Patent Grant
7940056
This application claims priority to GB 0805409.0, filed on Mar. 25, 2008 and entitled “Time-Domain Reflectometry,” the disclosure of which is incorporated herein by reference in its entirety.
The invention relates to time domain reflectometry and a time domain reflectometer, and, in particular, to determining the location and characteristic of line discontinuities.
Time domain reflectometers, TDR, provide a means to determine some characteristics of faulty and normal electrical transmission lines, by sending an excitation signal and receiving a reflected response with subsequent analysis.
Power Supply—which provides the necessary power to the various circuits
Processor/Memory—which as in many examples of modern instrumentation, provides overall operational control, processing of user actions, control of information provided to the user, management for the generation of test signals, management for the acquisition of measured signals, mathematical analysis of measurements and the application of signal processing algorithms. In this context, the term “user” might also mean a separate piece of linked system control equipment as well as a human operator.
User Interface(s)—in a portable TDR these would typically be a keypad for entering commands/data and a screen for the display of measured signals responses, derived measurements and system information. In fixed installation, the user interface might consist of serial communication port such as RS232 or USB.
Test Signal Generator—provides a test signal for application to the transmission line(cable) under test.
In practice it might be in the form of a voltage source or a current source. It might also be presented in either a single-ended (unbalanced) or a balanced form.
Traditional TDRs use a substantially rectangular pulse, or pulses that are smoother in nature such as a half-sine shape or a raised cosine shape. Other TDRs use a step waveform, which does not return to zero over the duration of the measurement. More complex waveforms may be used.
Line feed resistor(s)—these provide the correct matching impedance for the line being tested. When a signal travelling along a transmission line encounters a change in characteristic, a reflection occurs. This is also true for a reflected wave returning to the TDR instrument. The instrument should therefore present an impedance characteristic sensibly close to the impedance characteristic of the line under test, if it is to avoid causing further unwanted signal reflections.
The line feed resistor(s) are therefore provided to give the correct matching characteristic for the Line under test. Multiple selections may be provided to cater for various Line types.
In practical realisations, the Test Signal Generator might consist of a voltage generator which is then arranged in series with the Line Feed Resistor(s) or a current generator arranged in parallel with the Line Feed Resistor(s). These are equivalent as per the well-known Thevenin and Norton equivalent forms.
Also the signals might be provided (and responses measured) as either a single-ended or a balanced form which are well known in measurement systems. The later analysis is presented in the single-ended (unbalanced) form although this is easily extended to the balanced form, as is well known.
Additionally, some form of dc isolation might be provided between the TDR instrument circuitry and the connectors providing the access to the Line (cable) Under Test. Typically this is done by the use of capacitors whose value is chosen to have minimal effect on the signals generated to and received from the Line (cable) Under Test. If these capacitors do have a significant effect, it can be compensated for by use of traditional analogue or digital filter techniques.
Signal measurements—this block provides the ability to capture the electrical signals appearing on the TDR Line (cable) Under Test, access point. It can therefore acquire signals with or without the cable actually connected. Typically an input amplifier of suitable impedance when considered in conjunction with the Line Feed Resistor(s) will pass the signal to an Analogue to Digital Converter (ADC) which is used to capture signal values on a point by point basis in time, which are then passed via the processor to a memory store for later evaluation.
In a practical TDR, the effect on the measured signal due to any dc isolation may again be compensated for by use of traditional analogue or digital filter techniques. Also, the input amplifier/ADC circuit may be presented in either single-ended (unbalanced) or balanced configurations.
Access point—provides the terminal connections such that the cable under test can be connected to the instrument's test and measurement circuitry.
Whenever a signal travelling along a transmission line encounters some change (discontinuity) in the line characteristic, a portion of the signal is reflected back towards the sending end of the line. The nature of the reflection signal is determined by the discontinuity characteristic, which might be anywhere between a short circuit and an open circuit.
The primary objective for a TDR is usually to estimate the physical location of a line discontinuity such that it can be repaired or replaced if the unwanted effect is severe enough. A second objective is thus to determine the nature of the discontinuity and hence provide an estimate of the degree of severity for any unwanted effects that it may produce.
In broad terms, it is accepted that a reflection of the same polarity as the applied test signal denotes a fault impedance higher than the natural impedance of the transmission line. Conversely, a reflection of opposite polarity to the test signal indicates a fault impedance that is lower than the natural impedance of the transmission line.
The amplitude of the reflection provides an indication as to the severity of the impedance mismatch caused by the fault, with small mismatches producing small reflection amplitudes and large mismatches such as short or open circuit conditions producing larger reflections. Whilst the relative size of the reflection amplitude is dependant on the fault characteristic, the absolute value is also dependant on how far away the fault is and how much signal loss is introduced by the cable on the signal as it travels to and from the fault location.
Prior art techniques include the use of simple cable loss models together with measurements of test signal and reflection signal timings and amplitude measurements to provide estimates of the fault location and the fault characteristic. Such methods will always be limited in that they only provide a rough approximation to the actual physics of the signals in the transmission line.
The limitation of these simple techniques is due to the fact that they do not accurately take into account the detailed frequency-dependant transmission loss, transmission delay and cable impedance characteristic that exists in real cables.
a shows the Line signal measured at the TDR, consisting of a test pulse and reflections for an open circuit at two different locations. As expected the high impedance fault produces a reflection signal of the same polarity as the test pulse. Also as expected, when the fault is further away, the reflection is smaller in amplitude due to additional loss due to the greater length of cable.
b shows the Line signal measured at the TDR, consisting of a test pulse and reflections for a short circuit at two different locations. As expected the low impedance fault produces a reflection signal of opposite polarity to the test pulse.
c shows the Line signal measured at the TDR, consisting of a test pulse and reflections for a 200 ohm fault at two different locations. Given that in this example the nominal cable impedance is 100 ohms, then again we see a reflection signal of the same polarity as the test pulse indicating a higher impedance fault.
d shows the Line signal measured at the TDR, consisting of a test pulse and reflections for 110 ohm fault at two different locations. Again, the nominal cable impedance is 100 ohms.
On this occasion, the simple rules for high and low impedance faults are no longer sufficient to evaluate the location and the nature of the fault.
In the first instance, the reflections are substantially opposite to the test pulse, indicating a low impedance fault. Secondly, the substantial portions of the reflections start significantly later in time than say the reflections for the 200 ohm faults seen in
The failure to use an adequate description of the processes, can lead in this case to both the nature and position of the fault to be miscalculated.
More recently, frequency-domain analysis using FFTs has been applied to certain TDR measurements, to provide enhanced analysis.
Embodiments of the present invention advantageously provide a method of time domain reflectometry including transmitting a test signal along a cable under test from one end and sensing and recording a reflected signal from the cable at that end; using the recorded, reflected signal to estimate the distance, Ldist, from the one end to a discontinuity on the cable; separating a test signal component from the remainder, Vr, of the reflected signal; estimating the impedance, Zfault, of the discontinuity from known, predetermined values of the characteristic impedance, Zline, and of the characteristic gain, T, of a reference cable, and from the said separated test signal and reflected signal components; calculating the estimation error as a difference between the model reflection signal, Vrmod, expected of the cable under test based on the characteristic impedance and characteristic gain and the estimated impedance, Zfault and distance, Ldist, and the actual reflection signal Vr; choosing new estimated values of Ldist and Zfault in accordance with a predetermined algorithm such as to approach optimal values; repeating the aforesaid step of calculating the estimation error until a predetermined condition has been satisfied; and outputting the values of Ldist and Zfault that gave the least error.
Embodiments of the present invention also provide a time domain reflectometer for testing an electrical cable, including a test signal generator connected to a pair of terminals for connection to the ends of the electrical cable under test; and a signal processor connected to the terminals to receive a reflected signal from a test signal transmitted into the cable under test; the signal processor being programmed to sense and record a reflected signal from the cable; transmitting a test signal along a cable under test from one end and sensing and recording a reflected signal from the cable at that end; using the recorded, reflected signal to estimate the distance, Ldist, from the one end to a discontinuity on the cable; separating a test signal component from the remainder, Vr, of the reflected signal; estimating the impedance, Zfault, of the discontinuity from known, predetermined values of the characteristic impedance, Zline, and of the characteristic gain, T, of a reference cable, and from the said separated test signal and reflected signal components; calculating the estimation error as a difference between the model reflection signal, Vrmod, expected of the cable under test based on the characteristic impedance and characteristic gain and the estimated impedance, Zfault and distance, Ldist, and the actual reflection signal Vr; choosing new estimated values of Ldist and Zfault in accordance with a predetermined algorithm such as to approach optimal values; repeating the aforesaid step of calculating the estimation error until a predetermined condition has been satisfied; and outputting the values of Ldist and Zfault that gave the least error.
In order that the invention may be better understood, a preferred embodiment will now be described, by way of example only, with reference to the accompanying drawings, in which:
a to 3d, referred to above, show typical signals measured from the cable attached to the TDR of
a, and
a, 6b and 6c represent sample signals obtained using the process of
a, 7b, 7c and 7d show further signals obtained using the process of
a and 10b show sample signals of reflection signals using the process of
a and 11b show sample plots of real and imaginary parts of a model characteristic line impedance;
a shows the magnitudes of an actual reflection signal from a cable under test together with a calculated reflection signal from an as yet unoptimised model, in accordance with an embodiment of the invention.
b shows the error representing the difference between model and actual values in
a and 14b show, respectively, for an untuned model, the phase of actual and predicted values of the reflected signal, and of the phase error between these signals;
a shows sample magnitudes of reflection signals, after tuning of the values for Ldist and Zfault in accordance with the preferred embodiment;
b shows sample phase signals corresponding to
c shows actual and model time domain signals after tuning of the values of Ldist and Zfault;
a shows one example of the process embodying the invention for optimising, i e tuning Ldist and Zfault; and
b shows an alternative example to that of
Glossary of Terms
The preferred embodiment will now be described.
A first step is to create a detailed physics reference model of the transmission line (cable) under test in respect of its effect on a signal applied at one end as it moves along the length of the cable a specified distance. This requires knowledge of the cable loss characteristic.
A second step is to create a detailed physics reference model of the effect of a general fault condition, which produces a reflection, when a signal is applied to it. This requires knowledge of the cable impedance characteristic, Zline.
The loss characteristic for a transmission line can be represented in a number of different ways. If T(jw) is the traditional frequency gain response then the transmission line loss characteristic is:
1/T(jw)=transmission line input(jw)/transmission line output(jw)
The rectangular form is particularly useful in that the Real and Imaginary description of the Transmission Line Loss characteristic can be multiplied by the FFT of a desired input signal to produce the resultant transmission line output signal by means of the well-known frequency domain convolution. This output signal may then be converted to a time domain signal by means of the inverse FFT, if desired.
A further traditional representation of the transmission line loss characteristic from the fundamental theory of transmission lines is:
propagation characteristic γ=α+jβ=√((R+Ls)(G+Cs)) (1)
It is noted that the loss α and the delay β are for a unit length of line and that the loss and delay for any length Ldist is thus α(Ldist)=αLdist and β(Ldist)=βLdist.
It is also noted that the Neper is a logarithmic measure: 1 Neper=8.686 dB.
Although not essential, it is highly desirable that the TDR can itself gather the reference data for T(jw) and later for Zline(jw), using its normal test signals, normal measurement capabilities and suitable signal analysis.
A user can thus gather the required reference data on a reference sample of cable and then make subsequent measurements on potentially faulty cables, all with the same instrument.
One reference, i.e., Application note AN-4, May 1989 from www.picosecond.com, Picosecond Pulse Labs, “TDR, Step Response and “S” Parameter Measurements in the Time Domain”, James R. Andrews, describes suitable methods derived from academic papers presented in the mid 1970s for the measurement of the loss data using pulse signals and FFT analysis.
Another reference, i.e., 1969 IEEE International Solid-State Circuits Conference, “Wide Band Impedance Measurement by Fourier Transformation of Network Pulse Response” by A. S. Farber and C. W. Ho, pages 170, 171 and 185, describes suitable methods for the measurement of impedance data using pulse signals and FFT analysis.
Obtaining the Reference Data by TDR Measurement and Calculations
When a reflection signal is used to measure cable loss for a cable of length Ldist terminated in an open circuit, then that signal has in fact travelled a distance of 2×Ldist from its point of origin.
The complex reference data for the loss in a cable of length Lref, 1/T2×Lref(jw) is obtained by:
Selecting a piece of reference cable with its far end in an open circuit condition.
The reference cable should be of sufficient length=Lref that a reflected signal can be seen in isolation from the applied test signal i.e. any effects at the input end should sensibly have reduced to zero (or can be made so by suitable filtering) prior to the reflection signal appearing.
Using the TDR to apply a test signal to a reference length of cable.
Using the TDR to measure the resultant signal Vp(t), which may have needed some transient filtering to ensure that Vt(t) and Vr(t) are separated by a period of zero signal.
Separating the test signal response Vt(t) from the Vr(t) reflection response.
Performing an FFT operation on the test signal response and the reflection response.
Taking the FFT(Vr(t))=Vr(jw).
Taking the FFT(Vt(t)=Vt(jw).
Dividing Vr(jw) by Vt(jw) to get the transmission gain characteristic T2×Lref(jw) for 2×Lref (signal travels the length of Lref and back again.) for the reference length Lref.
Cable loss for 2×Lref=1/T2×Lref(jw).
A nominal value for the velocity of propagation (VOP) of the reference cable may now be obtained by calculating 2*Lref/(time of first discernible response of Vr(t)−time of first discernible response of Vt(t)).
A “region of validity” should be determined for the use of T2×Lref(jw). This may be determined by looking at the values for Vr(jw). The reflection response is typically stronger at lower frequencies and diminishes to a negligible level as the frequency increases. If the TDR circuitry includes dc isolation, then very low frequency amplitudes will also diminish to negligible levels.
The information in T2×Lref(jw) should only be used for frequencies where |Vr(jw)| is significant and well-behaved i.e. free from sharp variations. An example of a suitable region of validity might be—From an initial frequency point whereby for increasing frequency, |Vr(jw)| rises to 75% of its maximum value for the last time up to a final frequency point whereby, |Vr(jw)| first falls to 10% of its maximum value.
The arrangement for this is shown in
Sample signals for Vp(t), Vt(t) & Vr(t) are shown in
In this example for an Lref of 551 m, the nominal value of VOP may be calculated from
VOP=2×551/(5.3 us−0.15 us)=213.98×10^6 m/second or 71.38% of the speed of light.
Sample signals for |Vt(jw)|, |Vr(jw)| and the detail of |Vr(jw)| are shown in
d shows the calculated transmission gain in logarithmic form, i.e. 20 Log10(|T2×Lref(jw)|).
A flow chart of the general process is provided in
The principles of this technique for TDR estimation of the cable transmission characteristic are based on the principles and references found in Application note AN-4 1989 referred to above, and incorporated herein by reference.
The cable impedance characteristic Zline(jw)=√((R+jwL)/(G+jwC)) describes the input impedance.
In general, Zline(jw) is complex and frequency dependent.
A measurement of Zline(jw) is required over a suitable frequency range, to enable the realisation of the fault characteristic model.
The complex reference data for Zline(jw) is obtained by:
A “region of validity” should be determined for the use of Zline (jw). This may be determined by looking at the values for Vr1(jw) and Vr2(jw). The reflection response is typically stronger at lower frequencies and diminishes to a negligible level as the frequency increases. If the TDR circuitry includes dc isolation, then very low frequency amplitudes will also diminish to negligible levels.
The information in Zline(jw) should only be used for frequencies where |Vr1(jw)| and |Vr2(jw)| are significant and well-behaved i.e. free from sharp variations. An example of a suitable region of validity might be: from an initial frequency point whereby for increasing frequency, |Vr1(jw)| rises to 75% of its maximum value for the last time up to a final frequency point whereby |Vr1(jw)| first falls to 10% of its maximum value. Similarly a suitable region of validity may be chosen for Vr2(jw), and then only that range where both responses are valid can be chosen.
The arrangement for this is shown in
Sample signals for |Vr1(jw)| and |Vr2(jw)| are shown in
Sample plots for Re Zline(jw) and Im Zline(jw) are shown in
A flow chart of the general process is shown in
The principles of this technique for TDR estimation of the cable impedance characteristic are based on the principles and references found in the above-referenced IEEE 1969 International Solid-State Circuits Conference paper “Wide Band Impedance Measurement by Fourier Transformation of Network Pulse response” A. S. Faber and C. W. Ho, which is incorporated herein by reference.
Constructing the Variable Length Transmission Line and Fault Effect Model
Having obtained the two sets of complex reference data from the reference cable, the model of the variable length, variable fault transmission line is constructed as follows:
The per unit length gain characteristic is found by:
Consider now the effect of a fault impedance Zfault.
The well-known complex reflection equation provides the description of the effect of a fault impedance Zfault on a transmission line with impedance Zline(jw) as:
ρ(jw)=(Zfault(jw)−Zline(jw))/(Zfault(jw)+Zline(jw))
In the usual signal processing notation, the complete variable-length, variable-fault transmission line model and signals in the frequency domain are:
Model Reflection signal=Vrmod(jw) and
Vrmod(jw)=Vt(jw)×T2×Ldist(jw)×(Zfault(jw)−Zline(jw))/(Zfault(jw)+Zline(jw)).
Use of the Model
The model is used as follows.
General
A measurement is taken of the actual signal applied to the transmission line(cable) under test. This signal can be represented either in the time-domain Vt(t) or alternatively in the frequency domainVt(jw), by use of the FFT operation.
The complex frequency domain model characteristic T2×LdSt(jw)×(Zfault(jw)−Zline(jw))/(Zfault(jw)+Zline(jw)) could also be represented in the time domain for use in a time-domain convolution operation, but preferably it is represented here as a frequency domain characteristic so that the required convolution with the applied signal may be done by straightforward complex multiplication as is well known.
The model reflection response in the frequency domain is obtained as:
Model reflection=Vrmod(jw)=Vt(jw)×T2×Ldist(jw)×ρ(jw), (2)
where all functions of jw are complex.
The principle for the determination of the location and nature of a transmission line fault can now be described as:
A “region of validity” should be determined for the use of the error calculations below. This may be determined by looking at the values for Vr(jw). The reflection response is typically stronger at lower frequencies and diminishes to a negligible level as the frequency increases. If the TDR circuitry includes dc isolation, then very low frequency amplitudes will also diminish to negligible levels.
The information in error values below should only be used for frequencies where |Vr(jw)| is significant and well-behaved i.e. free from sharp variations. An example of a suitable region of validity might be: from an initial frequency point whereby for increasing frequency, |Vr(jw)| rises to 75% of its maximum value for the last time up to a final frequency point whereby |Vr(jw)| first falls to 10% of its maximum value.
a) Produce a Model reflection=Vrmod(jw) in the frequency domain using equation 2. Calculate:
magnitude error=|Vr(jw)|−|Vrmod(jw)|, and
phase error=∠Vr(jw)−∠Vrmod(jw)
between the actual and the model reflections, using a frequency range within the regions of validity established for the reference cable transmission data, the reference cable impedance data and the fault reflection data.
Choose new trial values of Ldist and Zfault(jw) based either on a predetermined sequence or by an assessment of previous error values.
Return to step a) until the predetermined sequence is completed or error values reach a suitable value.
Choose the values of Ldist and Zfault(jw) that give the minimum error between the actual response and the model response, as the best estimates of the actual fault location and fault characteristic.
One arrangement for this is shown in
Sample signals for |Vr(jw)| and |Vrmod(jw)| are shown in
b shows a magnitude of error, |error| plot for the signals in
Sample signals for ∠Vr(jw) and ∠Vrmod(jw) are shown in
b shows a |error| plot for the signals in
The responses of
Sample signals for |Vr(jw)| and |Vrmod(jw)| are shown in
Sample signals for ∠Vr(jw) and ∠Vrmod(jw) are shown in
Sample time domain signals for Vr(t) and Vrmod(t) are shown in
This results in this example in the final estimate for Ldist as 550.93 m compared to the actual distance of 551 m. The final estimate for Zfault was 99.7 ohms compared to the actual fault of 99.6 ohms.
One flow chart of the general process is provided in
The block in
A variety of known search techniques can be applied to the task of efficiently tuning parameters to achieve the minimisation of a cost function.
The arrangement shown in
The many features and advantages of the invention are apparent from the detailed specification, and, thus, it is intended by the appended claims to cover all such features and advantages of the invention which fall within the true spirit and scope of the invention. Further, since numerous modifications and variations will readily occur to those skilled in the art, it is not desired to limit the invention to the exact construction and operation illustrated and described, and, accordingly, all suitable modifications and equivalents may be resorted to that fall within the scope of the invention.
| Number | Date | Country | Kind |
|---|---|---|---|
| 0805409.0 | Mar 2008 | GB | national |
| Number | Name | Date | Kind |
|---|---|---|---|
| 5481195 | Meyer | Jan 1996 | A |
| 5524281 | Bradley et al. | Jun 1996 | A |
| 7245129 | Wajcer et al. | Jul 2007 | B2 |
| 7429866 | Beene | Sep 2008 | B2 |
| 20040015311 | Furse et al. | Jan 2004 | A1 |
| 20060181283 | Wajcer et al. | Aug 2006 | A1 |
| Number | Date | Country | |
|---|---|---|---|
| 20100188095 A1 | Jul 2010 | US |
