On Sun, 24 Apr 2005 05:44:34 +0000 (UTC), "Reg Edwards"
wrote:

All electrical calibration and testing laboratories issue tables of
claimed accuracies of measurements. Measurement uncertainties stated
on calibration certificates are legally binding. All stated
measurement results must be traceable to International Standards or a
laboratory or testing station loses its status.

Consequently there is no incentive for a laboratory to overstate its
capabilities in its sales literature. Indeed, it is dangerous,
illegal even!

Naturally, laboratories can differ widely, one from another.

It would be interesting to compare laboratory uncertainties with
performance figures claimed by antenna manufacturers. Or anyone else.

Does anyone have typical examples of measurement uncertainties claimed
by antenna testing stations? Answers in decibels please.

A reply from a testing station, at HF or VHF, would be specially
appreciated.

As stated by Ian, there's no simple answer. The bane of antenna
testing is reflections reflections reflections.

It may come as a surprise to our correspondent who likes to disparage
"gurus" that "standard-gain" antennas are widely used as reference
standards. To head off the question of how the standard gain is
determined, that is done by testing three "identical" antennas in
pairs; each one against the other two, with one the source and the
other the receiver. A bit of algebra and you have the gain of each
one individually.

http://www.mi-technologies.com/literature/a00-044.pdf

The foregoing paper might help answer Reg's question about achievable
accuracy.

While not addressing hf and vhf measurements, some of the following
might be of interest.

Indoor measurements are usually conducted in anechoic chambers where
the shape is often tapered to control reflections and the walls are
covered in absorber material. A chamber will have a "quiet zone"
where the reflections are specified to be X db down. Very often the
antennas under test are being characterized for side lobe levels or in
the case of monopulse radar, the null depth of the difference
pattern(s). If you're trying to measure a 60 dB null, it doesn't pay
to have a quiet zone of -40 dB.

These measurements also require an amplitude and phase front that
mimics a source at infinite distance. This used to require huge
chambers, often hundreds of feet long. A new way to accomplish this
is to "fold" the range by using specially shaped reflectors to flatten
the amplitude/phase across the test aperature. This has the added
benefit of shorter cables between sources, DUT and measurement
receiver. At X and K band, cable loss can be a killer. Likewise
moving cables around and even temperature changes can affect the
measurments.

I have used such a range to measure antennas from L to Ka band.

Outdoor ranges often "feature" the ground reflection, since it is
difficult to eliminate it physically. This is particularly true at
hf/vhf. I have used a technique that utilized the time-domain
capability of a modern network analyzer (HP-8510) to identify the
reflection and then place absorber material to attenuate it.
Similarly, a frequency-domain measurement, that includes ground
reflection, can be transformed to the time domain where the reflection
is gated out and then transformed back to the frequency domain for
"reflection free" analysis.

See also:

http://www.lehman-inc.com/pdf/mag2.pdf

Wes,
What you have posted is very interesting and is not spewing out alot of
stuff
regarding isentropic gain etc that is really not relevent to an actual
testing range.
Rather than deflect away from Reg's needs may I go back to the "compared to
a dipole"
statement which Richard keeps brushing off. If the gains are different then
the
angle for max radiation is different and if you do not take this into
account by searching
for the individual point of maximum gain position then the the measurements
are in total error.
To put antennas at the same height and then measuring at the same stationary
point for receive,
switching back and forth
is not a true comparison because of the different elevation angles. If one
was to compare a long yagi
to a dipole ando make it a true comparison measurement one must surely take
into account the two degree
or so difference when positioning the listening posts and not relying on a
single
listening position which to me appears to be a NO No .
Richards response to the "error" question totally ignored TOA saying they
are usually
the same . He also ignored what he considered as an "equal" height for the
curtain,
i.e the top,bottom or the center line of the curtain array which alone would
introduce
error with respect to comparible measurement. If Richard was pointing out
that his was a typical
professional method of measurement then I would view his statement in
complete disbelief.
Your posting, thankyou, confirms my thinking in that the use of a dipole
only confirms the reliability
of the set up used and that is the end of it with respect to measurement of
a competing antenna
where I suspect a pro lab would identify the particular resulting elevation
measurement.
If the last sentence is in error I would apreciate a correction
Regards
Art


"Wes Stewart" wrote in message
...
On Sun, 24 Apr 2005 05:44:34 +0000 (UTC), "Reg Edwards"
wrote:

All electrical calibration and testing laboratories issue tables of
claimed accuracies of measurements. Measurement uncertainties stated
on calibration certificates are legally binding. All stated
measurement results must be traceable to International Standards or a
laboratory or testing station loses its status.

Consequently there is no incentive for a laboratory to overstate its
capabilities in its sales literature. Indeed, it is dangerous,
illegal even!

Naturally, laboratories can differ widely, one from another.

It would be interesting to compare laboratory uncertainties with
performance figures claimed by antenna manufacturers. Or anyone else.

Does anyone have typical examples of measurement uncertainties claimed
by antenna testing stations? Answers in decibels please.

A reply from a testing station, at HF or VHF, would be specially
appreciated.

As stated by Ian, there's no simple answer. The bane of antenna
testing is reflections reflections reflections.

It may come as a surprise to our correspondent who likes to disparage
"gurus" that "standard-gain" antennas are widely used as reference
standards. To head off the question of how the standard gain is
determined, that is done by testing three "identical" antennas in
pairs; each one against the other two, with one the source and the
other the receiver. A bit of algebra and you have the gain of each
one individually.

http://www.mi-technologies.com/literature/a00-044.pdf

The foregoing paper might help answer Reg's question about achievable
accuracy.

While not addressing hf and vhf measurements, some of the following
might be of interest.

Indoor measurements are usually conducted in anechoic chambers where
the shape is often tapered to control reflections and the walls are
covered in absorber material. A chamber will have a "quiet zone"
where the reflections are specified to be X db down. Very often the
antennas under test are being characterized for side lobe levels or in
the case of monopulse radar, the null depth of the difference
pattern(s). If you're trying to measure a 60 dB null, it doesn't pay
to have a quiet zone of -40 dB.

These measurements also require an amplitude and phase front that
mimics a source at infinite distance. This used to require huge
chambers, often hundreds of feet long. A new way to accomplish this
is to "fold" the range by using specially shaped reflectors to flatten
the amplitude/phase across the test aperature. This has the added
benefit of shorter cables between sources, DUT and measurement
receiver. At X and K band, cable loss can be a killer. Likewise
moving cables around and even temperature changes can affect the
measurments.

I have used such a range to measure antennas from L to Ka band.

Outdoor ranges often "feature" the ground reflection, since it is
difficult to eliminate it physically. This is particularly true at
hf/vhf. I have used a technique that utilized the time-domain
capability of a modern network analyzer (HP-8510) to identify the
reflection and then place absorber material to attenuate it.
Similarly, a frequency-domain measurement, that includes ground
reflection, can be transformed to the time domain where the reflection
is gated out and then transformed back to the frequency domain for
"reflection free" analysis.

See also:

http://www.lehman-inc.com/pdf/mag2.pdf

" wrote in message
news:u9_ae.18115$NU4.14900@attbi_s22...
Richard,
You state that you used a dipole to compare with, which was at the same
height !.
Which antenna was altered so that the elevation angle of maximum gain
was the same for both antennas.such that max gain measurements
were truly comparable?
Where was the height of the "curtain" measured or referred to
so that "same height" could be justified ?
( You also did say it was for SW use which is certainly different
to ground wave use)
Presumably, the comparison was for the same type of polarization
and ignored differences created by the side addition of other
types of polarization.
Without further information the "Facts" could be seen as
correct to plus or minus 100 percent measurement error!



And that sums up most antenna testing rather well!

--
Ed
WB6WSN
El Cajon, CA USA

Art Unwin wrote:
"Richard`s response to the "error" question totally ignored TOA saying
they are usually the same."

Propagation dictates the take off angle that the signal actually follows
regardless of what your antennas do. We made meadurements on different
days so that propagation may have been different on different days. We
were checking over nearly the actual paths under what might be typical
conditions. Did the curtain produce louder signals? You bet!

Even though the curtain antenna had sharper vertical directivity as well
as sharper horizontal directivity than the lone dipole, these were the
goals of the design. Produce more signal on target to try to overcome
the myriad of jammers that were trying to drown us out.

During our tests, the paths between transmitter and the receivers were
the same in most cases. The width of a curtain was only about one
wavelength and the dipole was immediately adjacent to the curtain. The
curtain was two dipoles high, two dipoles wide and two dipoles deep as I
recall. Those dipoles in front were all driven in phase. Those behind
were tuned parasitic reflectors. It wasn`t unique at all. I`ve seen many
since then which look very much like our curtains. They were well
behaved and brought in lots of fan mail. They obviously radiated ok. The
reflectors seemed to shield the villiage behind them from being drowned
in radio frequency energy.

Whatever differences there may have been between the conditions imposed
on the dipole and curtain, they were tuned and loaded for the same
transmitted power. Received signal differences were likely due to gain
in the curtain versus gain in the dipole. Averiging a large number of
samples likely straightened out inevitable minor differences. I would
wager our results were good enough.
My employer was satisfied and all the contractors got paid.

Best regards, Richard Harrison, KB5WZI

"Frank" wrote in message
news:1V8be.56318$yV3.14588@clgrps12...

"Ed Price" wrote in message
news:H_Vae.2007$pk5.904@fed1read02...

Everbody loves to argue about antennas; their calibration, application &
accuracy! In the EMC area (my side of the elephant), we are frequently
looking for emissions with a maximum limit so low (imposed by the
standard) that we have to be inside a shielded enclosure. Since the cost
of a chamber increases as the square (or maybe the cube) of its volume,
only extraordinarily well-funded (uhh, governmental) labs can afford
really huge chambers. Thus, most EMC testing happens in more modest
volumes (my chamber is 36' x 24' x 9').

Last place I worked with EMC facilities they only had a 3 m cube chamber.
The dimensions you quoted are huge compared to my experience. (I think
ETC, in Airdrie Alberta, had a similar chamber to yours; also General
Dynamics in Calgary had two similar chambers. Also Nortel has some EMC
capabiltiy.) The insides were covered in microwave absorber, and there was
some question as to how effective the absorber was at 30 MHz. It must
have done something, since before the absorber was installed it was
interesting to see the effects on a transmitter keyed inside a shielded
enclosure.

The MIL-STD-461E requirement for absorbed is a 10 dB return loss at 250 MHz.
I have 24" tall pyramidal foam, and that meets the requirement. As frequency
decreases, the foam essentially disappears. By 10 MHz, it has almost no
effect. The pyramidal foam is expensive, about $50 / sq ft. If you want more
return loss, you need taller pyramids; those mythical governmental labs have
had foam up to 72" tall (and the wall absorbers tend to droop a bit g).

A newer technique is to use ferrite tiles, especially on the floor. They are
less than a half-inch thick, and perform much better at low frequencies. And
the cost is about $100 / sq ft. I like to think of my walls and ceiling as
covered with $5 bills, and the floor carpeted with $10's.

Your anechoic chamber is never really perfect; however, it becomes "good
enough" when you run out of money.

With the dark blue pyramids and black tiles, a chamber looks like a bat
cave. One vendor decided that the new millenia needed white paint on the
foam; another vendor touts pyramids that have a 90-degree axial rotation
part way up the taper, and yet another truncates the pointy tips, telling us
that works better. It's just like the antenna game.

Here's what MIL-STD-461E says about conical logarithmic spiral antennas:

"Previous versions of this standard specified conical log spiral
antennas. These antennas were convenient since they did not need to be
rotated to measure both polarizations of the radiated field. The double
ridged horn is considered to be better for standardization for several
reasons.

Very interesting Ed, will forward your comments to my last company. Doubt
they will do anything tho, as they never want to spend any money. Assume
the recomended type of antenna is a linearly polarized log periodic.

No, 461 doesn't like log periodics either, saying:

"Other linearly polarized antennas such as log periodic antennas are not to
be used. It is recognized that these types of antennas have sometimes been
used in the past; however, they will not necessarily produce the same
results as the double ridged horn because of field variations across the
antenna apertures and far field/near field issues. Uniform use of the double
ridge horn is required for standardization purposes to obtain consistent
results among different test facilities."

The MIL-STD defines a 104 cm rod from 10 kHz to 30 MHz, then a biconical
from 30 MHz to 200 MHz, and finally, horns above there. Since pyramidal
horns are only good for about an octave, a smart Navy guy added
exponentially flared ridges to the horns, and came up with multi-octave
horns. A typical horn for 200 MHz to 1 GHz has an aperture of about 1 meter,
then another horn tries to go from 1 GHz to 18 GHz. That's a bit too far for
me, as the antenna factor really climbs above about 14 GHz, so I switch to a
common, non-ridged horn for 12 GHz to 18 GHz. For 18 GHz to 26 GHz and 26
GHz to 40 GHz, I use standard-gain flared horns. With a pre-selected
spectrum analyzer, really good coax, and a couple of low-noise pre-amps,
that lets me get comfortably below the most stringent RE102 limits.


I remember the Singer (Was it Singer-Metrics), and using it to measure
radiated spurious in a cow pasture at 50 m from a 1kW TMC linear
(Canadian Marconi, Montreal). The test monopole had a cylindrical base
with a rotary switch.

OK, just for trivia's sake. If the antenna base was cylindrical, painted
grey crinkle, had a 6-position range switch and a brown bakelite top
insulator, it was an Empire VA-105. But, if it was almost a cube, painted
battleship grey, had a black front panel and an 8-position range switch, it
was a Stoddart 92138-1 (that number is a hazy memory). Both were passive
antennas. The Empire was used with the NF-105 receiver, while the Stoddart
antenna was associated with the NM-22A (that's why the range switches were
different, to match the ranges on their associated receivers).


--
Ed
WB6WSN
El Cajon, CA USA

Richard, it is now quite clear that you were not undertaking a test
referenced to a dipole. All you were doing is confirming a target area
under average conditions to ensure the language used was compatable
to the target area.....Period
More important to me is your statement that :

" Propagation dictates the take off angle that the signal actually
follows regardless of what your antennas do"

This statement seems to echo a conclusion arrived at by a regular poster
( I should call him a guru) on this group tho leaving me unconvinced.
Would you kindly point out to me what book you are extracting this
statement from so I may examine the boundaries under which that
statement is deemed correct?
Thanking you in advance
Art


"Richard Harrison" wrote in message
...
Art Unwin wrote:
"Richard`s response to the "error" question totally ignored TOA saying
they are usually the same."

Propagation dictates the take off angle that the signal actually follows
regardless of what your antennas do. We made meadurements on different
days so that propagation may have been different on different days. We
were checking over nearly the actual paths under what might be typical
conditions. Did the curtain produce louder signals? You bet!

Even though the curtain antenna had sharper vertical directivity as well
as sharper horizontal directivity than the lone dipole, these were the
goals of the design. Produce more signal on target to try to overcome
the myriad of jammers that were trying to drown us out.

During our tests, the paths between transmitter and the receivers were
the same in most cases. The width of a curtain was only about one
wavelength and the dipole was immediately adjacent to the curtain. The
curtain was two dipoles high, two dipoles wide and two dipoles deep as I
recall. Those dipoles in front were all driven in phase. Those behind
were tuned parasitic reflectors. It wasn`t unique at all. I`ve seen many
since then which look very much like our curtains. They were well
behaved and brought in lots of fan mail. They obviously radiated ok. The
reflectors seemed to shield the villiage behind them from being drowned
in radio frequency energy.

Whatever differences there may have been between the conditions imposed
on the dipole and curtain, they were tuned and loaded for the same
transmitted power. Received signal differences were likely due to gain
in the curtain versus gain in the dipole. Averiging a large number of
samples likely straightened out inevitable minor differences. I would
wager our results were good enough.
My employer was satisfied and all the contractors got paid.

Best regards, Richard Harrison, KB5WZI

On Mon, 25 Apr 2005 14:48:38 -0700, Wes Stewart
wrote:

It may come as a surprise to our correspondent who likes to disparage
"gurus" that "standard-gain" antennas are widely used as reference
standards. To head off the question of how the standard gain is
determined, that is done by testing three "identical" antennas in
pairs; each one against the other two, with one the source and the
other the receiver. A bit of algebra and you have the gain of each
one individually.

http://www.mi-technologies.com/literature/a00-044.pdf


Hi All,

The method described by the paper offered above is a commonplace of
Metrology called "Reciprocity." I have calibrated precision
microphones against this method, and the error math offered is
consistent with my experience (much less the actual values offered as
examples).

As an aside, this method is also as old as the pyramids - literally.
The Egyptians planned their blocks of granite to have nearly flat
faces to within 10s of microinches using three blocks, by abrading one
against the other and then rotating their positions.

Accuracy is far more a matter of protocol or technique than it is
about a ruler (or other scale).

73's
Richard Clark, KB7QHC

wrote:
Rather than deflect away from Reg's needs may I go back to the
"compared to a dipole" statement which Richard keeps brushing off. If
the gains are different then the angle for max radiation is different
and if you do not take this into account by searching for the
individual point of maximum gain position then the the measurements are
in total error. To put antennas at the same height and then measuring
at the same stationary
point for receive,
switching back and forth
is not a true comparison because of the different elevation angles.

I don't think Richard is attempting to deny that.

His tests were not intended to measure the gain of the antenna. They
were intended to answer a much more practical question: "How much
stronger is the signal from the curtain array, as delivered into the BC
target area, compared with using a dipole?" That's what the station
owners wanted to know, and they specifically wanted that answer to
include all the variables of antenna patterns and ionospheric
propagation.

As you have correctly pointed out, in any environment except free space,
that number is not the same as the antenna gain in dBd. Anybody who has
thought about it is aware of the problem, and that clearly includes
Richard.

Everybody agrees with you, so you can stop banging on that open door.


--
73 from Ian GM3SEK 'In Practice' columnist for RadCom (RSGB)
http://www.ifwtech.co.uk/g3sek

" wrote about Richard Harrison's post:
Richard, it is now quite clear that you were not undertaking a test
referenced to a dipole. All you were doing is confirming a target area
under average conditions to ensure the language used was compatable
to the target area.....Period
More important to me is your statement that :

" Propagation dictates the take off angle that the signal actually
follows regardless of what your antennas do"
_________

Your arguments arise from trying to compare two different test goals, e.g.,
accurately measuring the free space az/el radiation patterns of an antenna
itself, versus how those radiation patterns may perform in a particular
application (height above ground, ground characteristics, ionospheric
propagation characteristics, reflection sources, target coverage zone, etc).

Classic antenna test ranges are designed to measure the az/el radiation
patterns of antennas themselves, independent of their environment. What
that radiation will provide in terms of a desired "coverage" result is
another matter, and is the responsibility of the RF system designer -- not
the antenna test range.

RF

Visit http://rfry.org for FM transmission system papers.

Thanks again Ed. From everyone of your posts I learn something new.

The MIL-STD-461E requirement for absorbed is a 10 dB return loss at 250
MHz.

Assume you would test the chamber return loss with a tuned dipole having
free space return loss 10dB. i.e. some physically realizable antenna with
a return loss of 40 dB at the test frequency. I suppose, with an inductivly
loaded dipole, you could test the return loss of a 3 m chamber down to 30
MHz. There were some questions raised about possible reflections in the 3 m
chamber due to imperfections in the installation of the pyramidal foam. I
tried sweeping from 1 to 10 GHz with the log spiral antenna, coupling to a
non-standard antenna, and performing an inverse FFT on the network analyzer
data to generate a time domain plot. I had very little success in actually
seeing reflections. For best resolution the ideal would have been to sweep
from 30 MHz to 20 GHz with two wide band antennas, but the company did not
want to spend the money for any new antennas. What I am thinking is that
careful return loss measurements may have shown if any reflections were
present.

I have 24" tall pyramidal foam, and that meets the requirement. As
frequency decreases, the foam essentially disappears. By 10 MHz, it has
almost no effect.

I think we were using 12" pyramdal foam, even on the floor, with inverted
foam to provide a walking area.

The pyramidal foam is expensive, about $50 / sq ft. If you want more
return loss, you need taller pyramids; those mythical governmental labs
have had foam up to 72" tall (and the wall absorbers tend to droop a bit
g).

With a 3m chamber, anything greater than 12" is not really practical.

A newer technique is to use ferrite tiles, especially on the floor. They
are less than a half-inch thick, and perform much better at low
frequencies. And the cost is about $100 / sq ft. I like to think of my
walls and ceiling as covered with $5 bills, and the floor carpeted with
$10's.

Your anechoic chamber is never really perfect; however, it becomes "good
enough" when you run out of money.

With the dark blue pyramids and black tiles, a chamber looks like a bat
cave. One vendor decided that the new millenia needed white paint on the
foam; another vendor touts pyramids that have a 90-degree axial rotation
part way up the taper, and yet another truncates the pointy tips, telling
us that works better. It's just like the antenna game.

I have heard of the ferrite floor tiles, and are probably a much better
solution than inverted pyamids fitted into the floor mounted pyramids.

No, 461 doesn't like log periodics either, saying:

"Other linearly polarized antennas such as log periodic antennas are not
to be used. It is recognized that these types of antennas have sometimes
been used in the past; however, they will not necessarily produce the same
results as the double ridged horn because of field variations across the
antenna apertures and far field/near field issues. Uniform use of the
double ridge horn is required for standardization purposes to obtain
consistent results among different test facilities."

The MIL-STD defines a 104 cm rod from 10 kHz to 30 MHz, then a biconical
from 30 MHz to 200 MHz, and finally, horns above there. Since pyramidal
horns are only good for about an octave, a smart Navy guy added
exponentially flared ridges to the horns, and came up with multi-octave
horns. A typical horn for 200 MHz to 1 GHz has an aperture of about 1
meter, then another horn tries to go from 1 GHz to 18 GHz. That's a bit
too far for me, as the antenna factor really climbs above about 14 GHz, so
I switch to a common, non-ridged horn for 12 GHz to 18 GHz. For 18 GHz to
26 GHz and 26 GHz to 40 GHz, I use standard-gain flared horns. With a
pre-selected spectrum analyzer, really good coax, and a couple of
low-noise pre-amps, that lets me get comfortably below the most stringent
RE102 limits.

I think they were considering horns and low noise amps to get above 10 GHz.
I did a lot of analysis to figure out what was required, but never got to
finish it, on account of being laid-off! Nobody ever seems to want to spend
the money to get it right.

OK, just for trivia's sake. If the antenna base was cylindrical, painted
grey crinkle, had a 6-position range switch and a brown bakelite top
insulator, it was an Empire VA-105.

Describes it perfectly

But, if it was almost a cube, painted battleship grey, had a black front
panel and an 8-position range switch, it was a Stoddart 92138-1 (that
number is a hazy memory). Both were passive antennas. The Empire was used
with the NF-105 receiver,

That was the one I used, now you mention it I remember the model number as
the NF-105

while the Stoddart antenna was associated with the NM-22A (that's why the
range switches were different, to match the ranges on their associated
receivers).


--
Ed
WB6WSN
El Cajon, CA USA

73,

Frank