View Single Post
  #3   Report Post  
Old October 17th 03, 04:34 AM
Mark Keith
 
Posts: n/a
Default

"Serge Stroobandt, ON4BAA" wrote in message ...
Dear Roy, Richard, Art and other readers,

I had a look at that thread (for other readers' convenience included at the
bottom of this message).
Although it is not immediately answering my question whether inverted
groundplane verticals present fewer losses than normal verticals, it
contains some
interesting observations. Again for the sake of the other readers, I took
the liberty to rephrase and summarise these in terms that are more common in
antenna literatu


Why are you calling them "inverted ground planes"? To me , an inverted
ground plane would be one that is standing on it's end, with the
radials on top.
It's just a ground plane. Period. A ground plane is any vertical that
uses elevated radials to supply the lower part of the antenna. "Not to
be confused with decoupling radials." Adding decoupling radials to a
half wave that is elevated, does not turn it into a "ground plane".
It's still an elevated 1/2 wave with a decoupling section. A ground
mount vertical, is well, a ground mount vertical.

The electromagnetic field around an antenna exists out of two parts: a near
field and a far field.

Due to maths, the near field rapidly becomes negligible at larger distances,
whereas the far field is the field we use normally to communicate.

Losses in the near and far field can be reduced by reducing ground losses in
the immediate vecinity of the antenna.


Reducing ground losses under the antenna will have no effect on ground
losses in the far field. Those are fixed per the existing ground cdx
at the time, and are unchangable by you, unless you move, change
seasons, or can have someone make it rain over the whole area for four
days. Rain dancing" :/ I'm not sure what difference that actually
makes as fresh water is not near as conductive as salt water, but I
think helps a small bit. But, yes, if you reduce losses at the
antenna, you have more to work with out in the far field. So if far
field losses were say 2 db to a certain point when using an angle of 5
degrees, if you reduce losses at the antenna by 2 db, "adding radials"
you will see that extra 2 db at the receiving end over the less
radialed version. But the far field losses in DB will be the same.
"2db".



Losses in the far field are in addition also caused by the ground properties
at larger distances from the antenna, and are therefore less controlable
(apart from choosing a coastal or salt-lake QTH).

Now, my assumption is that by inverting a vertical ground plane, some of the
near field ground losses could be prevented by presenting the high impedance
end of the radiator to the low impedance of the earth immediately
underneath. This tremendous mismatching would cause less coupling of the
antenna's near field into the lossy earth.


True, if you actually inverted a ground plane. But adding radials at
the top would be silly... But, yes, if you had a 1/4 wave vertical,
and fed it from the top, say with a feedline running up through the
middle, the point of max current would then be at the top, and high
voltage at the bottom. You would see less ground loss than a base fed
1/4 wave, but you still couldn't totally ignore ground losses under
the antenna. I would still lay out radials.


Here are the results:

Normal 80m (3.650MHz) ground plane vertical
with 4 elevated radials 4.5m above "good" ground (according to Christman
KB8I, see also John Devolder, ON4UN, "Low-Band Dxing," chapter 9)
radiator and radial lengths: 20m
maximum height 24.5m
azimuth 0°: 0.23dBi @ 22° elevation
azimuth 45°: 0.21dBi @ 22° elevation

Inverted 80m ground plane vertical
with 4 elevated radials 24.5m above "good" ground
radiator and radial lengths: 20m
maximum height 24.5m
azimuth 0°: 0.08dBi @ 22° elevation
azimuth 45°: 0.12dBi @ 22° elevation

Result: 0.09-0.15dB in favour of the normal ground plane. The radials up in
the air of the inverted ground plane probably cause some shielding.


This model is faulty as far as I'm concerned. The lower antenna is too
low to the ground to show low ground losses with only 4 radials. The
higher antenna would kick it's tail. I don't care what the model says.
In fact, the antenna at 75 ft would be VERY effective to DX at night.
Thats just over a 1/4 up. The pattern will be good.

The same difference remains when we lower both antennas to 1m above ground:

Normal 80m ground plane vertical
with 4 elevated radials 1m above "good" ground
radiator and radial lengths: 20m
maximum height 21m
azimuth 0°: 0.00dBi @ 24° elevation
azimuth 45°: 0.00dBi @ 24° elevation

Inverted 80m ground plane vertical
with 4 elevated radials 21m above "good" ground
radiator and radial lengths: 20m
maximum height 21m
azimuth 0°: -0.16dBi @ 25° elevation
azimuth 45°: -0.11dBi @ 25° elevation


Again, this model is faulty, and /or misleading when compared to real
world results. An 80m vertical at 3 ft off the ground, with only 4
radials would be a dog compared to the higher one at 63 ft.
If you had a ground mount with 120 radials, which is usually
considered a "near optimum ground system", you would need 60 radials
at 1/8 wave up to equal the 120 on the ground. At 1/4 wave up, you
would need about 8-10 radials for the same low ground losses. At 1/2
wave up, you only need 3 or 4. Once you get to a 1/2 wave, the ground
losses are very low, and start to compete with the efficiency of a
dipole at that height. But from personal use, I can tell you that even
just 4 radials at 1/4 wave up , will do a good job. But 4 radials at
1/8 wave up, will be a dog in comparison. When thinking of radials for
a ground plane, and the proper number of them to equal a certain
reference, you must think in terms of wavelength for the freq to be
used.


Being an appartment dweller myself, I once dreamed of putting up an inverted
ground plane antenna using the building as a support for the radiator and
the roof as support and hideawy for the two elevated radials.


It will work, but the ground losses will depend on the height above
ground in wavelength for the band in use, and the number of radials
used. Refer to the specs above to equal the "120 radial reference".
BTW, I think Reg squirms in his chair every time I bring up 120
radials. But I still think it's a good reference to compare to for
top results. Nothing I've ever seen in the real world has ever
departed from the "specs" I've given above as far as comparing the
number of elevated radials, to the number of radials used with a
ground mount system.


Ok, but now taking into account Roy's previous thread. How does a low dipole
at the same height of 21m stands out of this?
azimuth 0°: -5.77dBi @ 30° elevation
azimuth 90°: 4.23dBi @ 30° elevation

Result: In its preferred azimuthal direction, a low dipole, even at a height
of about a quarter wavelength, still outperforms any ground plane vertical
antenna.


Depends where you are talking, and the angle being used to make that
path. Sure, if you are on 80m, and work 200 miles away, the dipole
will almost always be better. But not so on long paths. The angles
used on long hauls will be fairly low. When modeling, compare the gain
at say 5-10 degrees off the horizon. But even this is deceptive. I
don't totally agree with the way most programs model and compare
ground mount to elevated verticals. I think most programs understate
the gain of elevated verticals, when used at the lower than 1 wave
heights on the lower bands. The patterns seem ok, but the way they
deal with the decreasing ground losses when elevated, seems a little
suspect to me. I find I have to crank up the ground quality to show
the gains I see in the real world. On the low bands, the increase
when using the ground plane vs dipole would usually be greater in the
real world, than is modeled. I think part of this is because the
programs don't seem to take into account the differences between
losses vs freq, and also take take into account the quirks of
propagation on the lower bands. It's my opinion that if you had two
antennas with the exact equal gain in a certain direction and angle,
and one was horizontal, and one vertical, the vertical would win on
160m long paths.
But a modeling program will not deal with that, so you have to
consider it on your own.


MK