Nearfield Systems Inc.
|
About
Issues Involved with Near-Field Measurements
Testing
antennas in the near-field sometimes leads to unexpected
results. Some of the more common unexpected surprises and
pitfalls are discussed below. Topics on this page are excerpted
from Dan Slater's book, "Near-field
Antenna Measurments".
1. Higher RF power is often required when testing large
antennas.
Higher
gain antennas under test typically (but not always) have
a higher insertion loss. This is because the power density
in the aperture of a high gain antenna is lower since the energy
is dispersed over a larger area. The insertion loss in the
near-field region is approximately equal to the relative
aperture mismatch loss. Much of the effect of the insertion
loss is, however, counteracted by the process gain in the
far-field transformation. The insertion loss based on area
ratios (valid for uniformly illuminated apertures) is equal
to the absolute value of the difference (in dB) between the
probe and test antennas.
2. Insertion loss is not affected by the direction of
RF travel, i.e., whether the test antenna is transmitting or
receiving.
The insertion loss for reciprocal antennas is not affected
by changing the probe antenna from a receive to transmit configuration.
This is a direct consequence of the reciprocity theorem.
3. Multipath errors are not affected by the direction
of RF travel, i.e., whether the test antenna is transmitting
or receiving.
Multipath levels and errors are not affected in any manner
by changing the probe antenna from a receive to transmit configuration.
This is a direct consequence of the reciprocity theorem.
4.
High gain probe antennas often provide significant advantages
over the commonly used open ended waveguide probe.
A common misconception is that only physically small,
low gain probes should be used, minimizing any disturbances
to the electromagnetic field. A large aperture, high gain probe can often provide
significant advantages by operating as a spatial filter to
minimize multipath, reduce sampling densities and reduce insertion loss. The
disturbances to the electromagnetic field are similar to those
produced by a low gain probe since fewer data points are taken with the
high gain probe antenna. The high gain probe is often not usable
for other reasons. As an example, it is not the probe of choice if wide angle
sidelobe measurements are required.
5. The gain of the probe antenna need not be less
than the test antenna.
The far field response derived from the phasefront measurement
is a measure of the convolution of the probe and AUT response.
If the response of one antenna is available, the response of
the other antenna can be obtained by deconvolution. As in most
deconvolutions noise buildup can occur. For high quality processing,
the probe gain should be high over the range of far field angles
required in the
output data.
6. The planar near-field to far-field transformation is a misnomer.
It does not transform the near-field to the far-field.
The "near-field to far-field transformation" is a misnomer
since it actually converts a phasefront into an equivalent
angular energy distribution at the phasefront location. The
correct terminology is a phasefront to angular spectrum transformation.
The phasefront at any distance from the AUT, including at far-field
distances and the angular spectrum, are a Fourier transform
pair.
The
term "near-field to far-field transformation" came about
since the antenna is normally, but not always, probed in the
near field region. The angular spectrum is equivalent to the
far field radiation pattern.
An example of the invariance of the far-field transform
to distance is demonstrated in VLBI and VLA radio astronomy.
The far-field transformation is used in VLBI and VLA radio astronomy
to form images of distant radio sources external to our galaxy.
In this configuration, the extragalactic radio sources are clearly
in the far-field of the radio telescope. In one VLA radio telescope,
27 large 85-ft. antennas, are connected to phase measurement
receivers to form an even larger aperture with a dimension of many miles.
This concept, called aperture synthesis, is produced through
the use of the "far-field" transform. The phasefront, as sampled by the
27 antennas, is transformed into an angular spectrum to form
the image of the radio star.
7. The planar far-field transformation is amplitude invariant
to the separation distance between the probe and test antennas.
The distance between the probe antenna and AUT does not
affect the far-field amplitude pattern or measured gain. The
phasefront of the antenna can be measured at any distance,
including in the far field. The only requirement is that all
significant emissions from the AUT are sampled at a suitable
density. The amplitude invariance with distance is due to the
conservation of the total energy since all energy
emitted by the antenna is sampled by the near field probe.
An application of Parseval's theorem indicates that the total
power is conserved in the near to far-field transform.
8. Probe correction is not required for on-axis gain
comparison measurements.
The probe correction only affects gain at off-axis angles.
Most antenna gain measurements are made on boresight, requiring
no probe correction. Multibeam antenna gain measurements usually
require probe correction. Probe correction is, however, required
for directivity measurements.
9. Good
nearfield data can be taken with a low receiver signal to
noise ratio.
The far-field transform has a large amount of process
gain resulting in the suppression of random noise. Patterns
can be successfully produced from near field measurements with the receiver
operating at a signal-to-noise ratio below unity and with
what appears to be random phase information.
As an example, assume a pattern has 16,384 points and
the receiver is operating at unity signal to noise ratio. The
receiver is operating in a linear range so that the principle of superposition
holds. The receiver output is the sum of the random thermal
noise and the coherent phasefront. The receiver amplitude
and phase measurements are converted from polar form into
a complex IQ form and then transformed into an angular
spectrum by a Fourier integral. The Fourier integral results
in the summation of all 16,384 points. The integration process
is equivalent to reducing the receiver IF bandwidth. The improvement in
signal to noise ratio, called process gain, is equal to the square root
of 16,384 or 128. This is equivalent to a noise reduction of 42 dB, indicating
that the far field pattern will have a sidelobe noise level of -42 dB relative
to the beam peak when the receiver signal to noise ratio is 0 dB.
10. The antenna pattern, beam pointing and gain are not
affected by the X,Y or Z position of the scan plane, cylinder
or sphere as long as all significant energy is measured
and no aliasing occurs.
This is a direct consequence of the spatial invariance
of the Fourier transform.
11.
Some far-field transformation programs have significant non-linearities.
The transformed far-field measurements are corrupted by non-linearities
anywhere in the processing chain. The primary non-linearities are usually
in the transformation program or the receiver. The non-linearities in the
transformation process are usually caused by non band limited interpolation.
An example of a non-linearity occurs when the angular
spectrum is interpolated incorrectly. The angular spectrum
is band limited, and if the interpolation is not band limited, aliasing
will occur. This effect would cause sidelobes and complex autotrack patterns
to become distorted.
12.
The phase reference cable will not provide better phase stability
when used at a subharmonic.
A common misconception is that lower phase stability is
needed when
the phase reference cable carries a subharmonic of the
test frequency.
Subharmonics are often used to drive varactor frequency
multipliers and harmonic mixers. As an example, assume a test
frequency of 10 GHz with a subharmonic frequency of 1 GHz passing through
the cable. If the cable stretches by 0.01 inches, the phase
will change by approximately 3 degrees at 10 GHz and 0.3 degrees
at 1 GHz. A x10 frequency multiplier, whether external or as
part of a harmonic mixer, will change the 0.3 degree phaseshift
to 3 degrees.
13. Generally only one phase stable cable that flexes
is needed to connect the probe antenna to the other RF components.
When the probe transmits, only one cable is required.
When the probe receives, usually the LO and IF signals need
to be carried to and from the probe. Usually the IF frequency is relatively
low. Therefore, time delay variations in the IF signal line will not have
a significant effect. If the IF signal frequency is a significant portion
of the RF
test frequency, then the IF signal line will also need
to become phase stable. The LO line always needs to be
phase stable.