Automated TDEMI measurement
The data acquisition (DAQ) process for the time-domain measurement starts with the sample process of the ADC. Then, the spectra via FFT are digitally computed. After FFT, the errors due to the frequency characteristics of the antenna, transmission line, amplifier and anti-aliasing filter are corrected by signal processing. Next, the analysis of peak-, RMS- and average-values of the EMI signal is performed. An additional noise-floor adjustment allows a comparison of the results obtained in a conventional EMI receiver.
The flowchart of the basic algorithm of an automated time-domain EMI measurement with the TDEMI is depicted in Figure 2. First, the parameters M and N are computed
with ΔTM as the observation time, Δf is the frequency resolution and fS as the sampling frequency. Then, the program occurs a loop that M-times will go through. With each run, a timedomain data vector of the length N is read in, transformed into the frequency domain and supplied to the detector modeling. With the expiration of M itera-tions, the resulting amplitude spectrum of the detector modeling comes into the logarithm procedure. Finally, a correction of errors originating from the frequency characteristics of the TDEMI system is made.
Stationary EMI signals
In general, the signals encountered in a typical EMI measurement scenario are stochastic in nature. Besides harmonic components and noise, they may also contain transients and bursts. It can be said, though, that a sample x (t) with tO ± t < tO + ΔTM of the stochastic EMI signal contains all the information about the process if the observation interval ΔTM is sufficiently long. In this case, the properties of x (t) are independent of the arbitrarily chosen starting time t0, and the signal may be considered quasi-stationary.
The radiated EMI of a commercial laptop PC was examined with the TDEMI system and a conventional EMI receiver, and the results of both were compared. Figure 3 shows a typical time-domain data vector retrieved from this setup. Besides noise, it consists of strong stationary harmonic components emanating from the various clock signals used in the circuit. Figure 4 shows the result spectra obtained from the TDEMI system and the conventional EMI receiver, both using the average detector and an observation interval ΔTM =5ms. The step size was 50kHz, and the receiver used an IF filter bandwidth of 120kHz. Figure 4 shows a match of the amplitude spectrum measured with the TDEMI system and conventional EMI receiver for narrow-band harmonic signals, the mean deviation is typically less than 0.5dB. Slight differences appear in the noise f loor. These are caused by the differing noise behavior of the TDEMI and the conventional noise receiver.