GB2065312A

GB2065312A - Location of Cross-talk Faults by Correlation

Info

Publication number: GB2065312A
Application number: GB8035385A
Authority: GB
Original assignee: Post Office
Current assignee: Post Office
Priority date: 1979-11-16
Filing date: 1980-11-04
Publication date: 1981-06-24
Also published as: GB2065312B

Abstract

Apparatus for locating cross-talk faults in telecommunication cables comprises applying a step function signal to one cable pair and detecting the cross-talk signal in a nearby cable pair, the detecting procedure involving the use of an analogue to digital converter (6) controlled to sample the received waveform at successively different time intervals from the pulse transmission. A microprocessor (7) stores values of waveforms that are calculated based on those that are expected from various fault location distances. Samples of the received and calculated waveforms are correlated to determine the location of the fault. <IMAGE>

Description

SPECIFICATION Location of Cross-talk Faults by Correlation This invention relates to apparatus for locating cross-talk faults in cables.

In telecommunication cables, and particularly in those conducting PCM signals, faults can give rise to cross-talk between channels. Time and effort can be saved if the fault can be located, say to within 10 metres, from one end of the cable. It is to this end that the present invention is directed.

According to the present invention there is provided apparatus for detecting the location of those faults in a signal cable that give rise to cross-talk in a nearby cable, the apparatus comprising.

means for applying a signal of predetermined shape to one of the cables, means for receiving, at a position on the other cable, the resultant cross-talk signal; means for sampling the received signal at predetermined intervals and means for electronically correlating the received waveforms with comparison waveforms represented in a local store in the apparatus by comparing each sample of the received signal with a corresponding sample in each comparison waveform, and wherein the comparison waveforms are related to the transmitted waveform by being the result of, or a domain transform of the result of, the operation on the transmitted waveform or its frequency transform, of a frequency-dependent transmission function.

Embodiments of the invention will now be described, by way of example only, with reference to the accompanying drawing which is a block schematic diagram of apparatus for detecting the location of cross-talk faults in repeater sections of telecommunications cable.

In a PCM telecommunications link the location of that type of fault in a cable pair of a repeater section which gives rise to cross-talk in an adjacent cable pair is achieved by connecting a signal source to one of the cable pairs and a receiver/measuring unit at a position on the other cable pair. The receiver and its associated processor records the cross-talk waveform and compares it with waveforms derived from those that would be expected from faults located at various positions along the cables.

A series of expected waveforms may be calculated, each specified as a function of time xL(t) and corresponding to a position of cross-talk coupling a distance L from the receiver. For convenience the signal source and the receiver are assumed to be located at the same point connected to different cable pairs. The cable has a frequency dependent transfer function and therefore the frequency components of the transmitted signal should be considered.

If P(w) is the frequency spectrum of the transmitted pulse, C(o) is the transfer function of the cable per unit length, and k is a constant proportional to the magnitude of the cross-talk then the frequency spectrum X(w) of the received cross-talk waveform originating from a distance L is given by: XL(W)=P(W). C()2L k jO (1) The corresponding time domain waveform xL(t) of the expected waveform is found using the inverse Fourier Transform of spectrum XL(4)).

To compare the received and expected waveforms a correlation integral AL is calculated;

where r(t) is the received waveform and y(t) is the inverse Fourier Transform of l/XL(a)).

Thus, if a number of samples of the received waveform are taken at intervals after the pulse transmission and these are multipiied by corresponding samples of yL(t) derived from an expected waveform from a fault at length L then a correlation co-efficient for a fault at that distance is calculated and if this procedure is repeated for other expected waveforms for different values of L the correlation co-efficients may be plotted against L to indicate the likely location of the fault.

Referring now to the Figure, a signal generator 1 is connected via a buffer amplifier 2 to a pair of cables. The signal generator 1 produces a unit step signal at intervals determined by a timing arrangement. The timing arrangement comprises a clock generator 3 providing a 1 6 MHz clocking signal, connected to feed a frequency divider 4. The frequency divider divides the clocking frequency by 512 and produces an output to time the signal generator 1.

The signal receiving part of the apparatus is connected to a second cable pair that is connected to the first by a cross-talk fault. The receiving part comprises a variable gain amplifier 5 connected to receive the cross-talk signal and connected at its output to an analog to digital converter 6. The converter 6 is operated under the control of the clock generator 3 via the count output from the frequency divider 4 through a comparator 8. The converter 6 thus samples the received waveform once for each step pulse transmitted, the samples being at 60 nano second intervals, and converts the samples to digital form. The output of the converter 6 is connected to a microprocessor 7.

The microprocessor 7 stores information on the waveforms that are expected to be returned by a cross-talk fault at various locations along the cable upon the application of a unit step signal. This information is stored in a read-only memory and is sufficient to derive values of yL(t) of the expected waveform at those intervals corresponding to the samples taken of the received waveform.

Alternatively the microprocessor may calculate the expected waveforms as required, but this may add a considerable amount to the total calculation time.

Each sampled value of the received waveform transmitted to the microprocessor from the converter 6 is multiplied by a corresponding sample of the calculated yL(t) waveforms and the resultant products are added to give a co-efficient of correlation approximately proportional to the correlation integral described above.

To investigate a complete cable length with an accuracy of about 10 metres then 1 90 calculated waveforms need to be compared with the received waveform. Since correlations requires the calculations of around 500 products, the number of calculations required is 85,000 which may be accomplished by a suitable microprocessor in a few seconds.

Some measure of signal averaging may be required to extract the signals from the furthest positions along the cable since these may be attenuated by 70 dB or more.

In the above, a unit step function is used as the applied signal shape but other functions may be found preferable. It may however be more difficult to calculate the stored comparison waveforms for other applied signal shapes. The following is an alternative method of generating comparison waveforms which is more general than that described above.

To calculate the received cross-talk waveform as a result of the passage of a pulse along a total cable length x, a transmission function is assumed for sinusoidal inputs V.exp(jwt), I.exp(jwt) of: V=Aexp (-p;:)+Bexp (+px) (3)

where =angular frequency ZO=characteristic impedance A, B=arbitrary constants and the high frequency skin effect has been included into the propagation constant p which can be expressed: p-kw112+j(kw12+Nw) (5) i.e. p is complex of the form P=+ip =attenuation term p=phase term The attenuation term is kw112 which is the characteristic assumed in subsequent calculations. N is the reciprocal of the Propagation Velocity so that the term Nx represents only the time delay for the pulse to appear at the receiving end. k is the attenuation constant. Typical values for a 24 pair PCQT 0.63 mm Copper Cable are: N1=Vp=2 17 x 1 08my~ k=4.9 x 1 0-7np m-'s""2 Using Laplase Transform methods (e.g. as described in Bylanski, P., and Ingram, D. G. W.,: 'Digital Transmission Systems'.Peter Peregrinus Ltd., London, 1 976. pp 133-143.) it can be shown that the impulse response of such a line is the inverse of:

where s=Laplase variable V=impuise content which inverts to:

From the impulse response, the response g(t) to any other input f(t) can be calculated by convolution: g(t)=f(t)*h(t) (8) *denotes convolution h(t) is the impulse response.

The method can be applied to an arbitrary input pulse shape. The pulse is broken down into n terms of the form f,(t) where i ranges from 1 to n so,

T1 is the delay intercept t is the gradient.

n can be chosen to be as large as necessary for the specified accuracy. Each section f,(t) is convolved with the impulse response h(t) from equation (7) to give n terms of the form:

where erfc (x) is the complementary error function. From the total of n terms evaluated in this way, the full solution can be built up. From the n time divisions used, the solution after k divisions (k < n) will be:

Where b, is odd or even depending on whether the gradient of the section is negative or positive. The technique enables approximate but analytic solutions to problerns to be made which otherwise would have had to be computed numerically.

Claims

1. Apparatus for detecting the location of those faults in a signal cable that give rise to crosstalk in a nearby cable, the apparatus comprising:- means for applying a signal of predetermined shape to one of the cables, means for receiving, at a position on the other cable, the resultant crosstalk signal;; means for sampling the received signal at predetermined intervals and means for electronically correlating the received waveforms with comparison waveforms represented in a local store in the apparatus by comparing each sample of the received signal with a corresponding sample in each comparison waveform, and wherein the comparison waveforms are related to the transmitted waveform by being the result of, or a domain transform of the result of, the operation on the transmitted waveform or its frequency transform, of a frequency-dependent transmission function.

2. Apparatus as claimed in Claim 1 wherein the means for correlating the received signal operates to form signals representing the products of the instantaneous values of the corresponding samples of the received and comparison waveforms, and a signal representing the summation of these products.

3. Apparatus as claimed in Claim 1 or Claim 2 wherein the comparison waveforms are each the inverse Fourier Transform of the reciprocal of a frequency spectrum, this frequency spectrum being the result of the operation of a frequency-dependent transfer function on the frequency spectrum of the applied signal shape.

4. Apparatus as claimed in Claim 1 or Claim 2 wherein the comparison waveforms g(t) are each the summation of a number of waveforms gl(t) each of which is the result of the convolution of a respective function f,(t) and a function h(t), where h(t) is the impulse response of the line, and f,(t) are separate parts of the applied signal shape.

5. Apparatus for detecting the location of cross-talk faults in a signal cable substantially as hereinbefore described with reference to, and as illustrated in, the accompanying drawings.