Hi Guy,
I'm not saying the tests are correct or incorrect, but there are
certainly some things being overlooked in the response below.
> The idea is seriously flawed that the ground path somehow invalidates the
> measurements, using the protocol specified in the comparison report. The
> root assumptions of the modelling will explain why. Mosely is just going
> to have to eat the test results and live (hopefully redesign) with it.
>
> 1) There is nothing about ANY of the antennas tested that would increase
> or decrease their succeptibility to ground attenuation versus the others
> tested.
What about a condition where an antenna does not have a perfectly
horizontal radiation pattern compared to another antenna, say from
feedline radiation or re-radiation from clutter in, near or around the
test site?
A problem with accuracy could easily occur, since the horizontally
polarized component is greatly attenuated over the same distance
compared to any vertical component. The slightest amount of
pattern tilt could greatly affect path attenuation. That's why it is
imperative, in a test range, that the antennas NEVER be tested
where site attenuation is high and/or where re-radiation could
greatly affect results.
If you want to see an example where this occurred, there is a web-
site that proclaims by removing an antenna tuner and using a stub
match, several S units were gain in field strength. The test protocol
was a groundwave path over a few miles on 80 meters.
By simply removing an antenna tuner and feeding the feedline
direct through a stub, the person providing that data is totally
convinced he gained twenty dB or more. Let's assume the stub
was 100 percent efficient. In that case the tuner would have been
1% efficient. Since he ran a kilowatt through the tuner, it would
have had to dissipate 999 watts as heat without failing.
The true data above is a shining example of why a cluttered
groundwave path with high attenuation should never be used to
determine efficiency or gain.
Testing a horizontally polarized antenna close to ground over a
large distance is always a bad idea, even in a controlled test range.
The larger the distance and the more cluttered the path and area
around the antennas becomes, the more unreliable the test
becomes (even in an A-B test), unless the antennas all have the
same exact pattern (including the reference antenna).
This isn't to say the results aren't valid, just that the test was far
from ideal. The results would be much more reliable if the test was
made at 500 feet distant, rather than 5000 feet or more. Especially
if the close spaced test was made in an open location.
> 2) The modeling presumes the ground is a perfect plane with uniform
> electrical characteristics. ANY departure from that and the ground effect
> gets ill-defined, losing its sharp horizon null. It is almost never as
> deep a null as predicted. You can partly model real ground using the
> program TA, which allows the entry of height variations and their
> distance. With just a little messing with the ground countour, the ground
> null can be significantly reduced (ground sloping down from the antenna
> for five to ten times the antenna height, for instance). The program
> cannot, however, add in the variation in the electrical characteristics of
> the ground which would further difuse the null. In any event it is a FAR
> FIELD effect, and if the tested antennas are all set at the same height
> over the same spot, will NOT contribute any false divergence for the test
> results.
It is not only a far field effect, it is also a near-field effect. The
directivity of each antenna can greatly affect the results, especially
in a cluttered test site, when scattering or re-radiation occurs. With
that in mind, what you say above is true ONLY if the antennas all
have the same pattern.
Let me give a specific example:
Let's assume we are comparing a dipole to a yagi for the purpose
of evaluating gain. Let's also assume something near the antenna
under test but NOT in the direction of the receiver is re-radiating
energy that re-enforces energy in the direction of the receiver.
A bidirectional antenna with a wide pattern, like the reference
dipole, would couple to the re-radiator much differently than a
directional antenna under test. The reference dipole could easily
have gain that does not show up at the receive site as a FS change
when the dipole is rotated, and that gain would not be available to
the directional antennas.
Let's also assume an object, let's say a telephone wire near the
antenna under test, re-radiates and skews the pattern off
horizontal. Under this condition an antenna under test that radiates
with a slightly different pattern than another, but basically the same
gain, could easily produce a stronger or weaker signal at the
distant receiver depending on how it couples to that near-field (or
far-field) re-radiator.
A cluttered test site should NEVER be used for gain measurement
meant to apply at different angles or locations where the results are
plotted, especially when the pattern of the reference antenna is
greatly different than the antenna being tested. I can't think of a
single test in the world that would ever be considered a valid gain
test when gain is measured OUTSIDE the main lobe of the
antenna, or on the slope of the main lobe, unless it is gain only in
that direction and angle that the tester is concerned with.
Which leads to another question...how perfect was the measured
pattern of the dipole in the actual test site environment?
The only way to answer this question would be to move around the
dipole and plot the pattern near the dipole. If the site suffers from re-
radiation or scattering, rotating the dipole and measuring FS at a
distance location might never show the problem, although if the
dipole pattern is not perfect when measured at the distant site we
could be certain the site is polluted.
> At one time, Mosley may have been the best in the field. That was then.
> Have they been setting pat on the old designs? Metal covered traps?
> Perhaps more durable, but you won't ever catch me winding a coil inside or
> around a metal pipe. It's lossy. There is induced circulation current.
> It's as if I added a slightly resistive closed loop coupled to the coil.
All traps, and all linear loading sections, add loss. At least with a
trap the loss is primarily confined to the conductors inside the trap
so we can get an idea about the loss from temperature rise, or
easily measure the traps. With linear loading, we have the same
losses distributed over a large area. Losses can be masked
because heat transfer is increased.
The bulk of the loss occurs primarily for two reasons:
1.) Anything that adds capacitance across an inductor increases
the circulating current in the inductor (or stub).
2.) A "shorted turn effect" will reduce mutual coupling from turn to
turn and also generate eddy currents that reduce coil inductance.
That means more conductor area with current flowing is required to
produce the same inductance.
The result of this is that the bulk of heat is in the coil, and not in
the shield. Even so, the small amount of heat generated in the
shield by eddy currents still contributes to trap heating. What this
means is virtually all of the power lost in the trap heats the trap
internally, and most of that heat is right next to the plastic form.
Failures in Delrin tuner rollers occur when as little as 40-80 watts of
power are lost in the inductor. Where are the melted trap coil
forms?
If the TA-33 has a trap problem why is the gain low uniformly on all
bands? On 20 meters, the trap offers minimal loading effect. It
goes into an entirely different mode on 15 and ten meters. Even if it
has slightly high ESR on 20 meters, why does ruin twenty meters
when it is located at a point where current is low?
How can the plastic in the traps handle hundreds of watts of
heating without failure? Has anyone measured a trap? If no one has
measured a trap, how do we know the trap is lossy?
One of the main reasons the tests were done, to my knowledge,
was to establish a good test protocol so we know what we are
buying. I think that is an excellent idea.
While the test results may (or may not) have established triband
"pecking order", they were a long way from establishing absolute
gain. I hope this discussion results in improved measurement
methods, rather than mud-slinging and name calling.
73, Tom W8JI
w8ji@contesting.com
--
FAQ on WWW: http://www.contesting.com/towertalkfaq.html
Submissions: towertalk@contesting.com
Administrative requests: towertalk-REQUEST@contesting.com
Problems: owner-towertalk@contesting.com
Search: http://www.contesting.com/km9p/search.htm
|