Departure of repeater housing crosstalk from the Gaussian amplitude distribution

For PCM digital transmission systems, used on symmetrical pair cables, the performance is usually limited by near-end crosstalk (NEXT), introduced into the disturbed pair by crosstalk paths in the first few hundred metres of the main cable. There are, however, also NEXT crosstalk paths inside the repeater housing and in the (electrically short) tail cables. Such crosstalk is referred to as `repeater housing crosstalk¿¿ or `apparatus case crosstalk¿¿ and in some circumstances may be the source of interference which limits system performance. The difference between repeater housing crosstalk and near-end crosstalk from the main cable is discussed. A model of repeater housing crosstalk is developed, and its use in simulating real conditions is shown. Two Monte Carlo simulations are shown to be necessary: the first, to show the variation from housing to housing-the `outer loop¿¿ the second, to calculate the magnitude of the 10¿¿6 peak total crosstalk-the - inner loop¿¿. The methods of calculation are fully described and applied to a particular example. It is concluded for this example that, at the cumulative 99.9% point for repeater housings, the `10¿¿6 peak voltage¿¿ level is approximately 4 dB less than that which would be obtained if the voltage were Gaussian distributed; the difference falls to about 3 dB at the 95% level and further reduces to about 1 dB at the median (50%) level. For the whole distribution of housings, the mean `peak factor¿¿ was 12.5 dB (compared with 13.5 dB for a Gaussian distribution) and the standard deviation was 0.7 dB.