[Top] [All Lists]

Re: [TowerTalk] effect of large metal roof

Subject: Re: [TowerTalk] effect of large metal roof
From: K4SAV <>
Date: Sun, 27 Jul 2008 10:41:05 -0500
List-post: <">>
Based on simulations I have done for similar situations, I would expect 
a lot of interaction with the roof.  (I have done this simulation for a 
dipole over a 36 x 64 roof on 10 meters, and for verticals mounted on 
roofs.)  Generally what happens is that the antenna is affected most in 
the direction where there is more roof in front of the antenna.  That 
means on a non-square roof or one in which the antenna is located off 
center, the pattern changes as you rotate the antenna.  In this case on 
10 meters, I would expect somewhat of a cross between a pattern that you 
would get with the antenna at 10 feet and a pattern at a higher height.  
In other words, a distorted pattern from what you would expect, with 
loss of low angle gain and increase in high angle gain.  In some 
directions (where there is more roof in front of the antenna) the loss 
of low angle gain will probably be severe (about the same as if the 
antenna were located at 10 ft height).  Even when the antenna is pointed 
in the direction of the short side of the roof, I would expect to see a 
significant loss of low angle gain compared to what it would be if the 
antenna were located in the clear, but not nearly as bad as when pointed 
in the long direction of the roof.  On the higher bands things get 
scaled up, so there will be more interaction, with the antenna looking 
more like it is located at a 10 ft height.

A vertical using a metal roof as a radial system also operates a little 
weird.  The feedpoint impedance may end up being considerably different 
from what you expect.  The resonant frequency may be different as well.  
It's kind of like a vertical with elevated non-resonant radials which 
are too long.  You should be able to readjust the vertical length and 
get the resonant frequency back again. (There may be exceptions to 
this.)  The patterns may not look like a vertical if the roof is large 
and unsymmetrical.  It should have a null overhead, but there will 
probably be some high gain, high angle, lobes.  The low angle gain will 
probably be about the same as a vertical over ground.  It won't be 
totally omni-directional.  The roof will distort the pattern in the 
azimuth direction. The smaller the roof and the more symmetrical it is, 
the more the pattern will look like that of a normal vertical.

Many people say that roof mounted verticals work very well, and that is 
understandable when you look at the patterns.  If the roof is small and 
the antenna is located in the center of the roof, the patterns should 
look very similar to a ground mounted vertical.  If the roof is large 
and unsymmetrical (and the antenna is not located on a roof corner) 
usually the low angle gain is at least a good as a ground mounted 
vertical even though most of the power may going up at a high angle.  If 
the antenna is mounted on a roof corner, it will look a lot like a 
dipole with one half vertical and one half horizontal (as much 
horizontal radiation as vertical radiation).

Jerry, K4SAV

Gregg Seidl wrote:

>I'm sure this has come up before but I don't recall reading it.I want to put 
>up the new Bencher Skylark antenna for 12-17 meters and a lower 10 meter 
>antenna when the band starts to perk up again.The trouble is all my towers 
>have enough on them already and don't want to push my luck.
>    I do have a large metal pole shed that I could put an extra 9 foot Glen 
>Martin roof tower on.The antenna would be about 10-12 feet above the METAL 
>roof.The antenna would be about 30 feet above ground level.Not great but 
>better then nothing.
>    How much interaction would the large metal roof have on the antenna?The 
>shed is roughly 36 by 90.Would a vertical be a better choice,then tie the 
>roof into the ground plane?
>    Looking for 12,17 and 10 meter coverage.15 would be a great bonus. 
>Gregg K9KL 


TowerTalk mailing list

<Prev in Thread] Current Thread [Next in Thread>