Call Search
     

New to Ham Radio?
My Profile

Community
Articles
Forums
News
Reviews
Friends Remembered
Strays
Survey Question

Operating
Contesting
DX Cluster Spots
Propagation

Resources
Calendar
Classifieds
Ham Exams
Ham Links
List Archives
News Articles
Product Reviews
QSL Managers

Site Info
eHam Help (FAQ)
Support the site
The eHam Team
Advertising Info
Vision Statement
About eHam.net

   Home   Help Search  
Pages: Prev 1 2 3 4 5 6 [7] 8 Next   Go Down
  Print  
Author Topic: home-brew sigma 4 11 meter antenna  (Read 78755 times)
KE1IZ
Member

Posts: 30




Ignore
« Reply #90 on: November 15, 2013, 06:27:12 AM »

I thought I mentioned this earlier but if I didn't it may come as a surprise to some. Over the last decade I've seen many different models of this design done in EZNEC. Every one of these models struggled to produce more than unity gain in free space. Not one of these models even remotely demonstrates what we see in the field with this design. Prior to L.B. Cebik's passing he was questioned on this topic and provided a few vague clues.

He called the design a "non apparent collinear" and went on to say that modeling it in EZNEC would be difficult. He also made it clear he was reluctant to comment further as he knew it would create what he called meaningless debate and confusion that he preferred to avoid at that point. The point here is you will find striking differences in gain and element phasing when using CST Microwave Studio versus EZNEC. Apparently this has to do with the engine used by the software.

Most modeling software is based off the NEC engine and that has prevented accurate analysis of the radiation present on the cone. The in phase currents present on the outside of the cone are likely the result of constructive use of normal Common Mode Currents seen with end feds. Something NEC has a hard time displaying.

On the other hand, CST uses the MOM engine and was the first model ever to demonstrate constructive radiation with phase inversion taking place at the cone and a free space gain of 2 dbd! CST views this antenna as a 1/2 wave over a 1/4 wave while NEC views it as a 1/2 wave fed through a coaxial 1/4 wave having unity gain and no phase inversion on the outside of the cone.

Some other points worth mentioning would be the 3/4 wave groundplane you may make will have negative gain on the horizon because the bottom 1/4 wave will be allowed to radiate out of phase and cause the high angle secondary lobes to contain more energy than the main lobe. That will all change as soon as you sweep the radials upwards and turn it into the "non apparent collinear" Cebik called it.

Unfortunately I never considered adding the second 1/2 wave collinear section without any phase shift in between so I cannot comment on that particular question. I can confirm that when you use a 90 degree shift to add the top 1/2 wave that you will see virtually no change in resonance or impedance once these added elements are properly tuned. The 180 degree shift will show a horrible match. Even when you retune the gamma to rematch the antenna, it has just about the same gain as the stock version.

Just about everything you have described is exactly what I did in the field 2 summers ago. It was refreshing to see you consider the same ideas with curiosity and I hope you have the time to follow up and reach a conclusion to share with us.
« Last Edit: November 15, 2013, 06:50:12 AM by KE1IZ » Logged
WB6BYU
Member

Posts: 13486




Ignore
« Reply #91 on: November 15, 2013, 09:35:35 AM »

I'm well aware of the reported differences between EZNEC and CST results, though the
CST animated current distribution that I saw didn't show any significant difference.

One of the interesting questions is that, if the cone really does behave differently than
a standard j-pole, how many cone wires are required for the results to diverge?  I modeled
the cone with 4 wires and with 2 wires, and the results were basically the same as for
the J-pole with one wire.  I can make the wires sloping away from the radiator or parallel
to it, so there is a 2 dimensional space with the number of wires on one axis and the
angle of the wires on the other.  If the proposal is that a single parallel wire (J-pole) acts
significantly different than 4 or more angled wires (and the results can be extended to
solid skirts of either type), where is the dividing line along this plot between the two
operational modes?  Is it gradual or abrupt?  Does that mean there are intermediate
versions that have a combination of attributes?  Does it require some minimum number
of wires in the cone, and if so, how many?

This is of particular interest when planning a test:  if, for example, it requires 6 or 8
wires in the cone (or a solid sheet) in order to see the difference, then if I test with
just 4 wires it won't really give us any information.  If it only takes 2 wires then I can
do the test just by adding a second wire to a J-pole, which simplifies the effort
considerably.

The more simple and clear I can define the test beforehand, the more likely I can
find time to run it.
Logged
KE1IZ
Member

Posts: 30




Ignore
« Reply #92 on: November 15, 2013, 10:46:20 AM »

When speaking to the engineer who actually made the CST model given to me, he explained that the cone "confines the out of phase radiation on the vertical". That is different than cancellation which would only require one radial. The antenna was originally designed by Avanti with 3 radials. The new design by Sirio uses 4 radials. I can see slightly more gain in the distance on the 4 radial version. Going beyond 4 does not seem to help although the cone can be made of solid or screened metal too. The problem there is the gamma match will extend outside of the cone.

It may not be a good idea to try and feed this prototype between the vertical and one leg of the cone as you would a J-Pole. That will introduce the same type of skewing we see in the J-Pole and may also disturb the constructive currents we are trying to demonstrate on the outside of the cone. Do not gap the base of the radials away from the vertical before angling upwards. Best performance is noted when the radials are mounted directly on the vertical and the upward angle is consistent along the entire length.
Logged
WB6BYU
Member

Posts: 13486




Ignore
« Reply #93 on: November 15, 2013, 11:47:52 AM »

The simplest construction would be to solder the cone wires into the 4 mounting
holes on a SO-238 chassis-mount connector and feed the vertical portion directly
from the center pin.  The EZNEC models show a low SWR with that method (for
some combinations of lengths / angles / etc.)  Whether it works that way in
practice or not will be another data point, but mechanically it is much easier
to build.  I have an aversion to gamma matches, particularly after seeing how
badly one skewed the pattern on a quad.

Otherwise I'll use my commercial ground plane with the cone wires attached
to the sides in place of the usual radials.  It's a couple inches across the
base, but that should still work OK.  That is much stronger for mounting
the other elements on top.
Logged
KE1IZ
Member

Posts: 30




Ignore
« Reply #94 on: November 15, 2013, 03:25:19 PM »

I've never tried to isolate the base of the vertical from the base of the cone to drive it directly. Although that will work to drive the 3/4 wave ground plane version. I just question if that method may reduce the CMC flowing on the outside of the cone which is where the gain comes from. I've also noted the same skewing of the pattern when using a gamma on a balanced antenna like a Yagi. To fix that I just went to a T match on the Yagi.

I do not see very much distortion when using the gamma on this design and the antenna has been rotated while under test to confirm minimum distortion in the pattern. The distortion was reduced by keeping the gamma rod as short as possible to make the match and placing it so it's mostly inside the cone structure.
Logged
WB6BYU
Member

Posts: 13486




Ignore
« Reply #95 on: November 16, 2013, 09:57:04 AM »

One thing that EZNEC does show (as will, I suspect, any other program) is that the
pattern is very susceptible to currents on the coax and/or mast.  A quick model
suggests up to a 3dB difference (mostly degradation) at the horizon even for
relatively short lengths.  That tells me that any experiments have to be closely
controlled to give valid results.  If I can't mount the detector right at the feedpoint
then I'll need to put several ferrite cores on the coax to minimize the effects.

I wouldn't be surprised if this is a frequent cause of measured differences between
antennas in the real world.


To avoid several possible sources of error I'll probably lay the antenna horizontally
on top of a wood fence at one side of my yard, and use a transmitter into a horizontal
beam at the other side.  That will reduce the differences due to height above ground.
It may also allow me to run the coax vertically to keep it out of the pattern being
measured, as well as reducing the physical strength requirements on the test antenna
because it doesn't have to support its own weight.

That probably means I can use the same radiator wire and just make different feed
systems (coax connector with various radial wires attached) that plug onto the
feed end.
Logged
KE1IZ
Member

Posts: 30




Ignore
« Reply #96 on: November 16, 2013, 01:38:10 PM »

With all due respect, you can't learn a thing about this antenna from EZNEC. Just about every clue it provides about this design is misleading. W8JI apparently ran the same test you did in EZNEC and came up with the same incorrect idea about the antenna being a CMC nightmare where these currents would degrade performance.

Consider these facts and then apply them to what EZNEC just told you. Since the vertical radiator currents have been prevented from radiating by the cone, we have effectively delayed the phase of the vertical allowed to radiate by 90 degrees with respect to the source. That one move has now caused the currents that radiate on the vertical to be in phase with currents radiating from the radials.

It is almost certain that the constructive current from the upward swept radials is the result of Common Mode Currents since they now match the phase of the upper 1/2 wave. The vast majority of CMC will be flowing on the outside surface of the tuned resonant cone. Exactly where we want. The percentage of CMC on these redials is determined by how well the elements of the antenna have been tuned for the frequency in use.

It can be detuned to the point where movements in the coax will cause undesirable changes in resonance and impedance. It can also be properly tuned so that any change in coax length or position will not cause any noticeable change in reflected power or resonance.

We should also consider that since the phase of the vertical has been delayed to match the radiated CMC on the cone, any residual CMC descending on the begining of the mast or coax would also be in a constructive phase, contributing to gain rather than degrading it. That is confirmed in CST if you look at the currents on the short mast just below the feedpoint. It's also confirmed in the field by isolating the mast and choking the coax. Any small change in gain that may be noted will be in a negative direction if the antenna was tuned properly to begin with.

I also would be hesitant to change the orientation from vertical to horizontal. Knowing we are trying to identify collinear gain from one section being stacked over the top of another, changes in the ground reflections could be different if the antenna were placed on its side. If you don't see the results indicated by CST, I would keep it vertical and consider using a 2 meter beacon or repeater as a distant signal source to compare against. Trying to do this too close to the near field will not be as noticeable as a stable signal that is more distant. The gain is most noticeable in the fringe area of the dipole.
Logged
MARCONI390
Member

Posts: 17




Ignore
« Reply #97 on: November 16, 2013, 02:33:21 PM »

WB6BYU, thanks for considering to do this testing.

Donald has tried to convince me of his idea now for several years and this is frustrating for us both. However, my model of my S4 is as close to scale as I can duplicate, excepting the model does not use tapper.

My quandary is my model does not show the free space gain like Donald is suggesting, 4.15dbi in free space. He further claims that Eznec does not consider the cone radiating in the far field. However, the only thing I see different using Eznec is his idea that the cone constructively adds a significant amount of in phase RF to the 1/2 wave radiator.

I see some constructive contribution, but I see this minimal to insignificant at best, and I consider these effects are due to cancellation of out of phase currents in the cone area.

In fact I see hardly any currents flowing on the mast wire #39 on my model as well but that might be because I pick a good height. I agree that testing in the horizontal will help minimize ground effects.

The cancellation I see in my model is almost complete, with a very small fraction of an amp flowing in phase with the top 1/2 wave radiator, so that part of our issue agrees.

My model only shows 4.06dbi at 8* degrees using Eznec's real Earth feature and about 2.29 dbi at 18* degrees in free space.

Here are my modeling output results for image, gain, and match.

http://img547.imageshack.us/img547/7186/346.pdf

The model has quite a few wires and I would like to attach the EZ file itself, but I don't find a way to send the file. I don't see and attachment feature on eham. If you would like the file for dimensions, then let me know and I can email the EZ file to you. As you know you can then scale it Eznec to any frequency.

Email me at edromans@comcast.net

Thanks again,

Eddie - Marconi390

Logged
WB6BYU
Member

Posts: 13486




Ignore
« Reply #98 on: November 16, 2013, 04:17:44 PM »

Hi Eddie,

Well, I haven't done the tests yet, but by talking them through to find the simplest
experiment that will confirm or disprove the assertions, it makes it more likely that I will
get around to actually building a model and running the tests at some point.  And that
is part of the reason for asking so many questions, such as the number of radials required,
because that affects the complexity of the task.

If we can make the test simple enough, others can try it as well and confirm the results.

It appears at the moment that I can build a radiator and fix it in a specific location,
then plug different feed methods onto it, to eliminate many of the potential variables.
Then we can see if the cone feed gives any difference compared to a reference dipole
or a standard J-pole using one or two parallel wires.  (The 2-wire case will be an interesting
intermediate stage to see which version it appears closest to.)  Then I can extend the
radiator with a standard quarter wave stub and half wave and see if I get the same
results.  If necessary I can make a vertical version and compare it to a more distant
signal source, though 90' should be a reasonable approximation of far field on 2m.

One of the hardest parts will be to prevent the common mode current from affecting
the readings - I have a receiver that I can connect directly to the feedpoint that will
give me comparative readings via audio output, but it isn't as easy to document the
results that way.  I may end up using my signal generator for the source and varying
the output in 0.1dB steps, though I don't know if the final results can be trusted to
better than 0.5dB or so.


Fortunately I've been active in transmitter hunting for a long time, so I have receivers
that are designed to give a good indication of signal strength, along with transmitter
boxes that I can use for remote signal sources (several miles away if needed.) 
If I can get the common mode current reduced sufficiently then I can borrow a network
analyzer and use that for the measurements, which will give better accuracy.   But
the more work, the less likely I'm going to get the measurements made:  I still have
to build the reference antennas, etc.
Logged
MARCONI390
Member

Posts: 17




Ignore
« Reply #99 on: November 16, 2013, 04:50:05 PM »

With all due respect, you can't learn a thing about this antenna from EZNEC. Just about every clue it provides about this design is misleading. W8JI apparently ran the same test you did in EZNEC and came up with the same incorrect idea about the antenna being a CMC nightmare where these currents would degrade performance.

Consider these facts and then apply them to what EZNEC just told you. Since the vertical radiator currents have been prevented from radiating by the cone, we have effectively delayed the phase of the vertical allowed to radiate by 90 degrees with respect to the source. That one move has now caused the currents that radiate on the vertical to be in phase with currents radiating from the radials.

It is almost certain that the constructive current from the upward swept radials is the result of Common Mode Currents since they now match the phase of the upper 1/2 wave. The vast majority of CMC will be flowing on the outside surface of the tuned resonant cone. Exactly where we want. The percentage of CMC on these redials is determined by how well the elements of the antenna have been tuned for the frequency in use.

It can be detuned to the point where movements in the coax will cause undesirable changes in resonance and impedance. It can also be properly tuned so that any change in coax length or position will not cause any noticeable change in reflected power or resonance.

We should also consider that since the phase of the vertical has been delayed to match the radiated CMC on the cone, any residual CMC descending on the begining of the mast or coax would also be in a constructive phase, contributing to gain rather than degrading it. That is confirmed in CST if you look at the currents on the short mast just below the feedpoint. It's also confirmed in the field by isolating the mast and choking the coax. Any small change in gain that may be noted will be in a negative direction if the antenna was tuned properly to begin with.

I also would be hesitant to change the orientation from vertical to horizontal. Knowing we are trying to identify collinear gain from one section being stacked over the top of another, changes in the ground reflections could be different if the antenna were placed on its side. If you don't see the results indicated by CST, I would keep it vertical and consider using a 2 meter beacon or repeater as a distant signal source to compare against. Trying to do this too close to the near field will not be as noticeable as a stable signal that is more distant. The gain is most noticeable in the fringe area of the dipole.

Hey Donald, if you're going to use a reference from W8JI, could you please post the link, so everybody could check it out?

Maybe you could also give us a link to your reference to Cebik's words too...where you claim he talked about an "non-apparent collinear effect." So far I've found no reference to the idea. Is this phrase also possibly non-apparent too? The word didn't even pass the spell checker on eHam. Are you sure the word wasn't unapparent? That is what the spell checker suggested.

A very simple little link to such references could really change my mind on all of this mystique in a flash.
Logged
KE1IZ
Member

Posts: 30




Ignore
« Reply #100 on: November 16, 2013, 06:25:28 PM »

Marconi, I'm not sure why it is you imply I've been dishonest so many times during your misinterpretation of the CST model. You've regularly hinted many times that what I've said is not truthful. Then you post these question to me like I'm suppose to remember every time date and spot  that I've read Tom posting in so I can prove I didn't make it up?

You act like you didn't grill Bob countless times over his email exchanges with Mr. Cebik before he passed and you're unaware of the source of this information. Ask Bob to forward you the emails between Cebik and himself if you're so distrustful. It would be a much more appropriate approach than the way you've handled this matter with both Bob and myself.

Rather than just stopping short of calling me a liar, if you can't find the things I've quoted Tom as saying about the Vector design online for yourself, email him and ask him directly. I think he's a member on this forum too.

CST confirming in phase current on each of the individual radials at approximately 25% of the magnitude of the vertical years after Cebik passed didn't change your mind about his "non apparent collinear" analysis. Seeing the written words and possible typos of Cebik or asking W8JI if he believes CMC on this design is detrimental to its gain will never change your opinion. Just like I couldn't convince you testing on SSB and with the antennas you are comparing against within each others near field was worthless even if it was the only way you could do it.
Logged
MARCONI390
Member

Posts: 17




Ignore
« Reply #101 on: November 16, 2013, 07:28:17 PM »

Marconi, I'm not sure why it is you imply I've been dishonest so many times during your misinterpretation of the CST model. You've regularly hinted many times that what I've said is not truthful. Then you post these question to me like I'm suppose to remember every time date and spot  that I've read Tom posting in so I can prove I didn't make it up?

You act like you didn't grill Bob countless times over his email exchanges with Mr. Cebik before he passed and you're unaware of the source of this information. Ask Bob to forward you the emails between Cebik and himself if you're so distrustful. It would be a much more appropriate approach than the way you've handled this matter with both Bob and myself.

Rather than just stopping short of calling me a liar, if you can't find the things I've quoted Tom as saying about the Vector design online for yourself, email him and ask him directly. I think he's a member on this forum too.

CST confirming in phase current on each of the individual radials at approximately 25% of the magnitude of the vertical years after Cebik passed didn't change your mind about his "non apparent collinear" analysis. Seeing the written words and possible typos of Cebik or asking W8JI if he believes CMC on this design is detrimental to its gain will never change your opinion. Just like I couldn't convince you testing on SSB and with the antennas you are comparing against within each others near field was worthless even if it was the only way you could do it.

Donald, as far as I can remember, I was ill at the time and had not been on the forum for a spell. I think I recall that when I got back, that is when I first learned about Bob and his contacts with Cebik. FWIR, I always thought Bob and Cebik talked on the telephone and until now I did not know for sure there ever were emails.

Do you have or have you seen such emails?

That would answer a lot of questions if they are available. I guess Bob would know for sure, and he is welcome to show me proof anytime. I would love to finally see what was said between them and it would prove a lot to support Bob's story. I have never said a thing to Bob about showing me the email record, and he has never said a thing to me about his emails to Cebik.

If such a record is still on the WWDX forum back from those days then you or anybody else could check that out. You can bet that I will try and find out if it happend as you say here. But I don't want to hang out any of my dirty laundry here on eham. I hope you've OK'ed this revelation with Bob.

So, I say no more on the subject, except that I am surprised to learn this now after all these years.
Logged
KE1IZ
Member

Posts: 30




Ignore
« Reply #102 on: November 16, 2013, 07:48:59 PM »

Stop trying to pit me and Bob against each other. You know more about Bob's exchanges with Cebik than I do. I didn't know he actually got to speak with him. Bob has also recently indicated to you he's not pleased with the way you reword the things he's described to you to fit your thoughts rather than attempt to understand them.

I'm just curious, why would seeing Cebik's opinion on the antenna outweigh CST's confirmation of his opinion years later? What's to prevent you from just saying "you typed it up" like you refer to the CST model as "my animation"? Cebik's opinion to me just gave the encouragement that Bob and myself were on the right track. CST and the collinear field testing with adjustable phase shift proved it.
« Last Edit: November 16, 2013, 07:55:17 PM by KE1IZ » Logged
WB6BYU
Member

Posts: 13486




Ignore
« Reply #103 on: November 16, 2013, 08:45:55 PM »

I've been trying to remain objective and not import past history from elsewhere
into this discussion.

I'm sure that, whatever the result of these tests (whoever runs them), someone
won't be happy with the result and will insist they were not run properly.
Clearly a number of people have a lot invested on both sides, and my results
certainly won't end the debate.

So I'm going into this knowing that I'll likely get flak from someone regardless of
how well I run the tests.  Well, life's like that some times.

What I'm hoping to do is to set an example of how debates about antennas can be
resolved (to some extent) by simple testing, and that, while I happen to have
access to some specialized equipment, it really isn't that difficult for just about
anyone to run such an experiment.  In this case I'm building 2m scale models of
the antenna in question and appropriate reference antennas and comparing the
results.  It's similar to what I recommend for those who want to compare HT
antennas or anything else:  find a weak signal and try the two antennas to see
which one seems to work better.  Sometimes the radiation patterns are different,
and there may be other differences that affect the accuracy of the results, but,
if one antenna is significantly better than another, we should be able to measure
that difference.  Even a 1dB step attenuator and a coax switch may be enough
to get a reasonable estimate of the difference.

That's one reason for taking care in explaining how I plan to run the tests
beforehand - I want to make sure that I have agreement that the results will be
sufficient to test both competing theories.  And if I can make the test simple enough
that 8 others can run it on their own, that will give us even more information:  if
they all get the same result, in spite of individual differences in equipment and
technique, that's pretty good support for one of the theories over the other.  If
they get mixed results, then clearly the actual performance is dependent on some
other characteristics that we haven't controlled.  In that case, we look further
into the issue.


I've seen enough cases on eHam over the years where folks argue over the
performance of relatively simple antennas, where everyone could have tested
their assumptions in less time than they spent arguing about it.  So this seemed
like a good opportunity to set an example about testing a hypothesis rather
than arguing about it. 


To do that, we look at the two competing theories / models to find a particular
situation where they predict different performance and try to reproduce that
condition.  In this case there appear to be two differences:

1) about 2dB difference in claimed gain for the same antenna

2) a difference in the required phase shift when adding another half wave element
in phase with the original antenna.

The first one is the main purpose of this experiment, but in the process I may have an
opportunity to try out the second.  One would expect that the results of the two cases
would both point in the same direction, but, if not, then perhaps we need to develop
a third theory to explain the results...
Logged
KE1IZ
Member

Posts: 30




Ignore
« Reply #104 on: November 16, 2013, 09:01:26 PM »

All I can say is thank you for being objective and keeping an open mind.
Logged
Pages: Prev 1 2 3 4 5 6 [7] 8 Next   Go Up
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.11 | SMF © 2006-2009, Simple Machines LLC Valid XHTML 1.0! Valid CSS!