Historically, the two types of antenna which have been used for emission measurement have been the biconical and the log-periodic. These are both electric field linearly polarised and typically cover complementary frequency ranges of 30-300MHz and 300-1000MHz respectively. They have been combined into the BiLog® which exhibits similar characteristics to each over the relevant frequency range.
There are other antennas available but these do not always follow the same principles. An example is the log spiral, which has been widely used for military susceptibility tests. This receives or generates circularly rather than linearly polarised fields and so it is not suited for commercial IEC/CISPR tests. Other designs of biconical are sometimes used, varying in size or method of terminating the conical section.
The standard which defines the requirements on antennas for EMC measurements is CISPR publication 16-1 : 1993, Specification for radio disturbance and immunity measuring apparatus and methods, Part 1.
CISPR measurements officially require tuned dipoles, but a note in most standards (EN55103 is an exception) allows the use of broadband antennas where these can be shown to give equivalent results. For this to be the case the antenna must be calibrated to give the antenna factor.
The allowed system uncertainty in CISPR-16 is ±3dB, made up of the receiver, antenna, cable loss and mismatch uncertainties, but not including the test site. Directivity should be low so that signals from ground plane reflections, are not significantly attenuated. CISPR-16 mandates linear polarisation. The VSWR should be less than 2:1, which strictly speaking demands an attenuator on the output of most biconical antennas.
CISPR-16 Broadband Antenna Requirements 30-1000MHz
Polarisation Substantially plane polarised
Cross Polarisation 20dB below aligned polarisation
Polar Pattern Direct/ reflected ray response within 1dB
VSWR Less than 2:1
Accuracy ±3dB for antenna plus receiver
The maximum achievable field strength is related to the power that can be radiated from the antenna. For a dipole, the field strength E (V/m) at a distance d (metres ) with P (watts) radiated is
E = 7*(SQRT(P)) / d
Power handling is principally a function of the balun design. Some power is dissipated in the balun and the resulting temperature rise limits the power that can be applied. Additionally, good VSWR is essential to minimise the power reflected from the antenna and hence make best use of the capabilities of the power amplifier. For these reasons many antennas that are designed for radiated emissions test cannot be used for radiated immunity.
Low directivity maximises the area that can be covered with a constant field strength. IEC 1000-4-3 mandates a uniform area of field at the position at which the EUT is located. This is partly affected by reflections from the screened room walls but also by the uniformity of coverage of the antenna. To cover a given area, a narrow beamwidth antenna will need to be located further from the EUT. If even a broad-beam antenna is too close, then different parts of the elements will be at significantly different distances from the EUT and uniformity will suffer.
Low absolute value - high system sensitivity
No sharp resonances - easy interpolation
Minimum sensitivity to external factors
The field strength E, is obtained by multiplying the voltage V across the load by the antenna factor AF, allowing the attenuation in the connecting cables. The conversion can be expressed by;
E (dBµV/m) = V (dBµV) + AF (dB/m)
Since the noise floor of the measuring instrument (minimum measurable V) is fixed, a good system noise floor (minimum measurable E) depends on a low antenna factor. Even with this setup, the noise floor exceeds the Class B limit line for 10m distance. A lower sensitivity antenna makes the problem worse.
It is becoming common practice to quote an antenna factor which applies to free space conditions and assumes a 50R load. The AF changes with proximity to the ground plane and the EUT. Errors due to mutual coupling cannot be avoided but if all test antennas use a similar design they will at least be invariant. Mutual coupling with small EUT's even at 3m is unlikely to cause errors greater than 1dB. Variation of AF with height above the ground plane is most pronounced in horizontal polarisation and is typically 1dB at resonance (around 70MHz) falling to 0.5dB above 300MHz for a Bilog or a biconical antenna. Errors due to mismatch depend on the match at the receiver input and the antenna VSWR. Variations in the slope of the AF curve due to resonances should be minimised. The AF values must be programmed into test software which interpolates for a particular frequency, & for minimum error a smooth curve is essential.
Antenna VSWR (Voltage Standing Wave Ratio) affects two aspects of EMC testing; the accuracy of emission measurements and the power required to perform radiated immunity tests. The mismatch uncertainty on receiver measurements is given by:-
U (dB) = 20log10 (1 ± |rA| |rR|)
where rA and rR are antenna and receiver reflection coefficients, given by;
| r | = (VSWR-1) / (VSWR+1)
The CISPR16 requirement is for a maximum antenna VSWR of 2:1. This is rarely met in practice below about 80MHz, but with this figure for both antenna and receiver the uncertainty from the above equations is +0.9/-1.0dB.
When power is fed to an antenna, some is reflected back to the power source and is dissipated in it. This reduces the power that is radiated from the antenna to generate the required field strength, and also over-stresses the power amplifier; some amplifiers can be damaged by this condition. A VSWR of 3:1 will cause half the power applied to be reflected. The lower the VSWR, the more efficient the antenna (assuming no resistive losses) and hence the less power is required for a given field strength. VSWR can be improved by inserting an attenuator pad but this itself absorbs power and reduces sensitivity.
Polarisation, Polar Pattern & measuring distance
When a single antenna factor is specified it assumes that the antenna is used under conditions of maximum gain. For the log periodic antenna this is in the direction towards which the antenna is pointing, for the biconical it is perpendicular to the antenna axis. In other directions the response of the antenna falls off and the antenna factor is no longer valid. For a dipole the polar pattern response is within 1dB of the on-axis value over an azimuth variation of 45º. For a log-periodic array the beam is narrower. This is significant particularly when the antenna is used at high frequencies with a height scan from 1 to 4m and a close-in distance of 3m. Under these conditions the antenna is no longer properly aligned with the EUT and an error can result.
Polarisation of the antenna refers to the plane of polarisation of the electric field component. CISPR16 requires that the corss-polarisation be better than 20dB, which implies that the design of the antenna must ensure it is linearly polarised.
Most EMC antennas use similar mechanical designs and as a result may of their properties are comparable. One parameter where this is not so relates to balun design. The antenna balun converts from the unbalanced coax feed to the balanced termination between the antenna elements. The resulting balance is assessed by comparing signals received in the two possible vertical orientations. A well designed balun will keep the difference to less than 1dB, while a poor design can show differences greater than 10dB.
Poor balance will show up in several ways. The antenna factor is clearly going to depend on which way up the antenna is mounted, and will be different for horizontal and vertical polarisation. Equally important, poor balance will severely affect the uncertainty due to proximity of the ground and the antenna cable in vertical polarisation. An unbalanced termination results in appreciable common-mode RF currents flowing on the sheath of the antenna cable. These then couple to the vertical antenna elements and cause a change in the polar pattern and in the antenna factor. This problem can be reduced by taking the antenna downlead far away from the antenna and by applying ferrite sleeves to the cable, but good balance reduces the need for such measures.
Ground proximity cannot be avoided and the only way to minimise error is to improve the antenna balance.
Uncertainties can be attributed to several distinct sources, including instrumentation (the sum of antenna factor, cable loss, receiver calibration and mismatch errors) and that due to the measurement method. This latter is summarised in the table below for biconical and log periodic antennas over their respective frequency ranges, at the three common measurement distances. The contributions are added on the root-sum-of-squares principle: see NAMAS information sheet NIS81 for the full method.