Cyclostationary Characterization of Radiated Emissions in Digital Electronic Devices

Stochastic electromagnetic fields originating from processes in digital circuitry affect the operation of the circuit, increase the error rate in the transmission of digital signals, degrade the circuit's performance, yield malfunction of the system or even lead to a complete failure of the system's operation. Interference from stochastic fields are not only disruptive but are becoming more common due to the increasing complexity of modern electronics. Stochastic electromagnetic fields under common circumstances are due to a mixture of emissions from different sources, which makes a quantification of stochastic fields very challenging. Since digital signals are clocked, the ensemble averages of the electromagnetic field transients representing the digital signal exhibit a periodic time dependence with the period of the clock interval. The probabilistic model of the cyclostationary random process was composed by appropriate transformations of the initial discrete random process of the binary symbols. Exploiting the cyclostationary characteristics of stochastic fields greatly simplifies their analysis. The physical signal propagating along the transmission line is considered also a source of the radiating electromagnetic field which preserves the cyclostationary properties of the binary sequence. The stochastic process of the measured near-field emissions inherits probability characteristicsfrom the random process of the physical signal in the transmission line. This paper presents an approach for quantifying emissions and coupling of stochastic fields using their cyclostationary properties. For practical evaluation of cyclostationary characteristics the statistical cyclic averaging of the measured signals are used. The important feature of the statistical cyclic averaging procedure is its capability to distinguish cyclostationary signals from a mixture of stochastic processes such as stationary noise with deterministic signals and even from other cyclostationary signals with distinct cyclic frequencies. The measurement setup for observing electromagnetic emissions from printed circuit boards (PCBs) hosting digital electronic devices comprises an automated positioning system with a scanning near-field probe attached to the moving scanner head. This scanning near-field probe and a properly position fixed or also movable reference probe are connected to the inputs of a multi-channel digital oscilloscope for simultaneous and synchronous sampling and registering of the magnetic field components. The registered data from the oscilloscope was transferred to a data acquisition computer. After the postprocessing procedure, the spatial localization of distributed radiating sources on the surface of the PCB was performed in conjunction with determination of their average power characteristics and specified cyclic frequencies. The proposed cyclic averaging was applied to the spatial separation of radiating sources with different bit rates and for the quantitative characterization of crosstalk and coupling between transmission lines carrying the data sequencesand subsequent estimation of bit error rate (BER) caused by the crosstalk using their cyclostationary properties.