Brent Waddoups, Dr. Cynthia Furse, Mark Schmidt

Department of Electrical and Computer Engineering
Utah State University
Logan, Utah 84322-4120
Phone: (435) 797-2870
FAX: (435) 797-3054
Furse@ece.usu.edu

Abstract

Chafed insulation has been identified as a major cause of failure of aging aircraft wiring.  Insulation failures can be caused by aging and brittleness, maintenance damage, vibration, etc.  Chafes can range from very small breaks such as hair-line cracks or pin-holes in the insulation to large frays several centimeters long.  While there are several methods that can be used to detect chafed insulation during extended maintenance, these methods do not lend themselves to routine pre- or post-flight testing of the total wiring system.   The objective of this project is to develop and evaluate a system that can be imbedded directly into aircraft wiring for routine pre- and post-flight detection of breaks in the insulation.  This is part of the Smart Wiring System project sponsored by the Office of Naval Research (ONR) and the Naval Air Systems Command (NAVAIR).  Both legacy and new aircraft are considered, with emphasis on total cost savings and prevention of fires and catastrophic events. 

This paper describes new and promising results for detection of chafed insulation using reflectometry.   Time Domain Reflectometry (TDR) is used to evaluate the full electrical nature of a wide variety of cable frays including several common types of aircraft wiring, different sizes and types of frays, and different environmental conditions surrounding the fray.  These results are used to develop simulation models of frayed cables, to give some understanding of the thresholds for detection of frays in various electrical configurations.  Finally, the frequency domain components of the TDR signals are evaluated to determine an optimal frequency configuration for the small, low-cost frequency domain reflectometry (FDR) system developed at Utah State University for the Smart Wiring System.

I.    Introduction

Aging aircraft wiring has been identified as an area of great national and international concern.  Typical wiring problems include broken wires (open circuits), short circuits between two wires or between wires and the metal aircraft structure, compromised electromagnetic shields, improper grounding, broken, bent, or corroded connector pins, cold or broken solder joints, incorrect reassembly during maintenance, chafed (frayed) insulation, and more. 

There are numerous systems available today or under development for detecting various wiring problems.  One major class of these techniques is electromagnetic reflectometry which includes time domain reflectometry (TDR), frequency domain reflectometry (FDR), standing wave reflectometry (SWR), and related methods.  These techniques generally place a low power signal (either a fast-rise time pulse or voltage step or a series of sine waves stepped in frequency) on the cable.  This electromagnetic wave propagates down the cable and is reflected back to the source from any “change” in the cable.  Large reflections occur at open or short circuits, making them the easiest anomalies to detect.  Smaller reflections occur from junctions where the cable branches into a “tree” formation, and these junctions can also be relatively easily detected.  Even smaller reflections occur at junctions such as connectors and solder joints, breaks in the electromagnetic shield of shielded wires, and regions where the cable insulation is chafed.  The smallest of all reflections are where there are very small cracks in the insulation.  

This paper will analyze the expected performance of reflectometry systems based for detection of chafed insulation.  Section II will describe the USU Smart Wire system and the reasons why reflectometry is being pursued for detection of chafed insulation.  Section III will describe the basic operation of the two major classes of reflectometry systems – time and frequency domain reflectometry systems.  Section IV will outline the methods that were used, and Section V will analyze the response of reflectometry systems to wires that have chafed insulation.  Section VI will summarize these results and draw conclusions for developing a reflectometry-based system for detection of chafed insulation.

II.    The USU Smart Wire System

The Utah State University (USU) Center of Excellence for Smart Sensors is teamed with Management Sciences, Inc. and  NAVAIR, USAF, United Airlines, and the Federal Laboratories Consortium to develop a new type of "Smart Wiring."  This in situ system will automatically self-test the wiring prior to take off and during flight as needed, detect the nature and location of a fault, provide warnings and system shutdowns as appropriate, maintain a log of wire health as it degrades over time, and give preventative maintenance recommendations.  All other flight-critical systems on board an aircraft have had this type of self-testing capability for years.    Such a system for wiring is long overdue.  This new "Smart Wiring" system is, to our knowledge, the only system presently being developed for routine in situ wire testing.  Patents are pending.  A frequency domain reflectometry (FDR) sensing modality for finding faults in aging aircraft wiring has been successfully demonstrated [7-10], and the full system is being developed for "iron bird" testing.  This system will be able to locate electrical faults on wires and is expected to dramatically improve wire inspections while saving millions of dollars in maintenance expenses. 

III.    Reflectometry Systems for Wire System Diagnosis


Reflectometry systems can be divided into two broad classes – time and frequency domain reflectometry.  Both of these systems traditionally require direct connection to one end of a wire under test, and place a low power voltage signal directly on the wire itself.  They are based on detection of the reflected signal from the end of the wire.  Although these methods have a reputation for complex signals that “take a PhD to analyze them,” significant advances have been made in the past few years that enable automatic detection of faults with both TDR and FDR methods.

Time domain reflectometry (TDR) launches a short rectangular pulse (shaped pulses can also be used) or step function voltage down the cable. The wave travels to the end of the cable, where it is reflected back, and circuitry at the sourcing end of the cable is used to receive this reflected voltage.   The cable impedance, termination, and length give a unique temporal signature indicative of the cable's health.  In essence, the TDR tells the length of the cable based on the delay that it takes for the reflection to return to the source.  Large changes in the wire (open or short circuits) cause large reflections that are easy to measure, and small changes in the wire (junctions, frays, etc.) cause smaller reflections that are more difficult to detect.  The phase and shape of the reflected signal can be used to determine the termination on the cable or other anomalies along its length.  TDR circuitry includes a fast-rise time pulse generator and fast voltage samplers. 

Frequency domain reflectometry (FDR) sends a set of stepped-frequency sine waves down the wire.  These waves travel to the end of the cable and are reflected back to the source.  Electronics at the source end are used to sense these reflected waves (or the combination of the incident and reflected waves that produces a standing wave).  Frequency Modulated Continuous Wave (FMCW) systems use a set of frequencies that are ramped up in time.  By measuring the difference between the frequency of the  reflected wave and the new frequency of the incident wave, the elapsed time and hence the length of the cable can be determined.  Standing Wave Reflectometry (SWR) systems measure the magnitude of the standing wave (combination of incident and reflected waves) at the source (or other location) as a function of frequency and determine the length and termination of the cable from that measurement.  The USU Frequency Domain Reflectometry (USUFDR) system uses a frequency multiplier (mixer) to determine the phase change between the incident and reflected wave.  The signature of this phase change as a function of frequency is used to determine the length and termination of the cable or anomalies along its length.  Confusion is common in the names of frequency domain reflectometry methods, so for our purposes they will be referred to as “FMCW”, “SWR” and “USUFDR” methods.  Be aware that other authors may use different names for the various FDR systems.  FDR circuitry includes a stepped (or variable) frequency sine wave generator and either a frequency counter (FMCW), a received signal strength indicator (RSSI) chip or similar method of measuring high or intermediate frequency voltage magnitudes (SWR), OR a frequency multiplier (mixer) and DC voltage measurement (USUFDR) hardware. 

TDR and FDR methods are strongly related.  In theory, TDR provides information on the reflected wave on the cable for “infinite” bandwidths (in practice, very large, but limited by the bandwidth of the pulse and sampling circuitry).  The FDR methods provide virtually the same information over a selected set of frequencies (usually a much smaller bandwidth than TDR).  FDR information can, therefore, also be obtained by using a TDR to measure the reflected wave over the large bandwidth and using signal processing (such as fourier transform) methods to convert from time to frequency domains.  Thus, the TDR is an excellent tool for evaluating the expected performance of FDR methods over “all” frequency bands.  This is what has been done in the present study.

IV.    Methods
   
The objective of this study is to determine the characteristics of reflectometry systems that could enable detection of chafed insulation.  This was done by using a commercial TDR system for collection of experimental data on a wide variety of cables with chafed insulation.  Conversion to the frequency domain for analysis of the response of FDR systems is done by taking the fourier transform of the experimental TDR waveforms for the various chafed wire configurations.  The level of response that is seen can be used to interpret what would be seen by FMCW, USUFDR, SWR, or similar systems to these chafed wires.

A time domain reflectometer model TDR100 was obtained from Campbell Scientific for use in the cable fray detection research.  The TDR100 transmits a 14 microsecond long pulse at about 50 kHz along the wire and has a sampling circuit with a resolution of 12.2 picoseconds [11].   The unit requires a 12V supply voltage and interfaces to a personal computer using a Campbell Scientific proprietary software called PCTDR.   
The TDR100 has a BNC output jack for connection to cables.  This BNC connector has an outer conductor and inner conductor, and therefore two connections are required.  In some cases, we have connected to two parallel wires.  In other cases, we have connected to a single wire, with the outer conductor of the BNC cable connected to a shield or a nearby ground.  In order to facilitate these connectors, a BNC-to-banana jack was used.  As will be seen later, this jack created additional reflections that were removed via software.

V.    Results

a.    Single Strand Polyimide-Insulated Wire

The first type of wire that was tested was a single wire strand with a polyimide insulation.  The wire was connected via the active signal side of the banana clip to the TDR unit, and the ground of the banana clip was connected to a nearby ground on the computer. 

The first step in gathering the time domain signatures of the single wire with no known flaws or imperfections.  This is to be used as the baseline for comparison with wires that are chafed.  Then a 1 cm fray was made on the wire by removing the insulation with a knife.  The insulation was only removed from one side of the wire in order to simulate a fray that may be caused by the wire rubbing or vibrating against some surface.  The first test of the frayed wire was to simply let the fray be in the surrounding air and not in contact with any other materials.  The results of this test are shown in Figure 1. 


Figure 1– Baseline response and response when a 1cm fray is introduced and is in open air.  Normalized voltage is on the left, and wire length is on the bottom axis.
The difference between the baseline response and the frayed response is minimal in this case.  It would be difficult to determine the location and severity of a wire fray when the chafe is left in open air. 

    The chafe was then held between the fingers of a person or held against the metal of an airframe.  These two results are very similar and are shown in Figure 2.


   




Figure 2- Baseline response and response when a 1cm fray is held in fingers.
    The clearness of the location of the fray is due to at least two factors.  The first of these factors is the difference in the dielectric properties of the insulation and human flesh.  The polyimide insulation used on the wire is, obviously, not a very good conductor.  However, human flesh with its large amount of water, is a very good electrical conductor.  Also, along with being a very good conductor, the human tissue is not matched to the impedance of the conductor of the wire.  Therefore, when the signal on the wire encountered the finger touching the fray, a portion of the signal was conducted into the finger and the mismatch between the finger and the wire caused a reflection that can be seen in the TDR response.  Similar results are seen when the fray is held against a metal object.

    Next, tap water was dripped onto the fray, and a TDR response was obtained.  As may be seen in Figure 3, the difference in response when the tap water was dripped on the fray and the baseline with no fray was minimal, only slightly better than the fray in air.


Figure 3 - Baseline response and response when a 1cm fray has tap water dripped on it.

The response of the fray when a 10 percent saline solution is applied to the fray is shown in Figure 4.  This is a situation that could commonly be seen in some regions of Navy planes, particularly the SWAMP areas.  The significance of these results are that we can expect to see a significant improvement in the reflectometry responses of cables immediately after they have completed a mission over water than before.

     


Figure 4- Baseline response and response when a 1cm fray has a 10% saline solution dripped on it.


    The next step in our analysis is to evaluate the response of each of these frayed wire scenerios in the frequency domain.  This is done by taking the fourier transform of the time domain data that is shown above.  Figure 5 shows the response of the cable fray in air.  Figure 6 shows the response when touched by a finger or metal object.  Figure 7 shows the response with tap water dripped on the fray, and Figure 8 shows the response with salt water dripped on the fray.
 



Figure 5– Baseline response and response when a 1cm fray is introduced and is in open air.


Figure 6- Baseline response and response when a 1cm fray is held in fingers.


Figure 7 - Baseline response and response         Figure 8- Baseline response and
when a 1cm fray has tap water dripped on it.    response when a 1cm fray has a 10% saline solution dripped on it.

 
One very significant result that can be observed from these figures is that the frequency range from 200-400 MHz is most sensitive to the fray.  From our additional work, it appears that this is independent of cable type and length. 

b.    Shielded Twisted-Pair Polyimide Wire

Next, a shielded twisted pair wire polyimide-insulated wire was analyzed.  In this case, the chafe was substantial.  The metal shield was penetrated, and the insulation of the wires themselves, although they were not touching in any way.  This type of fray would represent a major cable fault, but it is not outside the bounds of frays that have been encountered during maintenance and inspection of active aircraft wiring harnesses.

A baseline for the shielded twisted pair wire prior to chafing is shown in Figure 2. This is typical of the graph one would expect for a TDR signal from an open-circuited wire.  The first “peak” in the graph is the response from the discontinuity of the banana clip, and the later rise is the result of the open circuit at the end.  The two wires in the twisted pair were connected to two banana plug jacks, and the shield was left to float or was grounded nearby (with little change in the results). 

Figure 2 – Baseline response of the shielded twisted pair wires.

The response of the fray in open air is visually indistinguishable from the baseline signal.  The response is so small that additional signal processing is or will be necessary in order to distinguish the signal, and it is not clear at this time that the response could be distinguished from the noise and measurement error inherent in the system.

There are some conditions, however, when the fray became very apparent.  One of these was when it was partially grounded by touching the frayed area to nearby exposed metal of the aircraft body (typical of chafes created by rubbing against a metal part of the aircraft), as shown in Figure 3. 

 
Figure 3 – Plot of the magnitude of the reflection coefficient of the baseline twisted pair and the twisted pair when a fray is in contact with exposed metal
Another time when the fray became apparent was when tap water was dripped or misted on the wires and is shown in Figure 4.  In a test scenario, this could be accomplished by brushing the wire with water prior to testing.  
 
Figure 4 – Plot of the magnitude of the reflection coefficient of the baseline twisted pair and the twisted pair when a fray is exposed to tap water.

Another possible testing configuration would be when the wire has been misted with salt water.  Although it is unlikely that a maintainer would deliberately mist wires with salt water, many areas of the planes (particularly for the Navy) are exposed to salt water spray during their normal operation.  The response of the fray to a 10 percent salt water solution is shown in Figure 5, and the fray is clearly evident.
Figure 5 – Plot of the magnitude of the reflection coefficient of the baseline twisted pair and the twisted pair when a fray is exposed to salt water.
   
From these results, it is clearly evident that the response of a very significant cable fray on a twisted pair polyimide wire can be observed when it is in contact with exposed metal, or when it is dampended with either tap or salt water.  These conditions could occur during normal operation or can be induced prior to testing.  Being able to detect chafed wiring that is just exposed to air is not clearly possible from these experiments.  It is possible that advanced signal processing techniques can improve the detectable range, although existing noise in the system and measurement error makes this difficult and potentially impossible.

VI.    Conclusions

Time and Frequency Domain Reflectometry both show promise for detection of chafed insulation.  Some frequencies are more sensitive than others.  The frequency spectrum between 200 and 400 MHz is found from these preliminary results to give responses that are up to 5 times higher than the responses at 1 GHz, for instance.  This demonstrates that tuning the frequency of the reflectometer to the chafe can give optimal responses.

For more in-depth technical information, please contact us: