By John Thompson, Westek Electronics Pty Ltd

EMI measurements can be subject to a larger number of errors including incorrect bandwidth selection, incorrect selection of detector type (peak, quasi-peak or average), inappropriate scan time for the frequency range to be examined, overload caused by out-of-band signals, ambient noise adding to signal, and broadband noise preventing detection of noise peaks. To add to this, antenna and site accuracy have to be taken account of. An important factor is the constancy of the electric field source. As an example, a simple dipole antenna might be carefully calibrated with respect to a near infinite metallic ground plane but a different signal generator from the one used to calibrate the antenna will alter field parameters. When it comes to receiver calibration, the quasi-peak mode is the important one to get right.
 
The terms spectrum analyzer and EMI receiver are often used interchangeably. Although there are similarities in their architectures, for example the super-heterodyne mixing principle, the EMI receiver is front-ended by preselection filters and attenuators. The other important difference is in the IF section which has further filters as well detectors, principally the quasi-peak detector. The quasi-peak detector is basically a peak detector (i.e.: detecting the envelope of the filtered IF signal) followed by a lossy integrator. Historically the detector has its nascence in its ability to mimic the annoyance level experienced in listening to interference on an AM radio channel. The detector has a fast rise time and longer decay time so that the sequence of impulses separated by very short time intervals add to one another.
 
For analogue spectrum analysers such as the AFJ ER55 EMI receiver, the sweep speed in Hz/sec has to be matched to filter bandwidth: i.e. the bandwidth should be bigger than the square root of the sweep speed. That said, the big advantage of analogue analysers that there is no Nyquist limitation imposed by sampling speed, as is the case for digital instruments.
 
The preselection filter as mentioned earlier provides overload protection against broadband noise for the input mixer. In the case of the AFJ ER 55 EMI receiver up to 15 fixed and tuned preselection filters provide more than 40 dB attenuation. A particularly important consideration is the use of a receiver for conducted interference analysis. Spikes with high spectral energy density can easily wreck input stages. To prevent this the use of an additional attenuator is highly recommended. The AFJ PAT 20 dB attenuator can withstand voltage pulses with 1 joule energy.
 
Conducting EMI tests can be a complicated business, as mentioned already. The facility to operate the receiver under computer control can therefore be an important advantage. In the case of AFJ ER 55 EMI receivers, Windows-based software permits the operator to set data acquisition parameters in accordance to CISPR 16-1, or for that matter to other criteria. In terms of meeting calibration requirements, antenna k-factors and GTEM correction factors are accounted for in software. Receiver settings, measurement set-up and test data are saved in user-defined workspaces of the database.
 
Low noise floor performance is important, the reason being that in the use of attenuation in order to provide increased spurious effects-free dynamic range, there is increased possibility of low level emissions falling below the noise floor of the receiver. For the AFJ EMI 55, the noise floor can be as low as –13 dBV at 200 Hz IF bandwidth utilizing the quasi-peak detector.
 
In running EMI tests, overload situations must be kept in mind. As already mentioned input filtering is important in avoiding this. However the relationship between pass band of the preselection filter and the spectrum width of the broadband input signal will determine the overload-avoidance effectiveness. On the other hand, narrow band high-spectral energy signals lying within the pass band will get through to the mixer and possibly cause overload.

Caption:
AFJ ER 55 EMI receivers comply with CISPR 16-1-1 and CISPR 16-2. Three models cover the frequency spectrum to 30 MHz, 1 GHz and 3 GHz with measurement bandwidths of 200 Hz, 9 kHz, 120 kHz and 1 MHz (3 GHz model). A built-in tracking generator permits insertion loss measurement in accordance with CISPR 15 (as required for electronic ballast testing. A powerful CPU with 1 Mbyte dynamic memory supports testing routines. Interfaces include LAN 10/100 Mbit and USB.


<< Click here to contact us regarding this article >>