Hello,
thank you all for your replies to my questions (summary below)
SRI I caught a flu, hence my delayed answer...
1.)
Big kudos to W6RMK for his excellent suggestion of how to approach
this problem in a simple simulation model:
- instead of running amok trying to model the metal roof, just try modeling
the antenna 6m above perfect ground. Then, run again 2m above perfect
ground.
Why didn't I come up with this idea myself...? Must have been the flu... :-)
Anyway, I did these simulations and the SWR values they predict are
pretty close to what my friend has measured.
Thus, this approach seems to point in the proper direction.
The matching values are true for both measured/simulated SWR curves,
6m and 2m above the roof.
(with 2m separation, the SWR stays OK, but of course the other antenna
parameters are much more affected.) But you know amateurs tend to
believe the antenna is OK when SWR is OK...
Running this simulation again, shortening the 20m driver by approx 8cm each
side, the resonance comes back into the 20m band, at 6m above the roof.
The other parameters (gain & F/B) show quite a good behaviour again, too.
They are still somewhat affected but if the reality was close to what the model
predicts, I would not touch the parasitic elements...
2.)
Steve, you are of course right with your question, what conclusions or
adjustments would be made armed with the model... Of course field
testing is always better.
The problem in this special case is, that I can only speak to the guy
on the radio or telefone.
He must do all the actions himself, but claims himself to be unexperienced...
So in this special case the simulation helps me to prepare and (hopefully)
make better suggestions to him...
About "going overboard" with the interaction issue, yes and no.
While it is true that when you are trying to make an element act as a reflector
or director it will do nothing unless it is "just right" (within a few
centimeters exactly)
It is also true that when you are trying to make an unwanted element NOT to
act as a reflector or director, you will often have a hard time doing so...
Must have something to do with Murphy... :-)
N4KG used to say many times on this reflector:
>>> ANY conductor longer than 1/2 WL out to several WL
>>> WILL act as a reflector. ANY conductor between 0.3 and 0.5 WL
>>> WILL act as a director.
>>> The ONLY way to make a conductor 'invisible' to another antenna
>>> is to make it appear as something LESS THAN 0.3 WL at the
>>> operating frequency.
... which I often had to fight with, while developing the "spiderbeam", an
interlaced triband yagi design for 20m (3el) / 15m (3el) / 10m (4el)
3.)
Pete, thanks for the simulation software suggestion.
However, I am sing 4NEC2: www.qsl.net/wb6tpu/swindex.html
which will now even allow well over 10K segments.
Again, thank you all for your helpful contributions.
I hope we will have the antenna working soon...
73s Con DF4SA
##### SUMMARY BELOW #####
---------- VE6WZ wrote: ----------
As the Yagi is lowered closer to the ground the impedance, and resonance
will change. At VE6WZ the tower is a crank-up and the Yagis are raised
and lowered daily. When cranked down the resonance point is lowered
significantly on the 80 and 40m Yagi, but less so on the high-band yagi.
The low point of the tower is around 40 feet.
It seems that your friends installation will require re-tuning. If you
find your lowest SWR is higher than ideal....consider re-adjusting the
matching network at the feedpoint....it is likely the impedance will
have changed also at this lower height.
Because of the mutual coupling between the elements, if you first
shorten the driver to resonance, and then later shorten the REF and DIR,
then you may see the driver move back down a bit. You may need to do a
few iterations to get it right.
As for the pattern.....I think the best way to be sure the Yagi is
properly tuned is to do some actual field measurements of the F/B.
Re-adjusting all the elements the same amount may...or may not...be
optimum. It will likely be close, but could probably be fine-tuned.
gl, de steve ve6wz.
---------- W6RMK wrote: ----------
I don't know that capacitively loading is actually the entire story of
what's going on with a antenna above a metal roof. There's not only
capacitance from elements to the ground plane (several pF/meter), but also
the currents induced in the roof. The pattern will almost certainly change,
but whether for good or bad is hard to tell.
You're also going to form an "image" antenna in the roof, much as an
antenna over good conducting ground forms an image, which can change the
pattern. A lot depends on the size and conductivity of the roof.
It's no surprise that the interactions are very noticeable at 6m above the
roof. That's getting close to a quarter wavelength, and one would expect
significant impedance variations. (think of the "reflected" wave from the
roof coming back up to the antenna and being almost exactly out of phase
and, so, causing a cancellation of part of the element current...) Up
close, though, the coupling is very strong, but, also, close to being in
phase, so things like resistive loss might dominate, but field cancellation
less so.
Modeling in NEC, you might get a feel for it by modeling the antenna over
perfect ground, 2m above the ground. You could even attempt to adjust the
conductivity of the ground to try and match the roof, just to see what happens.
What you want to watch out for is that you don't want to use a reflection
coefficient approximation, which is more suited to figuring out the far field.
You might also try specifying the ground radials on the GN card, although,
the radials are at 0,0,0. I don't recall off hand how to appropriately
model the "cliff" problem.
---------- N4ZR wrote: ----------
I think your concerns are fully justified, Con. After all, electromagnetic
radiation obeys the inverse square law, so the current in the roof will
become greater the closer the antenna is to it. Serendipitously, the SWR
or resonant frequency may seem more normal, but it's unlikely that the
pattern is improved by being closer. I think Steve's wrong about just
adjusting the elements and living with the thing -- I think that resonant
frequency perturbations are a sure sign of bad things happening to the
pattern. Once in a great while, good things can happen through
interaction, too, but it's a lot less likely...
If you are hacing segment problems, may I suggest you try MultiNEC. It's
an Excel front-end for NEC-2 that can be used with various public domain
NEC-2 compilations with up to 5000 segments. It comes with a 1500-segment
version, and will also automatically produce a wire grid that can be used
to simulate a metal roof surface.
You can download a fully-functional trial version at
<http://www.qsl.net/ac6la/>
73, Pete N4ZR
--------- W6RMK wrote: ----------
>I think your concerns are fully justified, Con. After all,
>electromagnetic radiation obeys the inverse square law, so the current in
>the roof will become greater the closer the antenna is to it.
Worse than that... it's only inverse square in the far field, where the
source can be reasonably approximated as a point source. In the near field
[which you certainly are in (rule of thumb is 2 pi wavelengths for antennae
that are <1 wavelength in size)], this assumption isn't valid. For
instance the magnetic field around a linear conductor only drops off as 1/r.
Jim, W6RMK
---------- "Bob" wrote: ----------
Had to shorten 20 and 40m sections of the 4BTV trap vertical when
I moved it to the KH6 house with the metal roof. It's ground mounted and
about 18" from the house with 40m trap at roof edge level and a fairly
shallow pitch to the roof :-)
Pattern? Happy to get any antenna up at that house~
-Bob
---------- VE6WZ wrote: ----------
I think sometimes we go "overboard" with this interaction issue.
Using any modeling program, start with a dipole and then introduce
another wire and TRY to make a pattern.
Only when the second wire is close to resonance, and at the right
spacing are the currents correct to develop a pattern. This new wire
needs to be accurate within inches in length to make a pattern. After
all, this is why it can be so difficult and time consuming to adjust a
Yagi model (or in the real world on a tower) for optimum F/B. That
reflector (or director) has to be "just right" or it won't do a thing.
Modeling and experience shows that introducing
another...non-resonant...element will tend to load the existing wires
and hence lower the resonance point. The models show that some current
does flow in the added wire, but it's effect on the pattern ****at the
new resonance point****is negligible unless it is tuned "just right".
The important point here when modeling the interaction is to model the
new system at the ***lowered resonance point***, otherwise, the
conclusion would be that the pattern has changed. The system just needs
to be re-tuned.
As far as the steel roof.....good luck modeling that with confidence.
What conclusions or adjustments would be made armed with such a model ??
My recommendation to Con was to retune the thing and ***field test it***
for best F/B.
de steve ve6wz.
---------- my Original Message: ----------
>
>
>VE6WZ wrote:
>
>>Put both elements back pointing the same direction.
>>Shorten all the elements on the 20 the same amount to bring it back to
>>resonance.
>>
>>The fact that the 20m Yagi has been "loaded down" because of the 40m
>>does not mean that there will be a problem with the pattern AFTER it is
>
>>re-tuned. There IS interaction....but only to the extent that the 20m
>>elements now resonate lower in frequency.
>
>
>
>I recently have been asked a similar question by a friend who is trying
>to
>put up a 3ele 20m yagi 6m (18ft) above a sheet metal roof.
>
>The resonance (point of lowest SWR he measured with his SWR bridge)
>moves down to 13.8 MHz.
>The interaction seems to be much less pronounced when he moves the
>antenna to only 2m (6ft) above the metal roof.
>
>I was hesitating a bit to give the advise of shortening first the driven
>element, and afterwards shortening also the parasitic elements by the
>same
>percentage.
>
>Do you really think this will work?
>OK, I understand your point, the roof is "capacitively loading" all the
>elements and therefore all the elements have to be shortened
>accordingly. True? I am still afraid that even with the shortened
>elements, the antenna pattern will not be restored.
>
>I have tried to model the situation with NEC2 but had no succes in
>making a workable model (too many segments)
>
>Any more comments?
>Thank you
>
>73s Con DF4SA
_______________________________________________
See: http://www.mscomputer.com for "Self Supporting Towers", "Wireless Weather
Stations", and lot's more. Call Toll Free, 1-800-333-9041 with any questions
and ask for Sherman, W2FLA.
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
|