Gentlemen,

If a man was of a mind to try to get some approximate antenna gain
comparisons, how many wavelengths distant might you like to separate the
antennas?

The proposed scenario is this: make a pair of 2M dipoles, one for
reference, one for receive. I was planning on using the local high
school football field, which is on the order of 50 wl, give or take.
Transmit a few mW at the design frequency, measure the signal strength,
then repeat with an alternate antenna, say a j-pole, collinear, or
something else.

Now, this leaves out a whole bunch of useful information, that would be
tough for me to measure, like spherical gain distribution, etc. I'm
hoping for a figure of merit for the actual implementation of the tested
antenna. (Which, as you can imagine, I could model and save myself the
aggravation.)

I was pondering all this, when it occurred to me that I could not
easily determine when I get to the point where the square law behavior
dominates. I've seen a couple of equations relating the antenna
dimension to wavelength, but I must be really stupid today, because it's
just not sinking in.

Anyone care to comment?

73,
Steve
W1KF

On 31 jul, 19:33, Steve Reinhardt
wrote:
Gentlemen,

If a man was of a mind to try to get some approximate antenna gain
comparisons, how many wavelengths distant might you like to separate the
antennas?

The proposed scenario is this: make a pair of 2M dipoles, one for
reference, one for receive. I was planning on using the local high
school football field, which is on the order of 50 wl, give or take.
Transmit a few mW at the design frequency, measure the signal strength,
then repeat with an alternate antenna, say a j-pole, collinear, or
something else.

Now, this leaves out a whole bunch of useful information, that would be
tough for me to measure, like spherical gain distribution, etc. I'm
hoping for a figure of merit for the actual implementation of the tested
antenna. (Which, as you can imagine, I could model and save myself the
aggravation.)

I was pondering all this, when it occurred to me that I could not
easily determine when I get to the point where the square law behavior
dominates. I've seen a couple of equations relating the antenna
dimension to wavelength, but I must be really stupid today, because it's
just not sinking in.

Anyone care to comment?

73,
Steve
W1KF

Hello Steve,

When your "Antennas Under Test" are moderate gain devices, I would go
for several wavelengths. For low to moderate gain (up to 10 dBi), you
are in the far field within about 4 WL.

The reason for the short distance is that the direct signal is strong,
hence influence of reflections is less. You can reduce the effect of
reflections by taking a receive antenna with some directivity.

You can be sure that you are in the far field distance when
D 2*B^2/lambda, where B = overall size of the antenna (from one
extremity to another). For several antenna types (like yagis), you can
halve this distance when you are interested in main beam gain only.

Best regards,

Wim
PA3DJS
www.tetech.nl

Wimpie wrote:
On 31 jul, 19:33, Steve Reinhardt
wrote:

Gentlemen,

If a man was of a mind to try to get some approximate antenna gain
comparisons, how many wavelengths distant might you like to separate the
antennas?

The proposed scenario is this: make a pair of 2M dipoles, one for
reference, one for receive. I was planning on using the local high
school football field, which is on the order of 50 wl, give or take.
Transmit a few mW at the design frequency, measure the signal strength,
then repeat with an alternate antenna, say a j-pole, collinear, or
something else.

Now, this leaves out a whole bunch of useful information, that would be
tough for me to measure, like spherical gain distribution, etc. I'm
hoping for a figure of merit for the actual implementation of the tested
antenna. (Which, as you can imagine, I could model and save myself the
aggravation.)

I was pondering all this, when it occurred to me that I could not
easily determine when I get to the point where the square law behavior
dominates. I've seen a couple of equations relating the antenna
dimension to wavelength, but I must be really stupid today, because it's
just not sinking in.

Anyone care to comment?

73,
Steve
W1KF


Hello Steve,

When your "Antennas Under Test" are moderate gain devices, I would go
for several wavelengths. For low to moderate gain (up to 10 dBi), you
are in the far field within about 4 WL.

The reason for the short distance is that the direct signal is strong,
hence influence of reflections is less. You can reduce the effect of
reflections by taking a receive antenna with some directivity.

You can be sure that you are in the far field distance when
D 2*B^2/lambda, where B = overall size of the antenna (from one
extremity to another). For several antenna types (like yagis), you can
halve this distance when you are interested in main beam gain only.


This formula is actually an embodiment of the venerable Rayleigh limit,

It actually says that wavefront is flat to within a fraction of a
wavelength (about 1/13th or 22 degrees). The implications for gain
measurement is that the gain you measure at 2*b^2/lambda distance will
be the same as you'd measure if you were truly in the far field, to
within about a reasonable degree of accuracy (1% or there abouts).

The derivation is this:
distance to center of antenna = D
distance to edge of antenna Dedge = sqrt(D^2+(B/2)^2) {Pythagorean formula}

phase error = (D-Dedge)/lambda {wavelengths}

etc.


if you start getting lambda close to B, then the relative path length
difference gets quite large, and you have to start worrying about the
current distribution or illumination non-uniformity.

Jim Lux wrote:

When your "Antennas Under Test" are moderate gain devices, I would go
for several wavelengths. For low to moderate gain (up to 10 dBi), you
are in the far field within about 4 WL.

The reason for the short distance is that the direct signal is strong,
hence influence of reflections is less. You can reduce the effect of
reflections by taking a receive antenna with some directivity.

You can be sure that you are in the far field distance when
D 2*B^2/lambda, where B = overall size of the antenna (from one
extremity to another). For several antenna types (like yagis), you can
halve this distance when you are interested in main beam gain only.


This formula is actually an embodiment of the venerable Rayleigh limit,

It actually says that wavefront is flat to within a fraction of a
wavelength (about 1/13th or 22 degrees). The implications for gain
measurement is that the gain you measure at 2*b^2/lambda distance will
be the same as you'd measure if you were truly in the far field, to
within about a reasonable degree of accuracy (1% or there abouts).

The derivation is this:
distance to center of antenna = D
distance to edge of antenna Dedge = sqrt(D^2+(B/2)^2) {Pythagorean
formula}

phase error = (D-Dedge)/lambda {wavelengths}

etc.


if you start getting lambda close to B, then the relative path length
difference gets quite large, and you have to start worrying about the
current distribution or illumination non-uniformity.

This is a big help! The equations I read did not help me understand the
problem. (Though when I read 'Rayleigh', thoughts of optical flats and
oversized college physics texts popped into my head.)

So, if I have a 4 element collinear, measuring 2 wl, or about 4 meters,
and the frequency of interest is about 2 meters, then I'm effectively
far field when I reach a distance 16 wl.

Cool. The neat part about the football field is that the nearest
reflection is well over 1.4 times the distance between source and
measurement antennae. It's flat with no RF hard surfaces around the
perimeter. That's not to say there are no other sources of measurement
error, just that I think their contribution will be small.

I'll report back if I can get it done before school starts, and they
want my RF range back for their sports activities :-)

73,
Steve
W1KF

Steve Reinhardt wrote:
. . .
Cool. The neat part about the football field is that the nearest
reflection is well over 1.4 times the distance between source and
measurement antennae. It's flat with no RF hard surfaces around the
perimeter. That's not to say there are no other sources of measurement
error, just that I think their contribution will be small.
. . .

I'm not at all an expert on antenna measurement. I know just enough to
realize that it's extremely difficult to do with even moderate accuracy,
and that some professionals with the very best equipment tend to trust
modeling more than measurement. Besides problems with reflections, you
also have the problem of assuring a constant real power into antennas of
different impedances, feedline radiation, and a host of other
confounding factors.

That being said, I'm sure you'll learn a lot in the process, and you
might be able to get useful results in spite of the difficulties.

I was really saddened to see, some years ago, a published group of
measurements like you're proposing, on a bunch of different antennas.
The results, while quite believable, showed some pattern skewing and
other artifacts which almost certainly weren't really due to the
antennas themselves. What saddened me was that the people running this
detailed, meticulous, and time consuming project hadn't thought to
include a measurement of any antenna with a well known pattern and gain
such as a dipole. Please include some reference antennas which are well
known and/or easy to model! Otherwise, the accuracy of all the results
is purely guesswork. A dipole might not be the best reference because
its broad pattern is more subject to reflections than a Yagi. But a good
reference Yagi (such as one of the NBS standard Yagis) or two could be
constructed and included.

You'll get strong reflections from the ground between the antennas.
It'll be easy to calculate the elevation angle of the signal from the
transmitting antenna which will arrive at the receiving antenna after
reflecting from the ground. If your antennas are far apart, this angle
will be more nearly horizontal, where antenna gain is higher, than if
they're close. So there may will be a stronger reflection if the
antennas are farther rather than closer (although most moderately sized
horizontal Yagis have a broad elevation pattern, not very different from
a dipole). This ground reflection could have a profound effect on
measured gain, and could cause some large differences with only small
differences in, say, antenna height. If the antennas are close and the
angle steep, then the pattern at the elevation angle being reflected
might be quite different from the horizontal antenna pattern. This will
result in a distortion of the measured pattern relative to the free
space pattern. This is something which will show up more clearly in a
plot of a reference Yagi with gain closer to the actual antenna under
test than something like a dipole with a simpler pattern.

Have fun! I'll be eager to see how well your reference antenna
measurements come out.

Roy Lewallen, W7EL

"Steve Reinhardt" wrote in message
...
Gentlemen,

If a man was of a mind to try to get some approximate antenna gain
comparisons, how many wavelengths distant might you like to separate the
antennas?

The proposed scenario is this: make a pair of 2M dipoles, one for
reference, one for receive. I was planning on using the local high school
football field, which is on the order of 50 wl, give or take. Transmit a
few mW at the design frequency, measure the signal strength, then repeat
with an alternate antenna, say a j-pole, collinear, or something else.

Now, this leaves out a whole bunch of useful information, that would be
tough for me to measure, like spherical gain distribution, etc. I'm hoping
for a figure of merit for the actual implementation of the tested antenna.
(Which, as you can imagine, I could model and save myself the
aggravation.)

I was pondering all this, when it occurred to me that I could not easily
determine when I get to the point where the square law behavior dominates.
I've seen a couple of equations relating the antenna dimension to
wavelength, but I must be really stupid today, because it's just not
sinking in.

Anyone care to comment?

73,
Steve
W1KF

Hi Steve

It is fairly easy to record exact radiation patterns of "2 meter" antennas
at 137 MHz, using NOAA satellites at the Illuminator. If you have any
interest in the details, you can contact me anytime.
Patrik Tast developed a (free) program for me that produces elevation
plane patterns of the antenna as the NOAA satellite passes over. Since each
satellite passes over 6 or 8 times per day, decent hemispheric patterns can
be made.
I'd guess that the plot of the antenna pattern, when using Patrik's
program is more accurate than any other method when evaluating ground based
"2 meter" antennas. I am open to learning where I'm wrong about the
accuracy.

Jerry

"Jerry Martes" wrote in message
news:coTri.7566$yg1.763@trnddc04...

"Steve Reinhardt" wrote in message
...
Gentlemen,

If a man was of a mind to try to get some approximate antenna gain
comparisons, how many wavelengths distant might you like to separate the
antennas?

The proposed scenario is this: make a pair of 2M dipoles, one for
reference, one for receive. I was planning on using the local high school
football field, which is on the order of 50 wl, give or take. Transmit a
few mW at the design frequency, measure the signal strength, then repeat
with an alternate antenna, say a j-pole, collinear, or something else.

Now, this leaves out a whole bunch of useful information, that would be
tough for me to measure, like spherical gain distribution, etc. I'm
hoping for a figure of merit for the actual implementation of the tested
antenna. (Which, as you can imagine, I could model and save myself the
aggravation.)

I was pondering all this, when it occurred to me that I could not easily
determine when I get to the point where the square law behavior
dominates. I've seen a couple of equations relating the antenna dimension
to wavelength, but I must be really stupid today, because it's just not
sinking in.

Anyone care to comment?

73,
Steve
W1KF

Hi Steve

It is fairly easy to record exact radiation patterns of "2 meter"
antennas at 137 MHz, using NOAA satellites at the Illuminator. If you
have any interest in the details, you can contact me anytime.
Patrik Tast developed a (free) program for me that produces elevation
plane patterns of the antenna as the NOAA satellite passes over. Since
each satellite passes over 6 or 8 times per day, decent hemispheric
patterns can be made.
I'd guess that the plot of the antenna pattern, when using Patrik's
program is more accurate than any other method when evaluating ground
based "2 meter" antennas. I am open to learning where I'm wrong about
the accuracy.

Jerry

Examples of the radiation pattern data that can be acquired with Patrik
Tast's SignalPlotter program can be seen on one of his sites
http://213.250.83.83/~patrik/apt/log...22-2007/daily/

I have used this SignalPlotter program to make radiation pattern plots by
recording rssi voltage with a simple voltmeter with a RS232 connection and
with LabJack data recorder.

Jerry

On 1 ago, 02:39, Steve Reinhardt
wrote:
Jim Lux wrote:
When your "Antennas Under Test" are moderate gain devices, I would go
for several wavelengths. For low to moderate gain (up to 10 dBi), you
are in the far field within about 4 WL.

The reason for the short distance is that the direct signal is strong,
hence influence of reflections is less. You can reduce the effect of
reflections by taking a receive antenna with some directivity.

You can be sure that you are in the far field distance when
D 2*B^2/lambda, where B = overall size of the antenna (from one
extremity to another). For several antenna types (like yagis), you can
halve this distance when you are interested in main beam gain only.

This formula is actually an embodiment of the venerable Rayleigh limit,

It actually says that wavefront is flat to within a fraction of a
wavelength (about 1/13th or 22 degrees). The implications for gain
measurement is that the gain you measure at 2*b^2/lambda distance will
be the same as you'd measure if you were truly in the far field, to
within about a reasonable degree of accuracy (1% or there abouts).

The derivation is this:
distance to center of antenna = D
distance to edge of antenna Dedge = sqrt(D^2+(B/2)^2) {Pythagorean
formula}

phase error = (D-Dedge)/lambda {wavelengths}

etc.

if you start getting lambda close to B, then the relative path length
difference gets quite large, and you have to start worrying about the
current distribution or illumination non-uniformity.

This is a big help! The equations I read did not help me understand the
problem. (Though when I read 'Rayleigh', thoughts of optical flats and
oversized college physics texts popped into my head.)

So, if I have a 4 element collinear, measuring 2 wl, or about 4 meters,
and the frequency of interest is about 2 meters, then I'm effectively
far field when I reach a distance 16 wl.

Cool. The neat part about the football field is that the nearest
reflection is well over 1.4 times the distance between source and
measurement antennae. It's flat with no RF hard surfaces around the
perimeter. That's not to say there are no other sources of measurement
error, just that I think their contribution will be small.

I'll report back if I can get it done before school starts, and they
want my RF range back for their sports activities :-)

73,
Steve
W1KF

Hello Steve,

The result of the 2*B^2/lambda formula is in meters. For your antenna
with 4m size, you can be sure to be in the far field at 16m (AKA
Fraunhofer region). The minimum distance must be the sum of the far
field distance for both AUT and reference antenna.

As Roy also mentioned, real measurements are difficult. To play your
own devil's advocate, you could run gain measurements for different
(many) distances. From the Gain versus Distance graph you can get
an impression of the accuracy. The same you can do for various heights
to get an impression of the influence of ground reflection.

A complete other approach is to include the ground reflection in the
measurement. That might no be a bad option. The requirement for such a
measurement setup is that for both receive and transmit antennas, the
ground reflection must fall well within the main beam. Horizontal
polarization is preferred. VSWR of both antennas must not be affected
by ground reflection.

Best regards,

Wim
PA3DJS
www.tetech.nl

Roy Lewallen wrote:
Steve Reinhardt wrote:
. . .
Cool. The neat part about the football field is that the nearest
reflection is well over 1.4 times the distance between source and
measurement antennae. It's flat with no RF hard surfaces around the
perimeter. That's not to say there are no other sources of measurement
error, just that I think their contribution will be small.
. . .

I'm not at all an expert on antenna measurement. I know just enough to
realize that it's extremely difficult to do with even moderate accuracy,
and that some professionals with the very best equipment tend to trust
modeling more than measurement. Besides problems with reflections, you
also have the problem of assuring a constant real power into antennas of
different impedances, feedline radiation, and a host of other
confounding factors.

That being said, I'm sure you'll learn a lot in the process, and you
might be able to get useful results in spite of the difficulties.

I was really saddened to see, some years ago, a published group of
measurements like you're proposing, on a bunch of different antennas.
The results, while quite believable, showed some pattern skewing and
other artifacts which almost certainly weren't really due to the
antennas themselves. What saddened me was that the people running this
detailed, meticulous, and time consuming project hadn't thought to
include a measurement of any antenna with a well known pattern and gain
such as a dipole. Please include some reference antennas which are well
known and/or easy to model! Otherwise, the accuracy of all the results
is purely guesswork. A dipole might not be the best reference because
its broad pattern is more subject to reflections than a Yagi. But a good
reference Yagi (such as one of the NBS standard Yagis) or two could be
constructed and included.


Well, as I mentioned, this is less about an absolute antenna gain than a
figure of merit. Using a dipole as the first, reference transmitting
antenna is part of the plan. I may be crazy, but I'm not entirely stupid
;-) (Well, maybe. Time will tell...)

You'll get strong reflections from the ground between the antennas.
It'll be easy to calculate the elevation angle of the signal from the
transmitting antenna which will arrive at the receiving antenna after
reflecting from the ground. If your antennas are far apart, this angle
will be more nearly horizontal, where antenna gain is higher, than if
they're close. So there may will be a stronger reflection if the
antennas are farther rather than closer (although most moderately sized
horizontal Yagis have a broad elevation pattern, not very different from
a dipole). This ground reflection could have a profound effect on
measured gain, and could cause some large differences with only small
differences in, say, antenna height. If the antennas are close and the
angle steep, then the pattern at the elevation angle being reflected
might be quite different from the horizontal antenna pattern. This will
result in a distortion of the measured pattern relative to the free
space pattern. This is something which will show up more clearly in a
plot of a reference Yagi with gain closer to the actual antenna under
test than something like a dipole with a simpler pattern.

Ah, yet another thing I have to consider. Since at least two of the
tested antennas will be primarily vertical, I was planning to make the
test antennas all vertical. So, I can talk myself into believing the
ground reflections are part of the real world installations, or I can
chuck it all and rely solely upon modeling. One is probably smarter, the
other more viscerally stimulating. I leave to the reader to sort out
which is which.

Have fun! I'll be eager to see how well your reference antenna
measurements come out.

Roy Lewallen, W7EL

Thanks for the guidance. The journey may be far more interesting that
the result!

Wimpie wrote:
On 1 ago, 02:39, Steve Reinhardt
wrote:
--snip--

wavelength (about 1/13th or 22 degrees). The implications for gain
measurement is that the gain you measure at 2*b^2/lambda distance will
be the same as you'd measure if you were truly in the far field, to
within about a reasonable degree of accuracy (1% or there abouts).
The derivation is this:
distance to center of antenna = D
distance to edge of antenna Dedge = sqrt(D^2+(B/2)^2) {Pythagorean
formula}
phase error = (D-Dedge)/lambda {wavelengths}
etc.
if you start getting lambda close to B, then the relative path length
difference gets quite large, and you have to start worrying about the
current distribution or illumination non-uniformity.
This is a big help! The equations I read did not help me understand the
problem. (Though when I read 'Rayleigh', thoughts of optical flats and
oversized college physics texts popped into my head.)

So, if I have a 4 element collinear, measuring 2 wl, or about 4 meters,
and the frequency of interest is about 2 meters, then I'm effectively
far field when I reach a distance 16 wl.

Cool. The neat part about the football field is that the nearest
reflection is well over 1.4 times the distance between source and
measurement antennae. It's flat with no RF hard surfaces around the
perimeter. That's not to say there are no other sources of measurement
error, just that I think their contribution will be small.

I'll report back if I can get it done before school starts, and they
want my RF range back for their sports activities :-)

73,
Steve
W1KF

Hello Steve,

The result of the 2*B^2/lambda formula is in meters. For your antenna
with 4m size, you can be sure to be in the far field at 16m (AKA
Fraunhofer region). The minimum distance must be the sum of the far
field distance for both AUT and reference antenna.

As Roy also mentioned, real measurements are difficult. To play your
own devil's advocate, you could run gain measurements for different
(many) distances. From the Gain versus Distance graph you can get
an impression of the accuracy. The same you can do for various heights
to get an impression of the influence of ground reflection.

A complete other approach is to include the ground reflection in the
measurement. That might no be a bad option. The requirement for such a
measurement setup is that for both receive and transmit antennas, the
ground reflection must fall well within the main beam. Horizontal
polarization is preferred. VSWR of both antennas must not be affected
by ground reflection.

Best regards,

Wim
PA3DJS
www.tetech.nl


Wim,

Thank you. I really don't know where my head is at lately. Of course,
the square term means the result will have units of length, not wavelength.

As I mentioned in my reply to Roy, the tested antennas are designed to
be vertical, which may play havoc with the whole idea.

More to follow, and thanks for reminding me my basic math skills have
ducked into a hole for the moment!

73,
Steve
W1KF