The Comparison Noise Emitter and other reference radiators and their uses in EMC Measurements

Technical Information Notes
Technical Application Notes
Technical Papers

Professor A.C. Marvin. Technical Director, York EMC Services Ltd, University of York, York YO10 5DD UK.


The Comparison Noise Emitter (CNE) has been used for a number of years in the assessment of the performance of EMC measurement facilities. Examples of an early version were used by NAMAS in the UK for an intercomparison of EMC test facilities and a major multinational electronics company has used the device for a similar survey of its measurement facilities world-wide. The CNE’s name was chosen to indicate the fact that it is an intercomparison device and is not intended as a standard noise source. However, the NAMAS report indicates that the device has an output stability of ±0.5dB when calibrated at intervals between its tests at the various EMC measurement sites [NAMAS Ref. NPD/027/18/05, 1992]. Measurements were taken over a period of three years. The main function of the CNE is to ensure that the response of the facility is stable and to indicate problems associated with such as damaged cables connectors or antennas or changes associated with equipment placement errors. An example of the use of a CNE for this purpose is illustrated in the section below. The CNE can also be used as an inter-comparison device to assess the relative performance of different test facilities. However, as is shown below, reference sources of this type cannot be used in a simple way as a kind of transfer standard to enable measurements of equipment’s-under-test (EUT’s) performance made on one test facility to be converted to an estimate the EUT’s performance at another facility. This is a common misconception with reference sources of this type and such claims should be treated with scepticism.

The CNE comprises a stable wideband noise source with a frequency response extending from 9kHz to 2GHz. The noise is amplified to the required output power. The unit is housed in a rectangular enclosure of dimensions 188mm × 120mm × 62mm equipped with internal batteries. It is equipped with a choice of three monopole antennas covering 30MHz to 100MHz, 30MHz to 1GHz and 1GHz to 2GHz for radiated measurements. A version enclosed in a 100mm diameter spherical dipole is also available. An IEC adapter is provided to allow coupling to mains cables for conducted measurements. Other devices (reference sources) exist which can be used for the same purposes and have similar dimensions. These all rely on the use of periodic signals with fast rise and fall times to give line spectra that extend over the desired frequency range although the number of frequencies generated is limited by the clock frequency and does not extend below that frequency. In this paper consideration is given to the possible uses and limitations of these devices and the differences between line spectrum sources and continuous noise sources are discussed.

Test Site Checks with the CNE

Fig 1 and Fig 2 which show the daily response taken on the Open-Area Test-Site (OATS) of York EMC Services on a normal day and on a day after a heavy storm had resulted in the ingress of water into the cables from the antenna. It is very clear from this measurement that something has changed in the measurement set-up. The periodic nature of the frequency response gives a strong indication that cables or their interconnections are a problem.

Fig 1 OATS daily test response (normal)

Fig 2 OATS daily test response showing the effects on the system of water ingress into measurement cables

Frequency Response Measurements

In general, both types of device are used to measure the frequency response of an EMC radiated emission measurement facility. This is necessary as all these facilities suffer from multiple energy propagation paths between the EUT and the measurement antenna. The advent of anechoic rooms for radiated emission measurements will overcome this problem to a large extent but no room is perfectly anechoic and a performance check will always be required. The frequency response arises because the direct and reflected waves are summed at the measurement antenna.

Fig 3 Side view of OATS

Fig 4 Frequency response of an unlined screened room taken with a CNE and with a line spectrum source of similar external appearance

The simplest such scenario is provided by the OATS where the field measured by the antenna is the summation of the direct radiation from the EUT and its ground reflection. At a particular frequency, the summation depends on the directional properties of the EUT and the measurement antenna and also on the path length difference expressed in wavelengths between the direct radiation path and the ground reflection path. This path length difference is a function of the actual physical path length difference and the ground reflection coefficient. The situation is illustrated in Fig 3. Consider the situation where at a particular frequency the two waves arrive at the antenna in antiphase. A minimum or null is obtained in the frequency response. If the frequency is changed the path length difference measured in wavelengths and hence the relative phase of the two waves also changes and the response moves away from the minimum. The width of the null in the frequency response depends on the rate at which the relative phase of the two waves changes with frequency. For an OATS as shown with a typical path length difference of 2.5m it can be calculated that a null with a width between its -20dB points of around 8MHz results. A device with a line spectrum with 10MHz intervals could easily miss such a null, whilst a device with a continuous spectrum would show the null. In an unlined screened room the situation is more complex with multiple reflections giving rise to many peaks and nulls. The room behaves as a resonant cavity. The frequency responses shown in Fig 4 were taken in such a room with dimensions 6m x 3m x 3m using a CNE and a line spectrum source of similar external appearance. Whilst the spectral lines follow the continuous spectrum in as much as there is a constant offset between them at any line frequency, it is clear that significant frequency response information has been lost with the line source.

This example raises the question of what is the narrowest feature in the frequency response of a measurement system that can be observed using a source with a continuous spectrum.

The use of a continuous spectrum device and a swept receiver to assess the frequency response of the system effectively means that a narrow band filter is used to sample the system frequency response. This filter is the intermediate frequency (IF) filter of the swept receiver. The frequency resolution is determined by the bandwidth of the filter. Fig 5 shows measurements of high Q factor resonances in a screened room with receiver IF bandwidths of 10kHz and 1MHz. It can be seen that the narrower the bandwidth the finer the frequency resolution. A further effect of using a continuous spectrum device is that the apparent signal to noise ratio is unaffected by the receiver bandwidth. This is due to the fact that both the wanted signal power, the noise spectrum of the source, and the internal receiver noise power are proportional to the receiver bandwidth. The reduction in measured signal level using the continuous spectrum source with a lower measurement bandwidth is compensated for by the reduced receiver noise floor obtained with the reduced bandwidth.

Fig 5 Frequency response of an unlined screened room showing the effect of measurement receiver IF bandwidth

The use of a continuous spectrum noise source requires that some averaging (or equivalent video filtering) is performed in order to reduce the measurement uncertainty. It can be shown that, for a given measurement bandwidth, the measurement uncertainty is inversely proportional to the square root of the measurement time. In other words, the measurement uncertainty associated with a noise measurement can be reduced by increasing the measurement time which is equivalent to reducing the post-detector (video) bandwidth. The plots of Fig 5 were taken with a video bandwidth of 100Hz. This does increase the measurement time compared to a line spectrum source, the trade off being that more of the frequency response information is available from the CNE.

Prediction of Measurements on an OATS from Pre-Compliance Measurements

The idea of predicting compliant OATS measurements from pre-compliance measurements is very attractive. It is frequently proposed that this can be done by adding a correction factor to the measurements made on the pre-compliance site and that this correction factor can be obtained by measuring a reference source (CNE or similar device) both on the OATS and on the pre-compliance site and taking the ratio (dB difference) as the correction factor. This is a very simplistic interpretation of the situation and it cannot be relied upon to work.

A reference source is a simple radiating structure which acts as an elemental electric dipole at frequencies where it is electrically small, say below about 300MHz, and as any other normal radiating dipole antenna at the higher frequencies. It radiates a linearly polarised wave, vertical, horizontal or slant polarisation depending on its orientation.

A typical EUT is a distributed source of radiation possibly extending over several metres with cabling. In general it will radiate an arbitrarily polarised wave. The polarisation will change with frequency. This is a much more complex situation than that posed by the CNE. Measurements made on an OATS include both the horizontal and the vertical components of this arbitrary polarisation and, as such, are a pragmatic engineering attempt to evaluate the worst case threat to radio communications systems posed by the EUT’s emissions.

A relatively simple EUT, i.e. one that is small with no long cables, can be considered to be three orthogonal dipoles, a vertical dipole, a horizontal dipole and a further horizontal dipole aligned along the measurement site which we term a longitudinal dipole. The relative strengths of these dipoles are unknown and change with frequency. On a perfect OATS the vertical and longitudinal dipoles of the EUT would not couple to the horizontally polarised measurement antenna. They are cross-polarised. Similarly the horizontal dipole of the EUT would not couple to the vertical measurement antenna. The longitudinal dipole of the EUT can couple to the vertical measurement antenna as they are not completely cross-polarised. On pre-compliance test sites and to a much lesser extent on imperfect (i.e. real) OATS’s other cross-polar coupling can take place.

Fig 6 Difference between OATS measurements and ‘corrected’ measurements made on a pre-compliance test site

Fig 7 Difference between OATS measurements and ‘corrected’ measurements made in an unlined screened room

Typical causes of this coupling on OATS’s are groundplane edges, antenna masts, antenna cables and residual antenna imbalances [Turnbull & Marvin, 1996]. On pre-compliance sites other nearby structures and the general uncontrolled construction of the site plays a major part. If the pre-compliance site is in an unlined screened room, then the internal field structures give rise to cross-polar coupling. Thus the measurement antenna receives a wave the source of which is an unknown cocktail of waves radiated by all three dipoles of the EUT. This cocktail is very site dependent. A simple correction factor between two sites made with a CNE or similar device simply cannot cope.

This is illustrated in Fig 6 where a correction factor of the type described has been applied between measurements made on a pre-compliance type site set up in an otherwise empty carpark and a UKAS accredited OATS. The EUT was an old, once very common, microcomputer known to radiate a relatively high level of interference. The bar chart shows the differences between the predicted OATS measurements and the actual OATS measurements at each of the harmonics of the microcomputer’s clock frequency. In both cases care was taken to set up the EUT with the same cable layout and orientation.

Fig 7 shows a similar bar chart for measurements of the microcomputer taken in an unlined screened room and on the OATS. In neither case can the corrected measurements be said to be equivalent to the OATS measurements with errors of up to 20dB between the real OATS measurement and the "corrected" measurements made on the pre-compliance site. The errors shown in Fig 7 for the screened room are even greater, approaching 30dB at some frequencies. Beware claims that this is a reliable technique!


In this short article I have indicated the uses and limitations of Comparison Noise Emitters and similar devices reliant on line spectra. All these devices will give useful information about the performance of various types of test facility including OATS, anechoic chambers, screened rooms and TEM cells. Substantial care should be taken in using them to attempt to characterise pre-compliance test facilities in order to estimate emissions on other compliant sites.


NAMAS ‘Report on Initial Phase of Interlaboratory Comparisons of Radiated Emission Measurements using a Comparison Noise Emitter’, Ref. NPD/027/18/05, 1992

L. Turnbull & A.C. Marvin ‘Effect of cross-polar coupling on open area test site measurement correlation and repeatability’ IEE Proc-Sci. Meas. Technol., Vol 143. No4, July 1996