Proceedings of the 7Th Annual ISC Graduate Research Symposium

Proceedings of the 7th Annual ISC Graduate Research Symposium

ISC-GRS 2013

April 24, 2013, Rolla, Missouri

Kenneth Smith

Department of Electrical and Computer Engineering

Missouri University of Science and Technology, Rolla, MO 65409

1 © 2013

High Temperature Sensor Based on Fabry-Perot Interferometer in Ceramic Coaxial Cable

1 © 2013

Abstract

The application of a coaxial cable Fabry-Perot interferometer (CCFPI) as a high temperature sensor is analyzed. Coaxial cable sensors provide the advantages of fiber optic sensors, such as the ability to be embedded and multiplexed, while overcoming the limitations created by the fragility of fibers. The fabrication process of the CCFPI, using a computer controlled drill system to create holes in the cable is discussed. Through the process of subjecting the CCFPI sensor to multiple heating and cooling cycles, the low sensitivity of the sensor to temperature cycling is shown. The CCFPI sensor is shown to have the same temperature sensitivity while being heated as while cooling. The high temperature performance and survivability of the CCFPI is investigated. The CCFPI sensor demonstrates an operating range of 180°C to 650°C with a linear temperature sensitivity of -108 kHz/°C.

1. Intoduction

In recent years, a significant amount of research has been focused on creating distributed and embeddable sensors for structural health monitoring (SHM) systems. The need for such research is in large part driven by the ever deteriorating conditions of the current infrastructure. Many of the long-span bridges used today were built in the 1950s and were only designed for a 40 to 50 year life span [1]. Sensors for SHM can be used in a variety of structures. These structures include building, piles, roads, pipelines, tunnels and damns [1]. The most researched structure for SHM use is bridges. Within bridges, the possible areas for sensor deployment include bridge cables [2], fragility curves for highway bridges [3], rebar used to strengthen aging bridges [4] and especially in concrete bridges [1]. The sensors in SHM systems can be embedded during the manufacturing process [5] or surface mounted after construction. In either case, SHM can identify problems in early stages and verify the effectiveness of repairs [6].

Current methods for monitoring structures include routine inspection and maintenance and monitoring using a limited amount of convention sensors. Visual inspection is not always sufficient to detect all damages in a structure. In addition to their unreliable nature, inspections require the structure be taken out of service [7]. Conventional sensors are made up of strain gauges, pressure cells, accelerometers, temperature sensors and more. These sensors have significant drawbacks. Due to their large size and lack of robustness, conventional sensors are difficult to embed. Conventional sensor can also be affected by electromagnetic radiation [1]. Most of the sensors can only measure one parameter. This increases the number of sensors needed to monitor a structure and the number of algorithms needed to analyze the data.

Fiber optic sensors (FOSs) provide an answer to many of these problems. These sensors are light weight, small, highly sensitive, immune to electromagnetic interference, corrosion resistant and have the ability to be multiplexed [8- 13]. Fiber optic sensors also have the ability to measure multiple parameters such as strain, temperature, pressure and magnetic field [10]. These parameters can even be measured simultaneously in some circumstances. This can lead to problems however and must be compensated for. This is commonly overcome by temperature compensation from an isolated second sensor [14] or by using the wavelength shifts from sensors with two different types of gratings [15]. One the most common type of FOSs is the extrinsic Fabry-Perot interferometer. This sensor has the ability to measure with high sensitivity, but suffers from a limited dynamic range [16]. In fact, fiber optic cables in general are fragile; an uncoated fiber grating can break under 1% strain [16]. To overcome this limitation, special protective coatings can be applied. However, even with protective coatings FOSs still need suitable protective housing when embedded in concrete [17]. In certain types of FOSs, splicing and epoxy are used during the fabrication process. The difficulty in reproducing the effects of the splicing and epoxy results in a need to calibrate each sensor before use [18]. Other problems with extrinsic Fabry-Perot interferometers are the low coupling efficiency, required alignment and packaging problems [19]. In intrinsic Fabry-Perot interferometers, the reflections are created by physically changing the fiber with chemicals or lasers, thus eliminating the need to splice or use epoxy. These chemicals can be dangerous and the equipment needed for fabrication can be expensive. In fiber Bragg grating sensors, the fabrication process can be complex and the resulting grating can be fragile [20]. Fiber Bragg gratings also suffer from limited temperature-induced spectral displacements [21].

A possible solution to the limitations of FOSs can be found in coaxial cables. Many forms of coaxial sensors are already used in various detection methods. Leaky coaxial cables are used in object detection systems [22]. Open-ended coaxial cables are used in material analysis [23]. Coaxial cables have even been used to detect cracks and measure strain [24]. Coaxial cables and fiber optic cables operate under the same electromagnetic theory. Because of this, many principles developed in optics can be applied to electronic signals, such as periodic structures and lasing [25, 26]. This paper will address the application of FOS theory in microwave structures. The concept of fiber Bragg gratings have been applied to coaxial cables, resulting in coaxial cable Bragg gratings [27, 28]. These coaxial cable sensors are larger in diameter than their fiber optic counterparts and can survive higher strain [27].

2. Theory

The temperature sensor discussed in this paper operates on the principles of a Fabry-Perot Interferometer (FPI). A schematic of a FPI is shown in Fig. 1. The input signal propagates until reaching the first impedance change. At the interface, the electromagnetic wave (EM) is partially reflected. The transmitted portion continues until reaching the second impedance change, where it is again partially reflected. The EM wave will continue to reflect back and forth between the two impedance changes. Each time it reaches an impedance change, the EM wave will be partially reflected. All of the EM waves that are transmitted from the second impedance change will interfere in the transmission spectrum. Likewise, all of the waves propagating left from the first impedance change will interfere in the reflection spectrum. As the waves interfere constructive interference in the transmission spectrum will occur when the conditions of Eq. 1 are satisfied. Constructive interference in the transmission spectrum will result in destructive interference in the reflection spectrum. The distance between the two impedance changes is given by L. The index of refraction of the dielectric is given by n. The order of interference is m and λ is the wavelength.

2nL=mλ (1)

The basic sensing principle behind the FPI is the length L of the sensor will change under mechanical and thermal strain. As the material in the sensors is heated, it will expand according to the coefficient of thermal expansion. This will result in a longer length and wavelength. The resonance minimums found in the reflection spectrum were monitored. A shift in the frequency of each minimum occurred as the temperature changed. The response of the CCFPI is displayed as the change in frequency as a function of temperature.

Figure 1: Schematic of Fabry-Perot Interferometer in Coaxial Cable

3. Fabrication of sensor

Two reflections were created in the ceramic coaxial cable by drilling into the cable. The coaxial cable was attached to the platform of a computer controlled drill. One end of the cable was attached to a Hewlett Packard 8753E vector network analyzer (VNA) using a SMA connector. Using 1601 data points the VNA measured the time domain reflectometry (TDR) of the cable. When the drill bit comes in contact with the inner conductor, a short will occur. The TDR will display the short as a large peak. By monitoring the TDR of the cable, the depth of the hole can be controlled. Using a 2mm diamond tipped bit, a hole was made through outer sheath, outer conductor and dialectic layer. The bit was stopped when the short was indicated in the TDR measurement. The cable was rotated 180° and a second hole was drilled through the outer sheath, outer conductor and dielectric layer. The remaining dielectric around the inner conductor was removed with hand tools until the inner conductor was completely exposed in the opening. A second set of holes was fabricated using the same method. The spacing between the two sets of holes was 10cm. The resulting TDR after the reflections were created is shown in Fig. 2. The reflections created by this method had amplitudes of 0.04Units or 4% reflection.

Figure 2: TDR of Coaxial Cable with Reflections from Holes

4. Temperature cycling response

4.1. Experimental Set-up

To test the response of multiple heating and cooling cycles on the CCFPI sensor, a CCFPI was placed in a Lindberg/Blue Mini-Mite Tube Furnace. The schematic of the experiment can be seen in Fig. 3. Argon gas was pumped into the glass tube through a metal tube and rubber stopper at one end of the furnace. The CCFPI was attached to the VNA and inserted in the other side of the glass tube. The space between the glass tube and coaxial cable was filled with stuffing. The hole closest to the VNA was placed 3.1cm from the center of the furnace. The second hole was 6.9 cm from the center of the furnace, in the opposite direction. The Ar gas was set to flow at a rate of 70mL/min. Before the temperature was increased, Ar gas was allowed to flow for 5 minutes.

Figure 3: Schematic of CCFPI Experiment in Tube Furnace

The VNA was set with a start frequency of 30kHz and a stop frequency of 6GHz. The total number of data points was set to 1601. The TDR of the CCFPI was measured. Using the built in gate feature, a gate was applied around the two reflections in the time domain. The gate had a start location of 8.26ns, stop location of 9.733ns, was centered at 8.997ns and had a span of 1.473ns. The resulting interference pattern from S11 can be seen in Fig. 4.

Figure 4: Interference Pattern of Coaxial Cable Fabry-Perot Interferometer

The furnace was set to a temperature of 25°C. Once the furnace reached the set temperature, it dwelled for 5 minutes. An averaging factor of 15 was used to reduce random noise in the system. The interference pattern was measured immediately before the dwell was complete. The temperature was then increased to 50°C. After allowing the furnace to dwell at the new set temperature, the interference pattern was measured. This process was repeated for 100°C, 150°C, 200°C, 250°C and 300°C. Once the measurement was taken at 300°C, the furnace was opened and allowed to cool to 20°C. Upon reaching 20°C, the furnace was closed and another heating cycle was started. The interference pattern was measured at the same temperatures as the previous cycle. The process was repeated once more, for a total of 3 cycles. A fourth cycle was completed, but the temperature only ranged from 50°C to 300°C.

4.2. Results

The interference pattern in Fig. 4 shows a resonance at approximately 3.7GHz. A cubic interpolation algorithm was applied to each data set over the range of 3.6GHz to 3.8GHz. This increased the number of data points by a factor of 10. The resonance frequency at each temperature was measured from the interpolated data. Using the resonance frequency from 25°C in the first cycle as a reference, the shift in resonance frequency was calculated for each temperature. The process was repeated for each cycle, still using the frequency for 25°C from the first cycle as a reference. The shift in the resonance frequency as a function of temperature was plotted in Fig. 5. There is a large shift from 25°C to 50°C during the first cycle. The frequency shift vs. temperature curves for cycles 2, 3 and 4 are identical from 50°C to 150C°. From 150°C to 250°C the curves deviate from one another very little. The largest discrepancy for a resonance frequency at a given temperature occurs at 300°C. This could lead to errors when measuring temperatures in the 275°Cto 300°C range. Overall, this indicates that after the initial frequency shift during the first cycle, the CCFPI sensor is not significantly affected by temperature cycling.

Figure 5: Resonance Frequency Shift vs. Temperature for Multiple Temperature Cycles

4.3. Summary

This section discussed the effects that undergoing many heating and cooling cycles had on the CCFPI sensor. A large change in resonance frequency was observed during the first heating from 25°C to 50°C. The resonance frequencies for 50°C through 150°C were the same for cycles 2 through 4. The resonance frequencies corresponding to temperatures 25°C through 200°C did not change much throughout the temperature cycling. The CCFPI sensor discussed in this paper has a low sensitivity to temperature cycling.

5. Response of Sensor During Heating and Cooling

The last section discussed the ability of the CCFPI sensor to relate the same resonance frequency to a temperature after temperature cycling. In this section, the ability of the sensor to relate the same resonance frequency to a temperature during heating and cooling is discussed.

5.1. Experimental Set-up

The CCFPI was attached to the VNA and placed into the Lindberg/Blue tube furnace. Argon gas was pumped into the tube at a rate of 70mL/min. A gate was applied around the reflections in the TDR measurement. The gate started at 8.166ns, stopped at 9.779ns, was centered at 8.973ns and had a span of 1.613ns. In order to achieve higher accuracy in the measurements, the interference pattern was measured over the range of 3GHz to 6GHz. A total of 1601 data points were used. The furnace temperature was increased to 50°C. After reaching the set temperature, the furnace dwelled for 5 minutes. The interference pattern from S11 and the TDR from S11 were measured before the dwell had finished. The temperature was increased in increments of 25°C. After allowing the furnace to dwell at each set temperature for 5 minutes, the interference pattern and TDR were measured. Upon completing measurements at 500°C, the furnace temperature was decreased in increments of 25°C. Again, the furnace was allowed to dwell for 5 minutes after reaching the desired temperature. Both the interference pattern and TDR were measured as the temperature was decreased.