Shielded Cable Transfer Impedance

B. Démoulin, L. Koné, You are not allowed to view links. Register or Login

TELICE-IEMN Group, Université Lille 1, (France), You are not allowed to view links. Register or Login

I. Introduction

In the previous articles [1], [2], we have described the setup of test benches for the measurement of the transfer impedance of shielded coaxial cables. We have demonstrated that, at frequen-

cies lower than 100 MHz, a conventional triaxial setup is suf-ficient to make this kind of measurements. The only restrictions of its use concern practical tests performed on shielded cables with high electromagnetic immunity. In this case, it is not

unusual to come across transfer impedances the absolute value of which falls below the mV/m. Under these conditions, the voltages of very small amplitude measured at the extremities of the test cables can be seriously perturbed by the radiation of the triaxial setup. Some adjustments were proposed with the aim of protecting the measurements from interference risks [1].

Keeping in mind that 1-m long cable samples fit very well for measurements of transfer impedances below 100 MHz, we demonstrated that in order to cover a frequency range up to 1 GHz it is necessary to reduce the cables length. Indeed, as the wavelength gets near or below the test tube dimension, propagation phenomena give rise to systematic measurement errors. In order to reduce this inconvenience, the dimension of the shield subject to the currents injected through the setup must be reduced to approximately 10 cm. Thus, this physical constraint calls for a total revision of the concept of measure-ment setup of transfer impedances itself.

The analysis performed in [2] mainly concerned two meth-ods respectively based on the the wire injection method and the shield discontinuity method in a triaxial setup. Despite this improvement, the measurements reveal that at frequencies close to or superior to the GHz range, the previous methods gener-ate new errors, this time due to the approximation of the TEM propagation. Indeed, be it the classic triaxial setup, the wire in-jection method or the shield discontinuity method, the transfer impedance measurement setup is based on the theory of trans-mission lines whose validity domain is necessarily dependent on the hypothesis of TEM signal propagation.

In order to extend the transfer impedance measurement to the microwave range, here assumed to cover the 1 GHz– 10 GHz bandwidth, at the beginning of the 80's we have started to characterize the cables shielding attenuation through prac-tical measures performed in shielded anechoic chambers. The cables, connected at both extremities to a matched load and to a spectrum analyzer installed outside the chamber, were submit-ted to an electromagnetic field generated by a large bandwidth antenna. This antenna, placed at 3 m distance from the cable, produces on it a local irradiation quite close to a plane-wave. The measure of the voltage amplitude gathered on the spectrum analyzer provides a clue on the attenuation generated by the cable shield. This rather simple process nevertheless presents three main difficulties. The measures will be hardly repeatable on account of the uncertainty of the radiation diagram of the ca-ble under test connected to the load and to the receiver through high-immunity coaxial cables. The physical contribution of the junction cables considerably influences the voltage amplitude induced on the shielding outer surface. As it is almost impos-sible to impose a repeatable setup configuration by means of a standard, the uncertainty becomes unacceptable. The second difficulty lies in the search for an objective reference magnitude, in the view of the shielding attenuation expressed as the ratio between two physical quantities having dimensions of voltage or power. Neither the electromagnetic field measured with a sensor, nor the power induced on a receiving antenna installed in the chamber could supply this reference. Their measures are still too dependent both on the position of the sensor and on that of the receiving antenna with respect to the transmitting antenna. They therefore do not account for viable indicators. Finally, we have to turn the shield attenuation in transfer im-pedance in order to compare the measurements with the results achieved below 1 GHz. The direct measurement of the induced

current on the outer shield surface could solve this issue, but it will cause huge uncertainties generated by the current collector and its connection with the outside receiver. This procedure was then abandoned.

During the same period, progress made in understanding re-verberation chambers allowed to use them to measure shielded cables attenuation [4]. As we will remind in section 1, the fields generated within reverberation chambers, submitted to differ-ent methods, offer interesting ways of producing repeatable at-tenuation measures through shielded cables or connectors. In fact, thanks to the oversizing of chambers in relation to wave-lenght, the objects installed in cavities with high conductive walls gain isotropic electromagnetic behaviour. This means that they are characterized by the absence of defined directivity and polarisation. Moreover, mode stirring provides uniform average amplitude of the field throughout the whole chamber. How-ever, the average amplitude is associated to a standard deviation that becomes constant for operating frequencies of the fields situated above a minimum value defined by the chamber di-mensions. The isotropic and uniformity properties of the fields allow using the power received on an antenna installed in any place in the chamber to measure the reference level. So it be-comes possible to evaluate the shielding attenuation by means of a ratio between the power received on the spectrum analyzer connected to the cable end, and the reference power captured on a broadband receiving antenna. Other advantages related to the properties of cables immersed in a reverberation chamber were added. Indeed, according to the statistics of the random data collected within a reverberation room, we have established analytical formulas which yield, through a very simple calcula-tion, the conversion of the shielding attenuation into the trans-fer impedance [12].

This third article will thus be entirely devoted to the de-scription of protocols which lead to measuring the shielding attenuation in mode stirring reverberation chambers. This con-version, given in terms of transfer impedance, will be demon-strated and then illustrated by means of examples. Section 1 will be devoted to remind us the properties of reverberation chambers. It will also include the description of the installa-tion of shielded cables (or the connectors) in preparation for the measurement of the shielding attenuation. The reader who wants to go deeper in the physical issues can refer to special-ized articles on reverberation chambers. Section 2 concerns the description of the protocols adopted to measure the attenuation of shielded cables. Section 3 will mainly address the conversion of the shielding attenuation into transfer impedance and vice-versa. Section 4 deals with the chambers calibration in view of determining their natural uncertainty margin and comparing transfer impedances taken from measures with a standard based on theory. Section 5 will show the results of different measure-ment performed on samples of shielded cables.

You are not allowed to view links. Register or Login