On 01/10/10 07:44, Jim Lux wrote:
Owen wrote:
On 01/10/10 03:51, Jim Lux wrote:

Thanks Jim.


I would think that the buried radials are more convenient (broad band,
etc.)

Yes, I understand that there are advantages to buried radials, but I
don't understand the preponderance of cases where I see 120 radials
pinned on the top of infertile dirt. They still present a trip hazard,
and less money spent on just a few elevated radials may perform just
as well.

never underestimate the power of tradition. It was written by BL&E that
120 radials work, and the FCC accepts that for broadcast, so by golly,
that's what we do. Why 120? it was at the point of diminishing returns
or practicality back when the study was done (e.g. there was no
detectable change from going to more)

As for laying on ground.. I think that's more the laying on grass, and
eventually, the wire sinks into the grass/turf.

There's also the whole "the radials must be resonant" misconception..



Look at the performance of your ankle biting radials when the dimensions
are changed slightly.. For instance, if you shorten them by 5%, does it
make a big difference? For the buried radials, the length is very, very
non critical.

Yes, of course the feedpoint impedance is more sensitive to change in
length or conversely change in frequency.

While for a buried radial system (probably because of the losses) it's
going to be less frequency sensitive.

I expect so.




Something else to look at is the sensitivity of "efficiency" (and your
definition of radiated power in the hemisphere/power into antenna is
fine) to soil properties.. if the soil conductivity or epsilon changes
(as it will with changing water content) does the efficiency change
rapidly?

Yes, efficiency is sensitive to soil parameters... for both types, but
not very sensitive.

Maybe less sensitive for the buried radials? Or, it was "good enough"
for BL&E, so being so written, so shall it be done.

BL&E were measuring ground wave, I think solely. My efficiency measure
is the hemisphere, so ground losses play a different role.



Because of the impedance change mentioned above, the impedance
transformation needs adjustment for wide range frequency operation.
Not such an issue in the intended application, the DX window on 80m
here is just 50kHz.




If I haven't got something quite wrong in the modelling, it would seem
worthwhile to prototype the shortened version with a view to extending
the system to a four-square if suitable.

The shortened version will, of course, aggravate the tuning sensitivity.

Yes, but the model suggests that the variation in R is very small, and
variation in VSWR (with shunt coil match) is small... in that band segment.





I have still to read Rudy's papers... I am away from home (less
bandwidth) and I will download them later today when I get home. I
suppose that the proposed design challenges the norm of a very large
number of buried radials. In our case, part of the property is quite
rocky, and a configuration with just a few elevated radials offers
deployment opportunities that aren't suited to buried radials.

So, my original question is no so much suggesting everyone else got it
wrong, but why don't I seem more people doing it this way. Could I be
forgive in thinking that the popular, nearly universal, way is to
uplift the BL&E research at MF and apply it to 80m?

Tradition is a powerful force. Look how many years it took for someone
(e.g. Rudy) to put the substantial work into doing a real quantitative
experiment. For most hams, they're only going to do something once, and
if works ok, that's how it stays. Almost none are going to do a well
controlled A/B study, especially if there's a (not necessarily valid)
tradition that says A works better (where better is ill defined and
probably a combination of radiation efficiency and installation
convenience)

Until recently, modeling tools available to most amateurs were not
suitable for making the call, although there have been some people who
did models and published it, but, in the face of decades of "lay down
120 radials" it was a tough sell.

The other thing is whether the difference is big enough to "make a
difference" in observed system performance. For a lot of operators, a 1
dB change in performance might not be noticeable. If you're in a "either
propagation is there, or it isn't" situation the difference between good
and bad is 10s of dB. There are relatively few people who work at 0dB
SNR (where tenths count) on a regular and continuing basis, and they're
not necessarily the ones who are interested in doing experiments on
antennas on the scale needed.

Yes, there will be differing view on what is significant difference. I
am not in the school of declaring less than one or two S points is
insignificant in general.

In the case of a four square in the DX segment, users are looking for
performance... and it seems to me that the elevated three radials, eight
wave vertical with capacity had is very close to quarter wave over
buried radials... depending of course on the soil type.

You mention the modelling tools, I am not so much concerned as to
whether the elevated radials model is good, but whether the NEC4 buried
radials model is good, and likewise for radials on and just above the
ground because those models are setting the benchmark for the
performance of the alternative.

Owen






Owen

Owen wrote:
On 01/10/10 07:44, Jim Lux wrote:

The other thing is whether the difference is big enough to "make a
difference" in observed system performance. For a lot of operators, a 1
dB change in performance might not be noticeable. If you're in a "either
propagation is there, or it isn't" situation the difference between good
and bad is 10s of dB. There are relatively few people who work at 0dB
SNR (where tenths count) on a regular and continuing basis, and they're
not necessarily the ones who are interested in doing experiments on
antennas on the scale needed.

Yes, there will be differing view on what is significant difference. I
am not in the school of declaring less than one or two S points is
insignificant in general.

yeah, but there's a big difference between 6-12 dB and 1dB.. I think
most users would care about 6 dB. Many fewer about 1 dB. And even
fewer care about 1 dB AND have the desire and means to perform the
experiment in a controlled way. (well, this latter category probably has
less than 10 people in it, and only 1 has published in the last 50 years)



In the case of a four square in the DX segment, users are looking for
performance... and it seems to me that the elevated three radials, eight
wave vertical with capacity had is very close to quarter wave over
buried radials... depending of course on the soil type.

Hmm.. and there the real question is what kind of performance are we
talking about: the power radiated in a desired direction (Tx) or the
ability to null unwanted signals (Rx). Given the generally high noise
levels on low bands for Rx, a 1 dB change in efficiency of the antenna
might not make any difference for the latter.

A bigger effect on a phased array is the relative phasing. For a 4
element array, you can have pretty big errors in phase on transmit
without changing the forward gain much (30 degree phase error on one
element might give you a 1dB change). But a 30 degree phase error on
receive could turn a -30dB null into a -7dB one..

And for that, the lower loss of your elevated radials might make things
"pickier".. that is, as frequency or surroundings change, the reactive
term for each element changes, which could change the power distribution
and phasing among the elements (depending on the feed system used).
(obviously, one of the "current forcing" drive schemes would be less
sensitive to this)



You mention the modelling tools, I am not so much concerned as to
whether the elevated radials model is good, but whether the NEC4 buried
radials model is good, and likewise for radials on and just above the
ground because those models are setting the benchmark for the
performance of the alternative.

The modeling performance of NEC4 for buried wires and wires just above
the surface is quite good. Where I would be suspicious is for a wire ON
the surface or partly embedded in the surface.

Look for that paper by Burke and Poggio on validating NEC3 and NEC4 (it
was published at some conference in Ankara Turkey)

Thanks Jim, all fair comment and noted.

The end application is a four square phased array for the 80m DX window.

The location is at another ham's property, a rural location with ground
ranging from dryish clay to rock.

I do expect noise to be lowish compared to residential precincts.

The excercise is really about a design for a monopole that gives
reasonably good performance if extended to the four square configuration.

Yes, I note your points about the phase sensitivity to components. That
would be a challenge even with buried radials as although we have been in
drought for a long time with 'controlled' low moisture content of the
soil, rain changes that and the soil is no longer as homogenous.

Nothing is as perfect as a modeller's world, but the discussion and some
of the links offered give confidence that a shortened vertical with
capacity hat and three radials, and shunt coil matched should give
similar performance to full quarter wave verticals with 32 buried
radials.

I have just reread Cebik's article on buried radials, and my own models
seem fairly consistent.

As you say, Rudy's work is further confirmation allowing for the
difference in configuration and the |S21| use.

Owen

On 10/01/2010 06:13 PM, Jim Lux wrote:
Owen wrote:
On 01/10/10 07:44, Jim Lux wrote:

A bigger effect on a phased array is the relative phasing. For a 4
element array, you can have pretty big errors in phase on transmit
without changing the forward gain much (30 degree phase error on one
element might give you a 1dB change). But a 30 degree phase error on
receive could turn a -30dB null into a -7dB one..


How come ?
Can you elaborate how can these differences happen ?

Thanks
--
Ing. Alejandro Lieber LU1FCR
Rosario Argentina

Real-Time F2-Layer Critical Frequency Map foF2:
http://1fcr.com.ar

On Oct 2, 4:28*am, Alejandro Lieber alejan...@Use-Author-Supplied-
Address.invalid wrote:
On 10/01/2010 06:13 PM, Jim Lux wrote:

Owen wrote:
On 01/10/10 07:44, Jim Lux wrote:

A bigger effect on a phased array is the relative phasing. For a 4
element array, you can have pretty big errors in phase on transmit
without changing the forward gain much (30 degree phase error on one
element might give you a 1dB change). But a 30 degree phase error on
receive could turn a -30dB null into a -7dB one..

How come ?
Can you elaborate how can these differences happen ?


it's the difference between the effect on a peak vs effect on a null.

consider a simple 2 element array.. for sake of argument, say it's 1/4
wavelength apart and phased 90 degrees, so it has a cardioid
pattern.... a gain of 2 in one direction (where the signals from the
two antennas align), and a gain of zero in the opposite direction.
The gain is 1+cos(phi - spacing*cos(theta)) where phi is the feed
phasing, and theta is the direction.. in the preferred direction
1+cos(90 - 90*cos(0)) = 1+cos(0) = 2
in the 45 degree direction: 1+cos(90-90*cos(45)) = 1+cos(90-90*.707) =
1.895
in the 90 degree direction: 1+cos(90-90*cos(90)) = 1+cos(90) = 1
in the 180 degree direction: 1+cos(90-90*cos(180)) = 1+cos(90-90*-1) =
1+cos(180) = 0

Now spoil the feed phase (phi) by 10 degrees... (80
on boresight: 1+cos(80-90*cos(0)) = 1+cos(-10) = 1.984
on 45: 1+cos(80-90*cos(45)) = 1.959
on 90: 1+cos(80-90*cos(90)) = 1.174
at 180: 1+cos(80-90*cos(180)) = 1+cos(80+90) = 1.52E-2

The gain on boresight didn't change much... from 2 to 1.984 (0.03dB)
But the null in the back came up from zero to 1.5E-2.. (instead of -
infinity, it's now -18dB)

Change the phase error to 45 degrees...)
@theta=0: 1+cos(45-90*cos(0)) = 1.707
@theta=180: 1+cos(45-90*cos(180)) = .292

So, from the 10 degree error case, the forward gain went from 1.984 to
1.707, about 0.6dB...
but the null went from 1.52E-2 to .292 (from -17dB to -5 dB)..


The thing to remember on any gain antenna is that it takes very little
power to disrupt a null (after all, a -30dB null means that if you're
radiating 1kW in the forward direction, you're radiating 1 W in the
null.. so just another watt will double the energy in the null,
turning it from -30dB to -27dB...)

(And, you can see why making antennas with sidelobes -60dB is VERY
challenging... )


Now, change the phasing to, say, 80 degrees.. in the preferred
direction, the gain is now 1+cos(10degrees)

On 10/02/2010 07:12 PM, Jim Lux wrote:
On Oct 2, 4:28 am, Alejandro Lieberalejan...@Use-Author-Supplied-
Address.invalid wrote:
On 10/01/2010 06:13 PM, Jim Lux wrote:

Owen wrote:
On 01/10/10 07:44, Jim Lux wrote:

A bigger effect on a phased array is the relative phasing. For a 4
element array, you can have pretty big errors in phase on transmit
without changing the forward gain much (30 degree phase error on one
element might give you a 1dB change). But a 30 degree phase error on
receive could turn a -30dB null into a -7dB one..

How come ?
Can you elaborate how can these differences happen ?


it's the difference between the effect on a peak vs effect on a null.

consider a simple 2 element array.. for sake of argument, say it's 1/4
wavelength apart and phased 90 degrees, so it has a cardioid
pattern.... a gain of 2 in one direction (where the signals from the
two antennas align), and a gain of zero in the opposite direction.
The gain is 1+cos(phi - spacing*cos(theta)) where phi is the feed
phasing, and theta is the direction.. in the preferred direction
1+cos(90 - 90*cos(0)) = 1+cos(0) = 2
in the 45 degree direction: 1+cos(90-90*cos(45)) = 1+cos(90-90*.707) =
1.895
in the 90 degree direction: 1+cos(90-90*cos(90)) = 1+cos(90) = 1
in the 180 degree direction: 1+cos(90-90*cos(180)) = 1+cos(90-90*-1) =
1+cos(180) = 0

Now spoil the feed phase (phi) by 10 degrees... (80
on boresight: 1+cos(80-90*cos(0)) = 1+cos(-10) = 1.984
on 45: 1+cos(80-90*cos(45)) = 1.959
on 90: 1+cos(80-90*cos(90)) = 1.174
at 180: 1+cos(80-90*cos(180)) = 1+cos(80+90) = 1.52E-2

The gain on boresight didn't change much... from 2 to 1.984 (0.03dB)
But the null in the back came up from zero to 1.5E-2.. (instead of -
infinity, it's now -18dB)

Change the phase error to 45 degrees...)
@theta=0: 1+cos(45-90*cos(0)) = 1.707
@theta=180: 1+cos(45-90*cos(180)) = .292

So, from the 10 degree error case, the forward gain went from 1.984 to
1.707, about 0.6dB...
but the null went from 1.52E-2 to .292 (from -17dB to -5 dB)..


The thing to remember on any gain antenna is that it takes very little
power to disrupt a null (after all, a -30dB null means that if you're
radiating 1kW in the forward direction, you're radiating 1 W in the
null.. so just another watt will double the energy in the null,
turning it from -30dB to -27dB...)

(And, you can see why making antennas with sidelobes-60dB is VERY
challenging... )


Now, change the phasing to, say, 80 degrees.. in the preferred
direction, the gain is now 1+cos(10degrees)

Thank you Jim for the explanation. Sorry I wasn't more specific.
I was refering to the difference between receiving and transmiting gain.
--
Alejandro Lieber LU1FCR
Rosario Argentina

Real-Time F2-Layer Critical Frequency Map foF2:
http://1fcr.com.ar

Alejandro Lieber wrote:

Thank you Jim for the explanation. Sorry I wasn't more specific.
I was refering to the difference between receiving and transmiting gain.

The gain effect is the same, but for a lot of radio applications, gain
is important on transmit, but less so on receive, where good back/side
performance (e.g. low gain in undesired directions) is important.

That is, my transmitter doesn't care about a strong interfering signal
from a different direction, but my receiver sure does.