A Clamp Fixture with Interdigital Capacitive Sensor for In Situ Evaluation of Wire Insulation

Transcription

1 A Clamp Fixture with Interdigital Capacitive Sensor for In Situ Evaluation of Wire Insulation Robert T. Sheldon and Nicola Bowler Center for Nondestructive Evaluation, Iowa State University, Ames, Iowa 50011, USA Abstract. An interdigital capacitive sensor has been designed and optimized for testing aircraft wires by applying a quasinumerical model developed and reported previously. The sensor consists of two patches of interdigitated electrodes, connected by a long signal bus strip, that are intended to conform to two sides of an insulated wire. The electrodes are deposited using photolithography upon a 25.4-µm-thick Kapton polyimide film. The two electrode patches are attached to the two jaws of a plastic spring-loaded clamp, with each jaw having a milled groove designed such that the electrodes conform to the curved surface of the insulated wire. An SMA connector and cable connect between the electrodes on the clamp and an LCR meter. Segments of pristine M5086/2 aircraft wire, each 10 cm long, were immersed in fluids commonly found in aircraft environments, to cause accelerated chemical degradation. The effects of Jet A fuel, deicing fluid, hydraulic fluid, aircraft cleaner, isopropyl alcohol and distilled water were studied. The frequency-dependent capacitance and dissipation factor of one pristine wire segment and of those degraded in the six fluid environments were measured within the frequency range 100 Hz to 1 MHz. Significant changes in capacitance and dissipation factor were observed for all degraded wires, compared with results for the pristine sample, suggesting the feasibility of detecting insulation degradation in the field. The results were also consistent with those of a similar experiment performed on sheets of Nylon 6, the material that comprises the outermost layer of M5086/2 wire. Keywords: Interdigital Electrodes, Dielectric Materials, Cylinders, Conductors, Capacitance Measurement, Capacitive Sensors, Dissipation Factor, Polymer Absorption, Aircraft Wires, Wire Insulation PACS: c, Ch, Gm, g, Jd, Ci, Me INTRODUCTION The insulation layer of a wire transmitting power and signals is designed to prevent the transfer of energy between its central conductor and potential external conductors. In theory, this insulation layer is a perfect dielectric at all frequencies and voltages with no ability to conduct electric current. However, in reality, the insulator has a small, finite conductivity that will suddenly jump to a very large value when placed in an electric field of sufficient intensity to cause dielectric breakdown. On-board an aircraft, this could cause short circuits of critical systems resulting in dangerous situations. The imperfect insulator is rendered yet more imperfect by extreme environments - extreme temperatures, humidity, mechanical stress, chemical reactions and absorptions, etc. [1]. Exposure to these environments over time will change the dielectric properties of the insulation. One common way of indirectly testing the conductivity of the insulation is through use of an insulation resistance tester, whereby one probe lead is connected to the central conductor and the other around the insulation. A high voltage is applied across the leads and the leakage current is measured; through Ohm s law the leakage resistance is then calculated [1]. Partial discharge analysis is another insulation characterization technique that relies on the transmission of a high voltage pulse down an insulated wire, which will partially discharge in degraded insulation regions and reflect some of the signal back toward the source. The pulse return time can locate the fault position and frequency-domain analysis provides insight into the insulation integrity [2]. The major downsides of these approaches are the high applied voltage that could permanently damage the insulation and the required connection to the central conductor, which may not be possible for a wire that remains actively in situ. Capacitive sensors have been employed to exploit the usefulness of the direct relationship between the material property of permittivity and the measurable quantity of capacitance. In [3] a two-electrode cylindrical coplanar sensor was developed and modeled for the specific application of detecting permittivity changes in wire insulation degraded by extreme heat. A rectangular coplanar array of interdigital electrodes, used to increase the measurable capacitance, was investigated in [4]. In this work, the cylindrical geometry and interdigital electrode concepts are combined and integrated with a plastic spring-loaded clamp for practical interrogation of wire insulation. The final design is tested on aircraft wire segments following immersion in common aircraft chemicals to determine the effectiveness of the sensor

2 Conductor Dielectric Insulation Layers g w l Side 1 ε1 σ Side 2 ε2 ε3 a b s c Electrodes with opposite polarity Conductor FIGURE 1. Cylindrical interdigital capacitive sensor. The radii of the central conductor, cylinder insulation, and sensor substrate are denoted a, b and c, respectively. The electrodes have width w, spacing s, and length l [5]. at detecting changes in the insulation dielectric properties. ELECTRODE DESIGN Sensing Method The sensing electrodes, when applied to the surface of an insulated wire, consist of an array of interdigitated conductor digits that alternate electric polarity around the outer circumference of the wire insulation. A model was developed in [5] and [6] to calculate the capacitance of such a structure when applied to a dielectric-coated conductive cylinder. This model used the geometry, shown in Fig. 1, of a perfect conducting cylinder of radius a and infinite length, surrounded by an insulation layer of radius b and permittivity ε 1, further surrounded by a substrate of radius c and permittivity ε 2, all of which being completely embedded in a infinite space of permittivity ε 3, which is always set to the value for free space. The electrode digits, with infinitesimal thickness, width w, spacing s and length l were placed at radius c. In Fig. 1, it is shown that there exists an azimuthal gap g between two separate interdigital electrode arrays. This represents a discontinuity in spatial periodicity of the digits around the circumference of the wire when applied with a clamping device where the two sides are affixed to the two jaws of the clamp. Thus one side will have N D positive or negative digits interdigitated with N D + 1 negative or positive digits, where N D 1. This arrangement is mirrored on the opposite side of the wire with opposite polarity, resulting in a total number of digits N T = 4N D + 2. Neighboring digits of opposite polarity accumulate charges along their edges nearest to adjacent digits but charge flow between digits is inhibited by the dielectric material separating the digits, namely the air, the electrode substrate, supporting structures, and, most importantly, the insulation of the wire that is being sensed. The separated charges generate a fringing electric field that penetrates and polarizes the material comprising the insulation layer, creating dipole moments that store electric energy. The amount of energy stored in the dipoles depends upon the insulation permittivity, and the ability of the generating structure to store this energy is defined as capacitance C. Because capacitance is directly proportional to permittivity, this makes a capacitor-type sensor a good method of indirect evaluation of the insulation dielectric properties and is easily measurable on a testing device such as an inductancecapacitance-resistance (LCR) meter. The clamp sensor was developed for testing of 20 AWG M5086/2 (MIL-W-5086/2) aircraft wires, manufactured by Allied Wire & Cable, Inc., the specifications for which are given in Table 1. Given the outer wire diameter of 2.2 mm, model simulations in [5] found that the optimal digit structure, achieving a balance between high capacitance necessary for minimizing measurement error and sufficient penetration depth of the wire insulation, is w = 0.1 mm, s = 0.3 mm and N T = 14. As the interdigital electrode digits are designed to be affixed to the two jaws of a clamp, the practical design must allow the jaws to completely open without restriction while maintaining total electrical contact. Thus, the two digit sides, each matching the jaw length with l = 25 mm, were spaced apart by a 100-mm-long signal bus, as shown in

3 TABLE 1. Composition and dimensions of the layers comprising a 20 AWG M5086/2 wire. Layer Size (mm) Tin-coated stranded copper conductor diameter 1.0 ± 0.1 PVC thickness (inner layer) 0.22 ± 0.03 Glass fiber thickness (middle layer) 0.22 ± 0.03 Nylon 6 thickness (outer layer) 0.15 ± 0.03 Total diameter 2.2 ± 0.1 Interdigital electrodes Signal bus LCR meter contact pads NT = 14 w = 0.1 mm s = 0.3 mm l = 25 mm FIGURE 2. Planar schematic of the interdigital capacitive sensor used in the plastic spring clamp. The bus strip is 100 mm long. Fig. 2. Two test pads are connected to the bus leads via separate traces, which are designed to allow an LCR meter connection away from the jaw region of the clamp. Both the digits and signal bus traces are 0.1 mm wide and composed of 17.8-µm-thick bare copper deposited photolithographically by American Standard Circuits, Inc., upon a 25.4-µmthick substrate film of Kapton polyimide. Excess substrate film around the outside shape of the entire structure was removed for placement in the clamp. Substrate Layer Effect Although the purpose of the sensor is to sense changes in the permittivity of the insulation layer(s) of a wire, the fringing electric field generated by the electrode digits must pass through the substrate layer that is structurally supporting them. As is the case with any fringing electric field, the magnitude of the field decays exponentially with distance normal to the generating surface, thereby decreasing its influence on the measured capacitance between the digits. This means that the substrate layer will have the greatest potential impact on the measured capacitance, depending on various factors, due to its intimate contact with the digits. Kapton polyimide, described in the previous section as the material comprising the substrate layer, is known to have dielectric properties that are sensitive to the ambient environment. According to [7], a change in the relative humidity at room temperature (23 C) from 0 % to 100 % will increase the real relative permittivity of Kapton from 3.0 to 3.8, an increase of nearly 27 %. To determine the impact of this potential permittivity swing on the capacitance and sensitivity of the interdigital sensor, a series of simulations were performed using the model described in [5] and [6], for the geometry of Fig. 1, with a = mm, b = mm, c = mm, w = 0.1 mm, s = 0.1 mm, and N T = 6. In the model, each digit is discretized into a number of elements and a cylindrical Green s function value is calculated for each element pair. Using the method of moments, the charge density is found for each element and the total capacitance C is calculated from the relation C = Q/V, where Q is the total charge and V is the voltage between the neighboring digits. The relative permittivity of the wire insulation ε r1 was varied between 1.0 and 10 for the two realistic substrate permittivity ε r2 extrema of 3.0 and 3.8 as well as two unlikely extrema of 2.0 and 10. The results are shown in Fig. 3. Here, sensitivity is defined as S = C/ ε 1 and is in units of meters. From the figure, it is observed that an increasing substrate permittivity increases the sensitivity to dielectric changes in the wire insulation. This is due to the fact that a higher substrate permittivity concentrates, or amplifies, more of the electric flux in its volume, which also aids in

4 Capacitance (pf) ε r2 =2.0 ε =3.0 r2 ε r2 =3.8 ε =10 r Relative permittivity ε r1 FIGURE 3. Sensitivity of the interdigital sensor to dielectric changes in the wire insulation as a function of the substrate permittivity. Sensitivity is defined as the slope, C/ ε 1, and is in units of length. TABLE 2. Sensitivity S of the interdigital sensor to dielectric changes in the wire insulation and substrate layers as a function of digit separation s. The geometry, from Fig. 1, is set as a = mm, b = mm, c = mm, and w = 0.1 mm. Parameters S (mm) s = 0.1 mm, ε r2 = 2.84, ε r1 varying 28.5 s = 0.2 mm, ε r2 = 2.84, ε r1 varying 28.4 s = 0.1 mm, ε r1 = 4.015, ε r2 varying 32.0 s = 0.2 mm, ε r1 = 4.015, ε r2 varying 21.1 amplifying the flux in the insulation layer. This may be a desired response, but a substrate permittivity value that is highly dependent upon humidity is undesirable for reasons of measurement repeatability. One possible solution is to adjust the penetration depth via the spacing of the digits to couple more of the field through the insulation instead of the substrate. A series of sensitivity simulations were performed to observe how the sensitivity of the sensor to dielectric changes in each layer depends upon the spacing of the digits, which varies between either s = 0.1 mm or 0.2 mm. The results of this simulation are shown in Table 2. From the table, it is observed that increasing the spacing of the digits decreases the sensitivity of the sensor to the substrate layer while leaving the sensitivity to the insulation relatively unaltered. This is due to the extension of the fringing field, which simultaneously increases the penetration depth and decreases the capacitance. Again, this method may reduce the impact of the substrate dielectric properties on the sensitivity, but the measurable capacitance remains affected by the layer and, hence, the ambient relative humidity. Kapton is thus not recommended for future use as an electrode substrate and should instead be replaced by a hydrophobic material such as Teflon polytetrafluoroethylene (PTFE). In this paper, the Kapton effects have been minimized by using a very thin layer (25.4 µm) relative to the insulation thickness ( mm). In the experiments that follow in the next section, a pristine wire is used as a reference to compare with wires that have sustained degradation, with measurements taken at approximately the same time and in the same ambient environment.

5 Adhesive Foam Clamp Jaw Electrode/Substrate Film Adhesive Foam FIGURE 4. Cross-section view of the spring-loaded clamp jaw designed to conform the electrode/substrate film to the cylindrical surface of the insulated wire. Layers of adhesive foam permit slight variations in the wire diameter and minimize air gaps between the electrode digits and the insulation. CLAMP FIXTURE INTEGRATION AND OPERATION The interdigital electrodes and substrate described in the previous section require conformation to the outer surface of the M5086/2 wire in order to maximize the sensitivity to changes in the insulation layers. Simple adhesion of the electrode/substrate film to the flat surface of the spring-loaded applicator jaw would not grip the wire and might cause it to be pushed out by the force of the spring. Circular grooves could be milled out of the surface of each jaw to cradle the wire, but imperfections in the wire diameter, such as may come from degradation, will prevent intimate contact between the sensor and the wire surface, if the sensor is rigid. This issue was mitigated by the addition of a layer of adhesive foam between the milled groove in the jaw and the electrode/substrate film, thus allowing a small range of differing diameters for wires of the same AWG specification, as shown in Fig. 4. The electrode/substrate film was affixed to the adhesive foam with the electrode digits facing, and thus covered by, the foam so as to protect the copper digits from wear. The LCR meter contact pads and their respective traces were then partially separated from the thin bus strip connecting the two digit sides and passed through a channel that runs through one of the handles of the plastic spring-loaded clamp, as shown in Fig. 5. The contact pads were electrically connected via two thin-gauge wires to an SMA female-type output connector that protrudes externally through the back-end of the clamp handle. The channel enclosing these wires and the connector was then filled in, or potted, with a black epoxy (not shown in Fig. 5) to protect the delicate traces and connections. The potting was also used to secure the traces and wires from being moved around, which could affect the measurement data. CHEMICAL AGING OF M5086/2 WIRES A chemical aging experiment was conducted on the M5086/2 aircraft wires detailed in Table 1 to determine the effectiveness of the integrated interdigital capacitive clamp sensor at sensing changes to the wire insulation permittivity. Six wire segments, each 10 cm long, were immersed in different fluids that may be present on-board an aircraft, with a seventh pristine segment acting as the control sample. Paraffin wax was used to encapsulate the ends of the wire so that degradation in only the insulation surface was induced by immersion. For 10 days the wire segments were immersed in separate glass containers filled with aircraft cleaner, deicer 50/50, distilled water, hydraulic fluid, isopropanol 70%, and Jet A fuel. After 10 days, the wires were removed from the fluid baths, thoroughly dried using KimWipes, and placed in separate plastic storage bags. Each sample was tested by the interdigital capacitive clamp sensor less than one hour after removal from the fluid bath. Although an insulator or dielectric material, such as may be found in a component capacitor, is often considered in simulations to be a perfect insulator (i.e. conductivity σ = 0, real permittivity ε only), in reality the insulation or dielectric material has some lossiness associated with it. This lossiness is expressed by the addition of an imaginary component to the permittivity, so that ε (ω) = ε (ω) jε (ω), where ω = 2π*frequency and j = 1. The real and imaginary components are frequency-dependent due to dipole relaxation times that vary with the varying sizes of chains within a polymer and the molecules comprising liquids [8] [9]. Both components of the complex permittivity may be significantly altered by degradation from chemical reactions or absorptions of liquids, and so it is of particular

6 Groove Electrodes Signal bus Output connector FIGURE 5. Plastic spring-loaded clamp sensor with inset showing detail of the jaws and interdigital electrodes attached to both orange jaws. interest to indirectly measure these parameters using the capacitive sensor over a range of frequencies. The LCR meter calculates capacitance by applying a voltage V across its output terminal and measuring the magnitude I and phase difference ϕ of the current response. This sets up an impedance relation Z = V /I(cos ϕ + j sin ϕ ) = R + 1/ jω C, assuming inductance is negligible, from which the capacitance C is easily calculable. The imaginary component of permittivity, ε, manifests itself in the form of the resistive (real) R-component of the complex impedance, leading to ohmic losses dissipated as heat. The LCR meter displays this dissipation factor alongside the capacitance as a ratio of the two impedance components, D = Rω C, which is also equal to the ratio of two permittivity components, D = ε /ε. The integrated clamp and interdigital capacitive sensor was connected to an Agilent E4980A Precision LCR Meter by way of a 122-cm-long SMA female-to-male cable and an Agilent 16095A Probe Test Fixture with BNC-type coaxial cable. An adapter facilitated the conversion between the two cable types. A calibration procedure was performed to eliminate the contributions of the cable and probe test fixture to the measurements. The sensor was then applied to each of the seven wire segments at ten different positions along the axial length and at the same ten positions rotated by a quarter-turn. At each of the twenty inspection points, the LCR meter measured C and D at 35 logarithmically-scaled frequencies between Hz and 1 MHz and for each frequency logged a mean value of 16 sequential measurements. The mean and standard deviation of the 20 C and D data points taken from each of the seven wires are displayed in Figs. 6 and 7, respectively. From these figures, it is readily observed that several of the fluids caused significant changes to the wire insulation during the 10 day immersion while others had little to no effect at all. Both isopropanol and deicer exhibit strong absorption within the insulation, which is highly characteristic of polar fluids in Nylon 6, the outermost layer of the M5086/2 wire [10]. Cleaner and distilled water demonstrated more modest absorption and appear to have a similar spectrum pattern except at low frequencies, possibly because water is a component of the cleaner. Hydraulic fluid and Jet A fuel, on the other hand, showed no absorption whatsoever, which is also quite characteristic of Nylon immersed in aliphatic hydrocarbons [10]. In addition to these large deviations in C and D from the pristine spectrum, there are several apparent dielectric relaxations at around 60 khz and 360 khz for deicer and isopropanol, respectively. It is also important to note the large change in standard deviation from wire to wire. This is due primarily to the 20 different positions along the wire from which the sensor obtained the measurements. Thus a large standard deviation

7 Capacitance (pf) Pristine Cleaner Deicer 50/50 Distilled water Hydraulic fluid Isopropanol 70% Jet A fuel Frequency (Hz) FIGURE 6. Plot of the mean capacitance C frequency spectrum of the interdigital capacitive sensor when applied to M5086/2 aircraft cable segments immersed in the listed fluids for 10 days. The error bars represent ±1 standard deviation of 20 measurements in different positions along the 10 cm wire segment immersed in the given fluid. Dissipation factor Pristine Cleaner Deicer 50/50 Distilled water Hydraulic fluid Isopropanol 70% Jet A fuel Frequency (Hz) FIGURE 7. Plot of the mean dissipation factor D frequency spectrum of the interdigital capacitive sensor when applied to M5086/2 aircraft cable segments immersed in the listed fluids for 10 days. The error bars represent ±1 standard deviation of 20 measurements in different positions along the 10 cm wire segment immersed in the given fluid. indicates that certain positions absorbed more fluid than other positions and is not an indicator of repeated measurement uncertainty. This information may also be useful in determining the extent to which a particular wire insulator has been degraded. Comparable results were obtained in [11] on pure Nylon 6 sheets immersed in the same chemicals using parallel

8 plate electrodes. The major difference between these two experiments, however, is that the interdigital capacitive sensor is reliant upon a fringing field to penetrate through not only the Nylon 6 outer layer, but also the glass fiber and PVC inner layers. These other layers provide some contribution to the capacitive measurements and may also have inherent relaxations that could be amplified by degradation. Thus, the interdigital capacitive sensor applied to insulated wire and interrogated over a range of frequencies (preferably a bandwidth larger than the 2 MHz capability of the Agilent E4980A used in these experiments) possesses the potential to both quantitatively characterize the extent of degradation sustained by a known insulator and to determine the material composition of unknown insulators. CONCLUSION An interdigital capacitive sensor has been developed and integrated with a plastic spring-loaded clamp for on-board evaluation and characterization of aircraft wire insulation. Aircraft wire segments immersed in six different chemicals for 10 days and tested by the sensor have demonstrated its ability to quantify and its potential to identify the types of chemicals exposed to the insulation. Future work may involve the use of a network analyzer for direct impedance measurements, instead of capacitance and dissipation factor, for the express purpose of utilizing higher frequencies to determine dielectric relaxations and identify unknown insulation layers. ACKNOWLEDGMENTS This work was performed at the Center for Nondestructive Evaluation at Iowa State University with support from Boeing Research & Technology under award No and in collaboration with the Boeing NDE Group in St. Louis, Missouri. REFERENCES 1. A Stitch in Time: The Complete Guide to Electrical Insulation Testing. Dallas: Megger, C. Desai, K. Brown, M. Desmulliez and A. Sutherland, Selection of Wavelet for De-noising PD waveforms for Prognostics and Diagnostics of Aircraft Wiring, 2008 Annual Report Conference on Electrical Insulation and Dielectric Phenomena, IEEE Dielectrics and Electrical Insulation Society, Quebec, Quebec, Canada, October 26-29, 2008, pp T. Chen and N. Bowler, Analysis of a capacitive sensor for the evaluation of circular cylinders with a conductive core, Meas. Sci. Technol., Vol. 23, (10pp), A. V. Mamishev, K. Sundara-Rajan, F. Yang, Y. Du, and M. Zahn, Interdigital sensors and transducers, Proc. IEEE, Vol. 92, pp , R. T. Sheldon and N. Bowler, An interdigital capacitive sensor for quantitative characterization of wire insulation, Review of Progress in QNDE, AIP Conf. Proc., Vol. 1511, pp , R. T. Sheldon and N. Bowler, An Interdigital Capacitive Sensor for Nondestructive Evaluation of Wire Insulation, IEEE Sensors J., 2013 (submitted). 7. Summary of Properties for Kapton Polyimide Films, E. I. du Pont de Nemours & Co., Wilmington, Delaware, USA. 8. S. Havriliak and S. Negami, A Complex Plane Representation of Dielectric and Mechanical Relaxation Processes in Some Polymers, Polymer, Vol. 8, pp , K. S. Cole and R. H. Cole, Dispersion and Absorption in Dielectrics I: Alternating Current Characteristics, J. Chem. Phys., Vol. 9, pp , D. C. Wright, Failure of Polymer Products Due to Chemical Attack, Rapra Review Reports, Vol. 11, No. 10, Report 130, R. Ding and N. Bowler, Influence of Accelerated Aging on Dielectric Properties of Extruded Nylon 6 Film, 2013 Annual Report Conference on Electrical Insulation and Dielectric Phenomena, IEEE Dielectrics and Electrical Insulation Society, Shenzhen, Guangdong, China, October 20-23, CD-ROM. Pages: 4.