In this paper, the effectiveness of applying an extra time delay to the data transmitted over redundant channels to cope with harsh electromagnetic interference is studied. The redundant system is compared with and without extra time delay and compared with a non-redundant system, which is the reference situation. The geometries under study are disturbed by strong incident fields, which represent reverberation room conditions. These conditions refer to the planewave integral representation, where multiple planewaves with random angles, polarisations and phases are combined and normalised, to create a reverberation room like environment. A reciprocity based technique is used to efficiently calculate the induced voltages and the resulting bit error rate for the different situations. Because of the properties of the geometries that are under test and the introduction of a time-delay in the transmitted datastream, the moment the induced voltage due to EMI is sampled is different, which causes an EMI-diverse system. The results show that using time-diverse signals optimized for the main disturbance frequency to be expected, is a very effective way of creating EMI-diverse behaviour. In a majority voting (2oo3) system it is possible to eliminate all triple occuring faults. Even for higher field strengths and different EMI frequencies, the time diversity method performs well (better than other EMI-diverse techniques).

This paper studies the effectiveness of inversion diversity to cope with EM disturbances caused by strong incident fields under reverberation room conditions. Here, reverberation room conditions refers to the situation where multiple plane waves are incident onto the system-under-study, each with a random angle-of-incidence, polarization and phase. Two different geometries are compared: a non-redundant system comprising a single trace on a PCB and a redundant system comprising two parallel traces on a single PCB. A reciprocity-based technique is used to efficiently calculate the induced voltages and the resulting bit error probability in the different geometries. It is shown that with increasing distance between the two traces, the probability on false positive (i.e. two identical, but incorrect bits at the comparator) increases.

No abstract

No abstract

No abstract