Search

The effects of radiated radio-frequency interference (RFI) on the operation of digital systems are studied by simulating the response of simple logic circuits to incident plane waves. The simulation is accomplished by combining a linear electromagnetic moment-method model of the wire structure with a nonlinear circuit model of the solid-state components. The complete model is analyzed in the linear and noulinear regimes as an example. It is shown how a circuit simulator, such as SPICE, can be used in the analysis of an arbitrary wire network loaded with logic circuits, by the process of representing the linear wire network as a lumped-element N-port [Pi] network and interfacing it to the nonlinear circuit simulator. Examples are given that demonstrate the occurrence of both static and dynamic failures under various RFI-field excitations and wire structure geometries. The prediction methods presented in this paper, can be used by eme engineers to assess the likelihood of failures in RFI-exposed digital systems.

Delays induced by radio frequency interference (RFI) in CMOS inverters are measured under radiated and capacitively coupled interference. Experimental and theoretical investigations of the effects of the RFI coupling mode (capacitive venus inductive) on the amount of induced delay are presented and a worst-case coupling mode is identified. A formula to predict delays caused by in-band low-level RFI in CMOS inverters is introduced. This formula uses experimentally determined parameters which are dependent on the coupling mode. The change in delay is computed as a function of the induced voltage disturbance, which in turn can be computed from the incident field using linear frequency-domain analysis. The formula accounts for the dependence of the induced delay on the phase and amplitude of the RFI signal as web as on the slew rate of the logic transitions. A delay growth phenomenon in a string of inverters is identified and characterized. A correction to the delay prediction formula is proposed in order to take this growth into account in worst-case predictions.

An investigation of the multifilament current method's (MFCM) ability to solve electromagnetic scattering and/or absorption problems involving inhomogeneous cylindrical structures is presented. Dielectric cylinders of arbitrary cross section covered by multiple layers of lossy dielectrics are considered. Both cases of wave incidence, TM and TE, are treated. Like some other moment-method solutions, the MFCM experiences numerical difficulties when dealing with a medium characterized by a high loss tangent or large electrical conductivity. To overcome this problem, a new boundary condition based on an impedance matrix equivalent circuit approach that accounts for the curvature of the surfaces has been developed. This new impedance matrix boundary condition (IMBC) extends the MFCM capability to the analysis of two-dimensional (2-D) structures involving conducting layers with ohmic losses. The usefulness of this extended method is confirmed by the study of metallic shells for which a strong energy coupling with the incident electromagnetic (EM) field is demonstrated at their structural resonance frequencies. Index Terms--Impedance matrix boundary condition, multifilament current method, shielding.

In this paper, a method to build equivalent models of radiating printed circuit boards based on complex E-field measurements taken in the close vicinity of the device under test is explored. It is shown that the inverse problem to be solved to retrieve the currents from the field data is ill-posed. An innovative regularization approach implementing a penalty on abrupt spatial variations of the currents is proposed to alleviate this difficulty. Various schemes to mesh the equivalent current distribution are also explored. Combining these with measurements, it is shown that accurate estimation of equivalent current models can be achieved, thereby allowing the identification of the emission sources. The method is tested for different circuit configurations with both synthetic and real data. Obtained results demonstrate the efficiency of the method. [ABSTRACT FROM AUTHOR]