> Hi Guy,
>
> If this gets too long, be sure to read the closing part. This was NOT
> the first published independent test of tribanders.
No. Certainly not. The one you mention below was actually contained in the
booklet.
> I must not have made my points clearly.
You think perhaps I disagree with you because I don't understand what you
write? Actually I have found your writing quite clear. Whether I agree with
WHAT you write on a given day is another matter. I think on this one we'll
just have to agree to disagree.
> Point 1
> You still seem to think I intended to say the polarization tilt was
> due to the antenna being tilted. I tried especially hard to not give
> that impression, even to the point of trying to directly say that is
> was due to re-radiation several times.
I was not just answering *your* post.
> While I NEVER said the cause was improper mounting, I might
> have made it clearer. I hope this does the trick.
> The most likely causes are nearfield re-radiation, far-field re-
> radiation, and feed systems. Two out of three of the above are
> different for every antenna measured.
Zero of the above are greatly different.
Coax feed? Taped to the boom. Antisnag loop. Taped to the same tower. What's
different?
Far field? Not unless the terrain has been rearranged by an earthquake.
Near field? Same near field stuff in the same place for each antenna. More
on this below.
> Point 2
> We ignore the absolute fact test equipment used, and HOW it was
> used, implies more than a few dB of tolerance in the numbers
> measured.
Why? Each measurement starts off with the same dipole reference. Thus the
procedure is self-calibrating to the dipole. Obviously not absolute, but
it's a comparison.
> It is virtually impossible to measure gain to 0.1 dB, let alone 1 dB,
> outside of a lab environment with freshly calibrated standards. I'd
> wager the tolerance in the equipment and methods used easily
> exceeded 2-3 dB.
Huh? Not only NOTimpossible, but common.
I used equipment working for Ma Bell that went to tenths of db's, every day.
Wasn't in a lab either. The stuff we have now is better and more stable. I
can measure to 1/4 db with the stuff in my shack. If I really needed 1/10 dB
for something, I could certainly get it.
> Take those measurements and the equipment
> outside in an uncontrolled range, and anything can happen.
> Especially when measuring in a null. Now this doesn't mean the
> results are that bad, just that we have no idea if they are or are not
> that bad...especially since they are totally unverified with any
> attempt at cross checking with another method.
I would say that the other test you said that wasn't considered, but is
actually reprinted in the booklet, is really a confirmation that this test
is in the ball park. Did you read the booklet? Just because Mosleys rated
comparable on another test does not mean that the one in this test wasn't
defective some way. Have to go with the one you got.
Did you hear the story a while back about a car company that delivered two
of a new model to a regional manager near a car magazine headquarters? One
to go to the mag for an evaluation article and the other for him to drive
and use as a demo. He went out and started one, and it ran rough cold and
was otherwise uneven in its power curve. The other one was fine.
The manager took the good one home. The other one went to the mag. The
article panned the car. Whose fault? Moral: best make sure the stuff you
send to the testers is all spiffy and working clean. Otherwise noone but
yourself to blame.
> Point 3
> I obviously wasn't clear enough in describing why the antennas are
> all different. They have different patterns, they were tested in an
> uncontrolled environment days or weeks or months apart, they all
> have VERY different near-field patterns and even a difference in far-
> field patterns.
NO, they DON'T have very different near-field patterns. The difference in
far field patterns is related to their gains, all except the Mosley, which
stands out on that account. Perhaps you define very different in the second
decimal place? I have almost a thousand modeling runs in a study I am doing
on interaction of guy wires, towers and beams. One interesting effect is
that if you are talking about a single yagi on top, anything on a halfwave
or less boom affected fairly equally by miscellaneous conductor *under* the
antenna. Talking about single percentage point variations. To the point that
it doesn't accomplish much to use anything except a 3 element yagi for the
frequency tested in an interaction run. Same general facts, a lot less
runs. I'm a little upset with myself that I didn't pick up on that earlier
than I did. VERY different??? Not.
> As a matter of fact, the reference antenna was a dipole. It's pattern
> was totally dissimilar to the other antennas, rendering it useless in
> an uncontrolled near-field or far-field environment when the
> measurement is made in a pattern null caused by polarity. It is the
> scattering and tilt that FILLS IN the null and makes the antennas
> all not have zero signal, along with the imperfect earth. If the
> patterns are different, we can be absolutely sure the scattering
> effects are different.
In that case the antenna with the *larger* gain is more subject to the tilt.
To move on to the comparison, then all the antennas EXCEPT THE MOSLEY are
subject to the tilt. I might get suspicious if ONE of the antennas was
hugely *better* than all the rest.
A dipole doesn't have THAT different a shape of pattern than the yagis
toward the receiving setup in the angles near the boresight. The deep diffs
in pattern shape are in the back 180 degrees.
BUT it doesn't matter any way. The dipoles sole purpose is to restart each
measurement against the same standard, not set an absolute gain. Even so, if
the dipole WAS really off, ALL the gains would be up or all would be down.
Since it's a comparison and we're not trying to tell anybody absolute gains,
it doesn't matter.
> I certainly agree, when all that is ignored the test is a perfect
> comparison with no room for error.
>
> The best article I ever read on testing was one about pathological
> science, and how a series of measurements were used to confirm
> a certain theory. Turned out the results were given latitude because
> of delta in the test equipment, and trying to measure differences
> smaller than the resolution of the setup.
>
> Each person making the tests had the same results as they
> repeated the tests over and over, yet the results disagreed with
> each other when the people were changed.
>
> None of this is all that sinister, and it certainly doesn't imply
> dishonesty. This is all just part of human nature, like the tendency
> to dismiss whatever argues with what we believe to be true.
Yup, Yup. Textbook stuff from the opening year in a degree in statistics.
Double blind, etc, etc, etc. NO argument. What is really neat, is that from
where I see it, this little tribander study has sidestepped the pitfalls,
mostly because the yagis are so similar where it counts.
> It would be nice, before one source is accepted as conclusive, if
> some attempt at a cross check was made.
Nobody on this list-serve got the authority to label anything conclusive.
Last I looked, everything subject to argument. Everybody gets to decide
themselves, what is conclusive, for themselves.
... Report actually published in booklet snipped.
> Before everyone leaps with both feet of top of Mosley, and ruins
> their business with techno-nonsense about how pure and accurate
> the triband tests are, we have to answer the question why a similar
> test a few years earlier reached what seems like different
> conclusions.
I don't think the REPORT will do much damage. What REALLY is going to hurt
them is that AD they published. It REALLY comes across bad. Here there's
just TowerTalk, and far, far fewer copies of the report booklet than
towertalkians. But the most damaging thing by far, THEY PUBLISHED IN QST!!!
UN-believable. How many copies of July QST went out?
Also, Tom, and this really bothers me about your line on this one, just
under the covers is the assumption (whether or not intended) that whoever
picks up the booklet can't work it out themselves, and needs us to explain
it to them, and if we don't, then Mosley will be unfairly driven out of
business. There are some number of posters here that won't accept the
report, for their own reasons. That's fine. Some will accept it. That's fine
too. Everyone will figure it out on their own terms. W4EF put up his post,
and his post actually cemented my views otherwise. I'm sure someone out
there has read my stuff, disagrees wholeheartedly, and my post cemented
their view otherwise. Thank God for a free country....
There were some other things... K7LXC does not work for Force 12 as
reported in the Ad. W8JI does do a lot of (very good) work for
MFJ/Ameritron/Etc (and now Hygain?)
As much as I insist so much of the above does not apply because it's a
comparison, I am sure that your penchant for the last good detail will
always tend toward a better product that I would rather buy than your
competitor's.
> 73, Tom W8JI
Likewise, Guy.
--
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
|