A Contribution to the Determination of the Electrical Length of Mated Data Grade Connectivity Based on Channel Measurements and Simulations

摘要

We demonstrate in the present paper that we cannot describe the insertion loss deviation of cascaded components (ILD = insertion loss of cascade - summed up component attenuations) as a monotonous increasing function. In fact, the ILD is cyclic over frequency. This behavior has been designated by the original investigators as "wiggly insertion loss curves" [1]; [2]; [3]; [4] for coaxial cables, depending upon the length of the repetitive impedance irregularities inside the cable or the spacing of the connectors. This has been confirmed in a paper by Walling [5] and Hess et al [6] for balanced cable channels. As the cables are electrically relatively long, see their cyclic variations in the ILD traces with a relatively low frequency repetition rate. However, connectors are short, and as a result, we their repetition rate over frequency can only be demonstrated at relatively high frequency. Though we did not consider this in the past, it has a serious impact upon any setting of specification limits, especially if we consider higher frequency ranges, I.e. for instance 10 Gbase-T and similar high frequency protocols. As a consequence the ILD has to be reflected as a sinusoidal curve over frequency for specification purposes.Our objective here is the experimental determination of a reasonable "electrical length" of the mated connectors to be used in simulations.