 1 MHz.
In the region of 90° phase difference, gap values and directional changes can be monitored by plotting the two
signals on opposite axes to create a Lissajous figure, as shown in Figure 5. By choosing source wavelengths a
few nanometers apart, the range of unambiguous phase detection increases. One can then track the rotations
around the circle and converting α to a displacement value, one finds the perturbation in terms of gap change.
This relative displacement is then converted into a proportional voltage. This system is an improvement over the
single wavelength system in that unambiguity is maintained over multiple fringes even if rapid changes in sensor
output occur, while maintaining a high-bandwidth.
α= tan-1(V /V )
Figure 5. Idealized Lissajous figure of dual-wavelength output.
The basis of the fiber-optic pressure sensors is shown in Figure 6. An optical fiber and glass tube fiber spacer are
bonded in the center opening of the sensing element using epoxy. The optical gap between the bottom of the
diaphragm and the end face of the fiber is a Fabry-Perot cavity. As described above, light interference resulting
from the internal reflection of light at the fiber end face (R1) and reflected light off of the bottom surface of the
diaphragm (R2) is used to monitor the optical path length. The optical gap varies with diaphragm deflection,
which in turn varies with applied pressure. The reference port on the bottom of the sensing element acts as a vent
through which air can pass to maintain a constant pressure on the reference side of the diaphragm.
S i l i c o n
D i a p h r a g m
Hollow Core Fiber
Figure 6. Assembled fiber-optic pressure sensor.
Figure 7. Packed fiber-optic pressure sensors: (a) screened and (b) flush-mounted version.
Optical fiber pressure sensors that utilize EFPI have exceptional tolerance to elevated temperatures. For
example, the data presented in Figure 8a and b show gap change as a function of both pressure and temperature
for an experimental EFPI-based pressure sensor. The pressure sensor demonstrated operation up to 300° C
(Figure 8a) with an expected thermal zero shift that is apparent in the data. Further experiments demonstrated that
the sensor is repeatable up to this temperature. Data in Figure 8b show that the sensor can operate at 600° C
without damage, though more data is required to determine if the sensor is repeatable at this temperature. These
results are from those uncompensated for temperature. It is possible, as currently being perfected, to measure the
temperature of the pressure sensor diaphragm simultaneous and with the same fiber as the pressure sensor. This
procedure is termed gap division multiplexing and involves determining an interference pattern that occurs
between the reflection on the front and back sides of the diaphragm while tracking the air gap that defines the
pressure measurement. The measured thickness of the diaphragm is directly related to its temperature. Through
this method, an accurate and effective means of temperature compensation can be achieved.
High frequency response characteristics for this sensor design were determined during a shock tube test at Wright
State University. Data from this test are plotted in Figure 9. In this figure, the fiber-optic sensor data are
compared to data acquired simultaneously from a co-located, high frequency, piezoresistive pressure sensor.
Both sets of data are smoothed (filtered) using a 10-point moving average to eliminate spurious signals. The data
from both sensors show virtually the same response to a moving shock wave, which clearly demonstrates the high
frequency capability of the fiber-optic pressure sensor. The power spectrum of the unfiltered fiber-optic sensor
data is also shown in Figure 9. The data show the resonant response of the sensor, which occurred at
approximately 137 kHz. This experimental sensor demonstrated a measurement resolution of approximately 0.003
psi when used with the most advanced demodulation system available.
Figure 8. Temperature behavior of a fiber-optic pressure sensor up to (a) 300° C and (b) 600° C
Piezoresistive Pressure Sensor
Fiber Optic Pressure Sensor
Resonant (Natural) Frequency
Pressure (psi) 16.0
Power ( psi
Figure 9. High frequency pressure sensor data from WSU shock tube and the power spectrum of the
results demonstrating the natural frequency of the sensor.
Overall, the fiber-optic pressure sensor demonstrates excellent qualities, including its large temperature range,
high resolution, and low combined uncertainties after temperature compensation. Table 1 shows the basic
requirements for a general-purpose high temperature dynamic pressure transducer as defined by the Propulsion
Instrumentation Working Group (PIWG), a gas turbine engine manufacturer consortium, and the comparison to
the Luna fiber-optic pressure sensor. As can be seen, the current fiber-optic sensor meets many of the goals,
significantly surpassing some of them. Where the sensor falls short, advances being developed (such as a SiC
sensor body as described below and temperature compensation as described above) to exceed the desired
parameters. These fiber-optic pressure sensors would be a natural choice many turbine engine applications, such
as the compressor surge control application, combustion control, or turbine characterization.
Table 1. Desired pressure transducer performance characteristics and comparison to the Luna fiber-optic
pressure sensor. vi
Luna Sensor Characteristics
100 psi (design feature)
> 125 kHz
450°C (900°C for SiC version)
Combined Uncertainties @ Ambient
Temperature Effect on Zero
+/- 2% FS
4% FS (uncompensated)
Temperature Effect on Sensitivity
+/- 5% FS (uncompensated)
Currently, gas turbine engine manufactures perceive optical techniques as having higher cost and lower reliability
as compared to more conventional electrical techniques. As a result, these sensors have seen little or no use
within the industry. However, there is interest and their use could become more prevalent with the development
of lower cost signal conditioning, effective packaging, and an evaluation of their accuracy and operational
characteristics in a “real world” gas turbine test. i With this in mind, sensors and signal processing electronics
currently under development by Luna Innovations were used to acquire point measurements during testing of a
transonic fan in the Compressor Research Facility (CRF) at the Turbine Engine Research Center (TERC),
WPAFB. The Compressor Research Facility Experimental Rig (CRFER) is two-stage transonic fan designed by
General Electric and owned by the USAF. Since 1984, CRFER has served as a test bed for the investigation of
various rotor designs and casing treatments, and for the aeromechanical evaluation of rotor blade response. The
1st stage rotor of CRFER is an advanced state-of-the-art design comprised of 16 low-aspect ratio blades with the
geometric properties described in Table 2. The design speed (100%) of this compressor is 13288 rpm, so the
maximum blade tip speed is 483.5 m/s, or approximately Mach number 1.7 for an inlet total temperature of 300K.
Table 2. Geometric Parameters of the 1st -Stage
S1 R2 S2
Average Aspect Ratio
Rotor Tip Radius (cm)
Inlet Radius Ratio
Average Radius Ratio
Average Tip Solidity
Figure 10. Test article schematic
Static pressure measurements were acquired from the “instrumented region” over the 1st stage rotor blade tips
shown in Figure 10. Ten fiber-optic pressure sensors were mounted on the fan casing in a removable fixture at
points starting near the leading edge of the fan blades. Historically, high-frequency piezoresistive-type pressure
sensors are normally mounted in this fixture instead of fiber-optic sensors. Therefore, high-frequency pressure
data from previous tests of this same test article were available for comparison to the fiber-optic sensor data
acquired during this test vii. Ten pressure ports for low-frequency static pressure measurements were also co-
located with the fiber-optic sensors. These ports were pneumatically connected to external pressure sensors that
are part of the TERC data acquisition system. Static pressure near the casing wall fluctuates as each blade passes.
The dominant frequency of this fluctuation is a function of the rotor speed and the number of blades on the rotor.
This is the blade passing frequency, which is the rotor speed multiplied by the number of blades. Capturing the
blade passing frequency and the time history of the static pressure fluctuation at the casing wall was the purpose
of this test.
Measurements were acquired at 77% design speed (10232 rpm) near the stall line and at 90% design speed (11959
rpm) near peak operating efficiency. Raw (uncalibrated) data acquired at 77% speed from fiber-optic sensor #6
are shown in Figure 11. Plotted with this data in the figure are ensemble -averaged data acquired during a
previous test using a piezoresistive type pressure sensor at the same measurement location. Although there is
considerable noise, the fiber-optic sensor data show a definite correlation with the piezoresistive sensor data. A
frequency spectrum plot of the fiber-optic sensor data (Figure 12) shows that the expected blade passing
frequency (2765 Hz) is evident in the data. Ensemble -averaging the fiber-optic sensor data further improves
agreement between it and the piezoresistive sensor data. Shown in Figure 11, the ensemble -averaged fiber-optic
sensor data seem to agree well with the piezoresistive data.
Fiber Optic Sensor Data (Not Calibrated)
Fiber Optic Sensor Data (Not Calibrated)
Piezoresitive Sensor Data
Piezoresitive Sensor Data
Figure 11. Data Comparison, (a) unaveraged fiber-optic sensor data and, (b), ensemble averaged fiber-
optic sensor data (blade period averaged) both compared to ensemble averaged piezoresistive sensor data
Blade Passing Frequency
Figure 12. Frequency spectrum (DFT) of fiber-optic sensor data
SiC Pressure Sensor Development
As described above, current EFPI-based fiber-optic sensors are capable of sustained operation up to
approximately 600° C (1100° F). This limitation is due to the small number of materials that can withstand
higher temperatures and still exhibit required mechanical properties for sensor operation. For example, silicon is
often used as a diaphragm material for EFPI-based pressure sensors and accelerometers. The melting point of
silicon is 1415° C, however it begins to deform plastically at a much lower temperature, making the useful
mechanical limit of the material somewhere between 550° C and 600° C (1025° F to 1100° F). Another limitation
is the operating temperature of the silica fiber, generally quoted as 900° C (1650° F), above which the core and
the cladding material begin to migrate. Extending the operating temperature range of EFPI-based sensors must be
accomplished by using materials with higher thermoplastic and/or melting points. Many research groups are
currently experimenting with more temperature tolerant materials. Replacing silica fiber with sapphire fiber has
been considered a viable method of increasing the temperature range of optical fibers for several yearsviii,ix,x.
Sapphire has excellent optical qualities, with a melting point over 2000° C (~3600° F). Current work has begun
on replacing silicon as the sensor’s structural material and replacing it with SiC. Silicon carbide has an
operational temperature of over 1100° C (2000° F) with enhanced strength and micromachining dimensional
tolerances. By combining a SiC sensor with sapphire optics, a pressure sensor can be fabricated that can
withstand the harshest turbomachinery environment.
0.014 mm Thick
SiC Sensor Body
Figure 13. Sketch and photograph of the SiC pressure sensor.
Silicon Carbide Pressure Sensor Burst Test
Figure 14. Pressure calibration of the fiber-optic, micromachined SiC pressure sensor.
The prototype micromachined silicon carbide-sapphire fiber-optic pressure sensor design is shown in Figure 13.
The design is similar to a proven Luna pressure sensor design based upon a silicon membrane, glass body and
silica fiber as illustrated in Figure 6. The SiC sensor is 3X3 mm and 500 microns thick, providing a small
envelope for detailed surface pressure measurements. Furthermore, these sensors can be easily packaged into
arrays, as already demonstrated with silicon/glass pressure sensorsxi, for complete surface pressure mapping in
propulsive and other extreme environments. A calibration of the first prototype of the SiC pressure sensor
combined with a silica fiber is presented in Figure 14. The sensor preformed as expected, demonstrating excellent
repeatability. Further development is in progress.
Testing at the CRF showed that current fiber-optic sensor technology provides sufficient frequency response to
resolve amplitude and frequency data in flow conditions representative of those found in operating turbine
engines. The successful nature of these tests demonstrates that these fiber optic pressure sensors are a possible
technology solution to many operational turbomachinery applications, including active compressor surge control.
The authors would like to thank Dr. Tom Jackson of the Air Force Research Laboratory (ARFL), Propulsion
Directorate, for sponsoring most of the research discussed herein. The authors would also like to thank Dr.
Joseph Schetz of Virginia Tech for his contributions to the development of this sensor technology.
i Cullinane, W., Strange, R., “Gas Turbine Engine Validation Instrumentation: Measurements, Sensors, and Needs”, SPIE
Conf. On Harsh Environment Sensors, SPIE vol. 3952, Sept 1999
ii Przybylko, S., “Active-Control Technologies for Aircraft Engines”, Micro Device Applications Workshop, Cocoa Beach,
FL, Jan 1998
iii Przybylko, S., “MEMS for Aircraft Engines”, DARPA ELTO MEMS for Harsh Environment Workshop, Oct. 1998
iv Murphy, K., Gunther, M., Vengsarkar, A., and Claus, R., "Quadrature phase-shifted, extrinsic Fabry-Perot optical fiber
sensors," Optics Letters, Vol. 16, No. 4, February 1991, pp. 173-275
v Dakin, J., and Culshaw, B., Optical Fiber Sensors: Principles and Components, Artech House, Boston, MA, 1988.
vii Russler, P, “A Technique for Acquiring Tip Flow Field Measurements in Turbomachinery,” Proceedings of the 44th
International Instrumentation Symposium, May 3-7, 1998.
viii Xiao, H., et al, “Absolute Sapphire Optical Fiber Sensor for High-Temperature Applications,” SPIE Vol. 3201, pp. 36-42
ix Murphy, K., et al,“Sapphire fiber interferometer for microdisplacement measurements at high temperature,” Proceedings of
SPIE, vol. 1588, Boston, MA, Sept. 1991.
x Claus, R., et al, “Sapphire fiber strain gage for static and fatigue measurements,”OFC, 1994.
xi Borinski, J., Meller, S., Pulliam, W., Murphy, K., “Optical Flight Sensors for In-Flight Health Monitoring”, SPIE
Proceedings, vol. 3986, Newport Beach, CA, Mar. 6, 2000