Bravo Eric,
I am glad someone took up the task of doing this kind
of modeling work. One question though - did your modeling
assume uniform flat ground, as I wonder if this is too ideal
an assumption to extrapolate results to the real environment?
If I had the time and the software, I think it would be
interesting to take a couple of different antenna models
(as you have done with EZNEC), put them into a program like
K6STI's TA, and then do a sensitivity analysis. I am still
wondering if the consistency you describe would survive when
non-idealities such as feedline radiation and side scatter
were introduced into the equation (I am a little less skeptical,
but not yet ready to take communion).
73 de Mike, W4EF
----------
From: Eric Gustafson[SMTP:n7cl@mmsi.com]
Reply To: n7cl@mmsi.com
Sent: Saturday, July 03, 1999 12:39 AM
To: towertalk@contesting.com
Subject: [TowerTalk] Antenna Measurement Discussion
Hi TTians,
I read with interest the recent discussion on Ward Silver and
Steve Morris's tribander comparison test. I was motivated to do
some modeling to check out one of the arguments that was
proposed. Namely that since the patterns are different, you
can't do a valid comparison near ground due to operating far down
in the ground effects null. This argument and the large distance
separating the emitter and receiving sites were the only two
things that really bothered me about the tests.
I'm posting a separate message with the results of the modeling.
But the bottom line is that NEC2 doesn't think that at the height
above ground used for the tests there was any relative advantage
given to any of the antennas tested due to differing ground
interaction due to differing patterns.
I must be strange in some way as I seem to be the only person who
was not shocked by the reported results of their test. But I
don't think that their results are really that far out of the
ball park. Here is a very non-rigorous "proves nothing" thought
process that I go through when thinking about the test results.
About all you can get from a three element 20 meter Yagi on a 26
foot boom is 7.3 dBd. And _that_ requires you to accept a F/B
ratio of only 8 dB and a feedpoint resistance of 7 ohms! By the
time you get this antenna efficiently matched up to 50 ohms, you
will have a very high Q system that works as advertised over only
a few tens of KHz.
On a 26 foot boom with a F/B greater than 20 dB, pretty much all
that we can expect (on 20 meters) is about 6.2 dBd. And that is
for a monobander with no element shortening due to trap
inductance, and no loading, matching or trap losses and no
extraneous aluminum for other bands in the near field zone.
Claims of substantially more gain that this on 20 meters for a
multiband antenna with a boom length equal to or shorter than 26
feet should be regarded as highly suspect.
I think it would be wise for the amateur community to lower its
expectations for the forward gain performance of small multiband
HF parasitic arrays. Then maybe the manufacturers wouldn't be so
strongly motivated to tell us what they think we want to hear.
Looking at the data from Steve and Ward's first report (I haven't
seen the second one yet.) for a representative three element
trapped tribander with a boom length close to 26 feet (TH7DX), I
don't see why there is all that much controversy over their
reported results. Doing a little arithmetic yields the
following:
The manufacturer claims a gain of 7.4 dB (unspecified reference)
and a front to back ratio of 27 dB. If the reference is an
isotropic radiator, then this equates to 5.29 dBd and is notably
within the realm of theoretical (lossless) possibility. The
measurement reported for 20 meters was 3.0 dBd. In the report
they claim their measurement uncertainty was 0.5 dB So we are
looking for 5.29 - 3.0 - 0.5 = 1.79 dB of "missing performance".
This is a trapped antenna. So there will be some loss in the
traps and some loss of performance on 20 meters due to element
shortening. The generally accepted number for total trap loss is
"about 1 dB". I don't know where that number came from and I
don't really want to argue about it. So I did a quick back of
the envelope calculation to convince myself of what it _might_
reasonably amount to give some reasonable assumptions. The
assumptions that I used were (a) the most power a single trap
could dissipate and survive is 30 watts and (b) the antenna must
be able to withstand 1500 watts key down for a minute or two.
Then, given the relative currents in the various parts of the
various elements, I came up with aproximately 1.2 dB as the most
total loss due to the traps that could be survivably tolerated.
As a sanity check, I asked one of the local Eimac grinders about
his experience with trapped tribanders. He said that the
tribander he still has up is a TH7. That is because it ts the
first one that didn't go up in smoke during an extended tuneup.
He said that of the five different tribanders of various
manufacturers he has had over the years, the TH7 was the only one
that could tolerate a 1500ish watt unmodulated carrier for any
length of time. So I think that 1.2 dB is a reasonable estimate
for the upper limit of the loss that could be expected from
dissipation in the traps. So now we are looking for only 1.79 -
1.2 = 0.59 dB of "missing performance".
I don't find it difficult to imagine that the sum total of all
the remaining loss factors could total 0.59 dB. These would
include matching system loss, balancing device (if any) loss,
extraneous aluminum (for the other bands) in the near field, user
construction variancees, etc. IMHO, 0.59 dB just isn't an
unreasonable amount to expect for "implementation" loss in a
multiband parasitic antenna produced as a commercial product for
an extremely price sensitive market.
Again, I don't think they are very far off the mark with their
comparisons.
But rather than argue about those tests I think it would be
_MUCH_ more useful for others to attempt the same type of
measurement and report the results to the ham community as Steve
and Ward have. I very much like Roger Cox's idea of an organized
test at one of the big 'fests. To that end, I think it would be
much more productive for us to offer our suggestions for
improvements to the test protocol so that the issues can be
discussed and possibly even agreed upon before the next test is
undertaken.
I have a few suggestions for possible improvements to the test
methodology. And here they are essentially as I sent them to
Ward and Steve back in January. My suggestion for the distance
to use is between 350 and 700 feet. That would represent between
5 and 10 wavelengths at 20 meters. I have seen 500 feet
suggested. 500 feet should do just fine. My other suggestions
are in the message appended below. Go ahead and shoot at 'em.
Lets get the methodology to use sorted out!
73, Eric N7CL
--------------------------------------
To: <towertalk@contesting.com>
Date: Tue, 12 Jan 1999 11:18:14 -0700
From: Eric Gustafson <n7cl@mmsi.com>
To: K7LXC@aol.com
CC: hwardsil@wolfenet.com
In-reply-to: <fb71a632.369a2342@aol.com> (K7LXC@aol.com)
Subject: Re: [TowerTalk] KLM 10-30 Log Perodic
Reply-to: n7cl@mmsi.com
Hi Steve, Ward,
>From: K7LXC@aol.com
To: <towertalk@contesting.com>
>Date: Mon, 11 Jan 1999 11:13:54 EST
>
>In a message dated 99-01-10 12:43:06 EST, you write:
>
>> 2. Include a couple of the commercially available Log Periodic
>> antennas that cover the same frequency range as the
>> tribanders for comparison.
>
>Hi, Eric --
>
>Tnx for your input - it's always appreciated.
>
>We have an X-9 already in the garage for the next round of
>tests. Also hope to test the C-31XR and maybe a monobander and
>an A3. There is no lack of antennas that we could add to our
>test.
>
>The big obstacles are getting one to test and the time involved
>in the assembly, testing and write-up. Needless to say, we spent
>a LOT of time on the initial testing and report.
An impressive effort!
Are Hy-Gain, Tennadyne, or KLM (or M^2, etc.) not approachable to
get you a LP to test? If not, I will be getting an LP from one
of those vendors sometime in the reasonably near future. Would
you like for me to have it drop shipped to y'all before it gets
to me? I could order it well before I'm to the point in my tower
project to be ready for it.
>
>As far as changing the antenna distances, since it would be
>different than the first test - it wouldn't be a standardized
>antenna comparison anymore. And we're not willing to re-test the
>original batch.
Right. I wasn't advocating the range change for your testing but
for others to consider if they actually do set out to do the same
kind of testing. I notice that there is already reference to a
similar recommendation in the report.
After reading the report again, I do have some additional
comments. These are just comments. Not criticisms. I spent a
number of years doing antenna pattern characterization for the
Army at Ft. Huachuca here in southern AZ. I realize that no
amateur is likely to duplicate what can be done on a real antenna
range.
1. You show that you are radiating from the AUT to the reference
for the pattern measurements. I would be inclined to radiate
from the highest directivity antenna I had to the AUT for
making pattern measurements. Particularly if I was operating
over a reasonably short path (not your case).
As a general protocol recommendation, I'd specify measuring
the pattern while receiving on the AUT. This permits the
greatest flexibility in selecting the test location since
only one direction must be absolutely clear of large scale
perturbations for a long distance. The more places that are
viable for making this type of measurement, the more likely
that someone may attempt to do the test.
2. I'd like to see a bit more dynamic range available in the
attenuator system. But what you have is probably adequate
for the stated purpose. It would be useful to locate the
pattern minimum, set the signal and attenuation levels so
that the RSS indication is up in the useable range (probably
s7 or above) and proceed from there. If this is the
procedure, you may discover a need for more than 30 dB of
step attenuation.
3. I notice that all of your patterns seem to show a side to
side asymmetry. The side null (which should be nearly
complete) is shown as much deeper on one side than the other.
The remarkable thing about this is that the deep side is the
same across many different antenna models. Since the
feedline decoupling was not identical from one model to
another (I could be wrong here but that is what I took away
from reading the report.), I must assume that this had to do
with the 15 degree sample intervals not aligning with the
null minimum on both sides. I'd recommend using a higher
sample resolution but I have doubts that most hams who might
attempt this test would have access to positioners with
indicators much better than this.
Another possible source of this kind of artifact might be a
large scale scattering object to one side of the test area.
Measurement systems where the radiator is rotated are highly
susceptable to this kind of error. Did y'all track down the
source of this assymmetry?
4. It might be useful to suggest that the feedline for the AUT
be of the "foil enhanced" variety. If the test is run at
close range, and the null depths are to be fully explored,
feedline leakage can become a significant factor. A good
test is to put a fully shielded load termination at the
antenna end of the feedline and verify that the leakage is
low enough to not affect the measurements.
Let me know what you think!
73, Eric N7CL
------------------------------------
--
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
--
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
|