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 69459 times)
KE1IZ
Member

Posts: 30




Ignore
« Reply #60 on: February 14, 2013, 09:55:30 PM »

I'm surprised someone with a good understanding of collinear antenna design has not chimed in here yet. Perhaps a specific question aimed at getting to the heart of the design might spark some comments? If the Sigma is just a cool looking 1/2 wave radiator why can't we see any gain improvement in the field whatsoever by adding a 180 degree phase delay to the top and another 1/2 wave radiator above it?

Another way to ask the same question would be: Why does the gain on the horizon only peak out when the phase delay between collinear sections is reduced to 90 degrees? Can anyone provide a valid theory on how this can be possible other than we were dealing with a 270 degree in phase radiator to begin with? I can't be the only one here who can see the logic behind my field tests and resulting comments.

This is a technically orientated site. Can't we get some realistic thoughts going on as to what could be happening here? I've shared more information about the design then Avanti ever did with the CST models and discoveries regarding proper phasing of a collinear version. I can't see where anyone can say it's a 1/2 wave J-Pole unless one just assumes everything I said is false.

In that case I ask how hard would it be to prove me wrong? You can't do it in EZNEC but you can consistently prove my findings correct in the field by making a collinear version with an adjustable phasing section that will go between 90 and 180 degrees of delay. When you get around 90 degrees and see the gain peak, ask yourself how it could have only been a 1/2 wave to start with?
« Last Edit: February 14, 2013, 10:53:53 PM by KE1IZ » Logged
MARCONI390
Member

Posts: 14




Ignore
« Reply #61 on: February 17, 2013, 09:03:47 AM »

Quote from: KE1IZ

We still have deep red and blue colors on the outside of the cone that indicates the maximum radiation current of 2.37 amps is still there in spite of the opposing currents inside the cone...



But that field strength is only when you are very close to one of the conductors in
the cone.  In the same way you can have high field strength around each wire in
a length of open wire line, but the net radiation is nearly zero when the currents are
properly balanced.  The same applies to the conductors in the cone:  when you are
much closer to one wire than the rest (whether on the inside or the outside) the field
strengths will be high, but it drops off quickly with distance.  That, and the fact that
the field is stronger near the end of the cone where the current are lower, would
suggest that it doesn't correspond with far-field radiation.

However, EZNEC will give you near-field levels and currents in each element - it would
be interesting to compare those with the results from the CST model to see if there
are any significant differences.


Dale, KE1IZ and I have an ongoing discussion on how Eznec handles the currents for the Vector.

Do you have any interest in following up on your suggestion...as noted above in bold type?

If you do...I can email you my Eznec model of the Vector to scale as boilerplate, and maybe save you some work. If so, then email me at my email address in my profile.

The problem I have with this New Vector 4000 model compared to the CST model is the current phase in the Eznec tabular currents log does not look to agree with the in-phase currents noted in the antenna view, similar to what KE1IZ is suggesting. My problem may be modeling error(s) or a lack of understanding on my part, but I've seen this phase issue crop up before with some of my other Eznec models, and I can't explain it. I tried to talk to Roy about it, but I could not seem to get my point across, and he does not allow for time to critique user's modeling techniques.

If you can help some more on this matter, then thanks in advance.
« Last Edit: February 17, 2013, 09:07:03 AM by MARCONI390 » Logged
WB6BYU
Member

Posts: 13172




Ignore
« Reply #62 on: February 17, 2013, 06:31:13 PM »

Quote from: KE1IZ

I'm surprised someone with a good understanding of collinear antenna design has not chimed in here yet.



Several of them have participated on this thread, but you don't appear
to have listened to them.  The principles are pretty simple, and well covered
in any of the standard reference books.

Since you have an endfed half wave element, you know that the current
distribution will be on it.  If adding a second colinear element doesn't give
the expected gain, then your phase shift network isn't working as expected.

What are you using for a 180 degree phasing line?  A 90 degree long shorted
stub?  180 degree long stub?  Parallel tuned circuit?  Half wavelength of
thinner diameter material than the radiator?  Some sort of coaxial decopler?

There are some such methods sometimes seen in ham antenna designs that
don't reliably give the expected results.  They might if the center point of
the stub could be grounded, but that rarely is practical.

It is possible, of course, that you have common mode currents on your
feedline, and they are affecting the radiation pattern.  Some of these
details we can't determine from a distance.  If I had to guess a cause,
I'd first make sure you were using a 1/4 wavelength stub (which
contains 1/2 wavelength of conductor) as your 180 degree phasing
transformer.  That doesn't guarantee you 180 degrees, but will work
better than using a 1/2 wave stub, which would give you 0 degrees
of phase shift.


Quote

... but you can consistently prove my findings correct in the field by making a collinear version with an adjustable phasing section that will go between 90 and 180 degrees of delay. When you get around 90 degrees and see the gain peak, ask yourself how it could have only been a 1/2 wave to start with?



The required phase shift has nothing to do with whether the original antenna
is 1/2 wavelength or something else.  If the radiation from the cone is in phase
with the radiator, and makes a significant contribution to the overall radiation
from the antenna (as you claim) then the added colinear element still has to be
in phase with the rest of the antenna via a 180 degree phase shift.  In fact,
if 90 degrees really gives you more gain at the horizon, then there must be
some part of the antenna that is radiating MORE than the half wave "radiator"
at a different phase:  the principles of phase combination / cancellation would
REQUIRE it for the antenna to behave that way.  That would mean that the
radiation from the cone is NOT in phase with the radiator above it, or that
the mast or coax is radiating more than the "radiator" and cone put together.

The simplest explanation is that your phase shifter isn't shifting the way you
expect it to.



Quote from: MARCONI390
Quote from: WB6BYU

However, EZNEC will give you near-field levels and currents in each element - it would
be interesting to compare those with the results from the CST model to see if there
are any significant differences.


Dale, KE1IZ and I have an ongoing discussion on how Eznec handles the currents for the Vector.

Do you have any interest in following up on your suggestion...as noted above in bold type?

...

If you can help some more on this matter, then thanks in advance.



I don't have CST handy, but I'd suggest starting by having it print out a table
of element currents and phases just as EZNEC does, and comparing them.  Just
looking at the colors on the display doesn't tell you everything because it is
near-field field intensity, and doesn't necessarily correspond with far-field
radiation.  My guess is that the two models will give similar currents in each
of the parts, but you'll need to get the tabular data out of CST to know for
sure.
Logged
MARCONI390
Member

Posts: 14




Ignore
« Reply #63 on: February 17, 2013, 08:01:45 PM »

I've requested the CST tabular currents information for the New Vector 4000, but I think Sirio's business policies will likely reject such a request.

Thanks for your reply.
Logged
KE1IZ
Member

Posts: 30




Ignore
« Reply #64 on: February 19, 2013, 04:08:26 PM »

Several of them have participated on this thread, but you don't appear
to have listened to them.  The principles are pretty simple, and well covered
in any of the standard reference books.
Quote

It's not as though I've ignored any of them. It's just that what has been said does not line up at all with 15 years of field testing. In countless applications commercial center fed 1/2 wave dipoles were replaced with the design, all noting improved coverage. Many designs are well covered in the reference books.


Extensive searching on this design will reveal a brief publication in 1939 and Herbs Avanti patent 40 years later that also references the 1939 work. It took another 40 years before any significant new research was done on the design. That was when I requested permission from Sirio to publish the new CST modeling results on the design.

Virtually no other information exists anywhere on this antenna. The closest cousin you will find are the links to the Skeleton sleeve monopole. If you are aware of any other 3/4 wave vertical that can confine the out of phase radiation in the lower 1/4 wave and replace it with in phase currents that can be modeled and confirmed in CST Microwave Studio, please point it out.

Since you have an endfed half wave element, you know that the current
distribution will be on it.  If adding a second colinear element doesn't give
the expected gain, then your phase shift network isn't working as expected.
Quote

That all assumes we had a simple 180 degree vertical radiator to start with. CST, Cebik, Herb, and many others including myself, strongly suggest we do not.

What are you using for a 180 degree phasing line?  A 90 degree long shorted
stub?  180 degree long stub?  Parallel tuned circuit?  Half wavelength of
thinner diameter material than the radiator?  Some sort of coaxial decopler?
Quote

Roy asked the same questions when I pointed out the discrepancies between his EZNEC+ program and CST. I assure you the phase delay that works is an electrical 90 degrees long. To keep things as simple as possible all designs of phase delays were tried except multi turn coils and parallel resonant circuits.

The phasing section was a thin electrical 1/2 wavelength conductor, folded over at the far end and connected across the insulator between collinear sections. The length of that section had to be cut in half before the gain of the collinear peaked well over that of the original. The length that peaks with EZNEC software is off by 100% longer.

There are some such methods sometimes seen in ham antenna designs that
don't reliably give the expected results.  They might if the center point of
the stub could be grounded, but that rarely is practical.

It is possible, of course, that you have common mode currents on your
feedline, and they are affecting the radiation pattern.  Some of these
details we can't determine from a distance.  If I had to guess a cause,
I'd first make sure you were using a 1/4 wavelength stub (which
contains 1/2 wavelength of conductor) as your 180 degree phasing
transformer.  That doesn't guarantee you 180 degrees, but will work
better than using a 1/2 wave stub, which would give you 0 degrees
of phase shift.

The required phase shift has nothing to do with whether the original antenna
is 1/2 wavelength or something else.  If the radiation from the cone is in phase
with the radiator, and makes a significant contribution to the overall radiation
from the antenna (as you claim) then the added colinear element still has to be
in phase with the rest of the antenna via a 180 degree phase shift. 
Quote

If the required phase shift has nothing to do with the radiator length below it then wouldn't all collinear antennas use a 180 degree delay? This makes sense when stacking 1/2 wave radiators but there are other designs to consider. Does a 5/8 wave over a 1/4 wave use a 180 degree delay?

In fact,if 90 degrees really gives you more gain at the horizon, then there must be
some part of the antenna that is radiating MORE than the half wave "radiator"
at a different phase.
Quote

This theory would require the original antenna to have less than unity gain and contradicts the CST model along with field tests. On the flip side of this idea is that the antenna radiates in phase along it's length and that using a 180 degree delay causes 50% of the additional 1/2 wave section to buck the phase below it. 

The principles of phase combination / cancellation would
REQUIRE it for the antenna to behave that way.  That would mean that the
radiation from the cone is NOT in phase with the radiator above it, or that
the mast or coax is radiating more than the "radiator" and cone put together.

The simplest explanation is that your phase shifter isn't shifting the way you
expect it to.
Quote


There is no reason to guess about the phase or magnitude of radiation taking place on the cone. The work has been done in CST and presented for anyone to examine. It very clearly demonstrates the cone contributes in phase radiation to the 1/2 wave above it. If you want to experiment with modeling this antenna you will have to leave your trusty EZNEC behind and try CST Microwave Studio.
Logged
MARCONI390
Member

Posts: 14




Ignore
« Reply #65 on: February 20, 2013, 09:09:05 PM »

Donald here are my Eznec results for my Sigma 4 in Free Space, with an average gain of 0.998 = -0.01 db, and I have the currents turned on. If you need the tabular results I will post that too, but it is pretty long.

You tell us such a model using Eznec will show 0 unity gain or less in free space, and that Eznec does not show the current phase similar to CST. I see out of phase currents in the bottom cone area, and they look to cancel with a fractional adantage for the radial currents that show to be in phase and maybe combining with the top 1/2 wave element, but the radiation difference is very small.

http://img542.imageshack.us/img542/5241/76215424.pdf
« Last Edit: February 20, 2013, 09:21:06 PM by MARCONI390 » Logged
WB6BYU
Member

Posts: 13172




Ignore
« Reply #66 on: February 20, 2013, 09:44:21 PM »

I suspect that joining the tops of the cone with just 4 wires (or
perhaps an octagon with 8 segments) is probably a good enough
model rather than using  as many as you did, but it shouldn't
hurt anything as long as the number of segments doesn't slow
things down too much.


I've been looking at the use of the 90 degree phase shift:  my
first question is whether the SWR changed significantly between
the original antenna and the extended one with both the 1/4 and
1/8 wave phasing stubs.
Logged
KE1IZ
Member

Posts: 30




Ignore
« Reply #67 on: February 20, 2013, 10:11:50 PM »

Donald here are my Eznec results for my Sigma 4 in Free Space, with an average gain of 0.998 = -0.01 db, and I have the currents turned on. If you need the tabular results I will post that too, but it is pretty long.

You tell us such a model using Eznec will show 0 unity gain or less in free space, and that Eznec does not show the current phase similar to CST. I see out of phase currents in the bottom cone area, and they look to cancel with a fractional adantage for the radial currents that show to be in phase and maybe combining with the top 1/2 wave element, but the radiation difference is very small.

http://img542.imageshack.us/img542/5241/76215424.pdf
The 2.27 dbi in free space at 17 degrees sure sounds like less than unity gain to me at zero degrees. One reason this angle is so high is because the cone has not been correctly positioned to confine all of the out of phase currents at the base of the vertical.

The crossover point in phase along the vertical should occur right at the top of the cone. If you do this, the angle will come down closer to 0 degrees but the gain does not improve. No significant radiation is shown from the cone in EZNEC. That's not what I see in CST and in the field.
Logged
MARCONI390
Member

Posts: 14




Ignore
« Reply #68 on: February 20, 2013, 10:56:05 PM »

Donald here are my Eznec results for my Sigma 4 in Free Space, with an average gain of 0.998 = -0.01 db, and I have the currents turned on. If you need the tabular results I will post that too, but it is pretty long.

You tell us such a model using Eznec will show 0 unity gain or less in free space, and that Eznec does not show the current phase similar to CST. I see out of phase currents in the bottom cone area, and they look to cancel with a fractional adantage for the radial currents that show to be in phase and maybe combining with the top 1/2 wave element, but the radiation difference is very small.

http://img542.imageshack.us/img542/5241/76215424.pdf
The 2.27 dbi in free space at 17 degrees sure sounds like less than unity gain to me at zero degrees. One reason this angle is so high is because the cone has not been correctly positioned to confine all of the out of phase currents at the base of the vertical.

The crossover point in phase along the vertical should occur right at the top of the cone. If you do this, the angle will come down closer to 0 degrees but the gain does not improve. No significant radiation is shown from the cone in EZNEC. That's not what I see in CST and in the field.

I knew you wouldn't even consider it.

Here is the pattern with the cursor set at 0 degrees. The gain does come down some, but it doesn't show unity or less at 0 degrees. You can figure that out just looking at the pattern. How do you know for sure what the maximum angle noted for the CST pattern is?

http://img248.imageshack.us/img248/4539/sigma4infreespaceat0deg.pdf

This model is set to the dimensions for the Antenna Specialists Sigma 4. The only exception is I removed the tappered elements for the radiator. I set the radiator diameter at .875 as an average and the model is what it is. I can easily change this diameter.

Describe for me the fix to reposition the cone you mention above, and I'll fix the model, and well see if it does what you suggest.
« Last Edit: February 20, 2013, 10:59:41 PM by MARCONI390 » Logged
KE1IZ
Member

Posts: 30




Ignore
« Reply #69 on: February 21, 2013, 07:56:52 AM »

Donald here are my Eznec results for my Sigma 4 in Free Space, with an average gain of 0.998 = -0.01 db, and I have the currents turned on. If you need the tabular results I will post that too, but it is pretty long.

You tell us such a model using Eznec will show 0 unity gain or less in free space, and that Eznec does not show the current phase similar to CST. I see out of phase currents in the bottom cone area, and they look to cancel with a fractional adantage for the radial currents that show to be in phase and maybe combining with the top 1/2 wave element, but the radiation difference is very small.

http://img542.imageshack.us/img542/5241/76215424.pdf
The 2.27 dbi in free space at 17 degrees sure sounds like less than unity gain to me at zero degrees. One reason this angle is so high is because the cone has not been correctly positioned to confine all of the out of phase currents at the base of the vertical.

The crossover point in phase along the vertical should occur right at the top of the cone. If you do this, the angle will come down closer to 0 degrees but the gain does not improve. No significant radiation is shown from the cone in EZNEC. That's not what I see in CST and in the field.

I knew you wouldn't even consider it.

Here is the pattern with the cursor set at 0 degrees. The gain does come down some, but it doesn't show unity or less at 0 degrees. You can figure that out just looking at the pattern. How do you know for sure what the maximum angle noted for the CST pattern is?

http://img248.imageshack.us/img248/4539/sigma4infreespaceat0deg.pdf

This model is set to the dimensions for the Antenna Specialists Sigma 4. The only exception is I removed the tappered elements for the radiator. I set the radiator diameter at .875 as an average and the model is what it is. I can easily change this diameter.

Describe for me the fix to reposition the cone you mention above, and I'll fix the model, and well see if it does what you suggest.
Don't be so quick to assume things Eddy because the model you just presented to us confirms exactly what I've said about EZNEC unable to show any gain over a 1/2 wave. The problem is you don't understand the difference between db over a dipole and db over isotropic and that is causing you to think that 1.64 dbi at 0 degrees is showing gain. When in fact you would have to get to 2.15 db over isotropic just to reach the unity gain of a dipole in free space. To fix the angle in your EZNEC model, shorten the whip or lengthen the cone.
Logged
KE1IZ
Member

Posts: 30




Ignore
« Reply #70 on: February 21, 2013, 08:30:32 AM »

I've been looking at the use of the 90 degree phase shift:  my
first question is whether the SWR changed significantly between
the original antenna and the extended one with both the 1/4 and
1/8 wave phasing stubs.

Thanks for taking the time to look at this because you are touching on some interesting areas here. When the original antenna was tuned for resonance and 50 ohms and a 180 degree phase shift was used to stack the next half wave, it messed up everything in terms of resonance and impedance. Took a good deal of time to tune much of that reactance back out in the gamma and still had no usable gain over the original.

Nothing changes accept the gain goes up if you try the same thing with a 90 degree delay. All of the original tuning settings continue to show an excellent match and the gain increases about 2 more db. Very similar to what I recalled from tuning my first Hy-Gain 5/8 wave over 5/8 wave 2 meter antenna years earlier. That tells me that the 90 degree delay is key in this design and what I had to refile at the PTO on after seeing EZNEC did not match up with the field tests.

Even knowing this when I look at the CST plot and see that perfect 180 degree radiation above the cone, logic says we need to shift the phase 180 degree before we could drive another section. My only theory for this not being the case is the top of the original antenna is 270 degrees from the source and as Cebik said some "non apparent collinear" action is taking place at the cone. This somehow creates a situation where EZNEC is off by 100% on the phase delay. Wish I knew why.
Logged
WB6BYU
Member

Posts: 13172




Ignore
« Reply #71 on: February 21, 2013, 10:13:19 AM »

Quote from: KE1IZ

Thanks for taking the time to look at this because you are touching on some interesting areas here. When the original antenna was tuned for resonance and 50 ohms and a 180 degree phase shift was used to stack the next half wave, it messed up everything in terms of resonance and impedance. Took a good deal of time to tune much of that reactance back out in the gamma and still had no usable gain over the original.

Nothing changes accept the gain goes up if you try the same thing with a 90 degree delay. All of the original tuning settings continue to show an excellent match and the gain increases about 2 more db. Very similar to what I recalled from tuning my first Hy-Gain 5/8 wave over 5/8 wave 2 meter antenna years earlier. That tells me that the 90 degree delay is key in this design and what I had to refile at the PTO on after seeing EZNEC did not match up with the field tests.

Even knowing this when I look at the CST plot and see that perfect 180 degree radiation above the cone, logic says we need to shift the phase 180 degree before we could drive another section. My only theory for this not being the case is the top of the original antenna is 270 degrees from the source and as Cebik said some "non apparent collinear" action is taking place at the cone. This somehow creates a situation where EZNEC is off by 100% on the phase delay. Wish I knew why.



You can't just blame it on EZNEC, as any standard engineering analysis of the
antenna would predict the same thing.  This goes back long before the advent
of antenna modeling:  Kraus, Laporte, Bruce, Jasik, and countless others over
the years have analyzed and used stubs,  and I haven't yet found any examples
where the standard analysis didn't hold up.

Have you actually modeled the antenna with and without the extension using
CST, or are you just looking at that plot provided by the manufacturer?  That
plot can be quite misleading because it shows field strength in the near field,
and doesn't necessarily correlate with the actual far-field radiation pattern. 
(As I said before, the fields are quite strong in the middle of a balanced feedline,
but the total radiation is low.)

Remember that a shorted stub does more than provide a phase shift, and the
effect will vary depending on where it is inserted into an antenna.  For a
pure transmission line stub a 1/8 wave shorted stub would look like an inductor
with a reactance equal to the transmission line impedance, so probably around
300 - 600 ohms for typical stubs made of aluminum tubing.  That only holds
when the currents are balanced in both sides of the stub, whoever, so it has
to be centered on a point of maximum current in the middle of the shorting bar.

Do you have the extended antenna up and functional?  If so, can you send me
the dimensions for it?  It shouldn't be difficult to build a scale model for 2m using
#12 solid copper wire and try it out.

I'm hosting an antenna workshop this Saturday where other hams can come over
and build antennas, and that might be a good time to try it.  We'll have the ability
to measure antenna gain, and I think I can find my 2m current sniffer. 
Unfortunately the barn isn't heated, and the weather is supposed to be stormy,
so I suspect everyone will be in a hurry to finish up and go home.  After that I'm
packing up my equipment in preparation for a move to a new house, so it may be
months before I can get back to it (and the rest of the pile of projects that have
been stacking up.)


But as I understand it, your observation is this:  with a standard J-pole, or any
other common half-wave radiator, one can add a quarter wave shorted phasing
stub and a second half wave element and the radiation will be in phase, giving
a gain increase at the horizon.  But if the antenna is matched with a tapered
cone in place of the tapped matching stub of a J-pole, then the shorted phasing
stub has to be 1/8 wavelength long to achieve the same effect.  Is that correct?

If so, it shouldn't be difficult to build a common radiator structure so we can just
vary the shape of the matching stub at the base between a single parallel wire
and a 4-wire cone to see how much difference it makes.  One approach might be
to build the antenna on an SO-238 socket with the center radiator attached to a
banana plug in the center and the cone build on the outer shell of a PL-259 that
screws onto the outside.

Or we could build a phasing stub and extension that works when attached to the
top of a J-pole, and then try it on the cone version and see if it acts the same
way.

If I made an open cylinder out of some tin cans, would you expect it to act more
like the conical stub or parallel wires?  That would be an easy test to run once the
basic structure was in place, as would the "open sleeve" version with 4 parallel
wires.

Maybe I can get some of the other hams at work interested in the problem and
we can run some tests in the lab using the network analyzer, but that probably
would need to be scaled for 440, as the 2m version is getting a bit tall and
the ceiling may detune it.
Logged
KE1IZ
Member

Posts: 30




Ignore
« Reply #72 on: February 21, 2013, 04:37:55 PM »

Quote from: KE1IZ

Thanks for taking the time to look at this because you are touching on some interesting areas here. When the original antenna was tuned for resonance and 50 ohms and a 180 degree phase shift was used to stack the next half wave, it messed up everything in terms of resonance and impedance. Took a good deal of time to tune much of that reactance back out in the gamma and still had no usable gain over the original.

Nothing changes accept the gain goes up if you try the same thing with a 90 degree delay. All of the original tuning settings continue to show an excellent match and the gain increases about 2 more db. Very similar to what I recalled from tuning my first Hy-Gain 5/8 wave over 5/8 wave 2 meter antenna years earlier. That tells me that the 90 degree delay is key in this design and what I had to refile at the PTO on after seeing EZNEC did not match up with the field tests.

Even knowing this when I look at the CST plot and see that perfect 180 degree radiation above the cone, logic says we need to shift the phase 180 degree before we could drive another section. My only theory for this not being the case is the top of the original antenna is 270 degrees from the source and as Cebik said some "non apparent collinear" action is taking place at the cone. This somehow creates a situation where EZNEC is off by 100% on the phase delay. Wish I knew why.


You can't just blame it on EZNEC, as any standard engineering analysis of the
antenna would predict the same thing.  This goes back long before the advent
of antenna modeling:  Kraus, Laporte, Bruce, Jasik, and countless others over
the years have analyzed and used stubs,  and I haven't yet found any examples
where the standard analysis didn't hold up.

Have you actually modeled the antenna with and without the extension using
CST, or are you just looking at that plot provided by the manufacturer?  That
plot can be quite misleading because it shows field strength in the near field,
and doesn't necessarily correlate with the actual far-field radiation pattern.  
(As I said before, the fields are quite strong in the middle of a balanced feedline,
but the total radiation is low.)

I do not have CST so I have not been able to do this. You're beginning to make me think CST may have the same issue with identifying the correct phase shift here too. In any event, I do see noticeable in phase radiation on the outside of this cone. Especially when you consider these near field currents we see on the cone are divided between multiple radials and we are comparing them to a single radiator taking the full current in the middle.

Remember that a shorted stub does more than provide a phase shift, and the
effect will vary depending on where it is inserted into an antenna.  For a
pure transmission line stub a 1/8 wave shorted stub would look like an inductor
with a reactance equal to the transmission line impedance, so probably around
300 - 600 ohms for typical stubs made of aluminum tubing.  That only holds
when the currents are balanced in both sides of the stub, whoever, so it has
to be centered on a point of maximum current in the middle of the shorting bar.


I'm in complete agreement here. the 90 degree long folded phasing section did in fact radiate noticeably in the horizontal plane and that cause the vertical omni pattern to become distorted as well. To alleviate this problem the phasing section was simply shaped into one big loop around the antenna.

There probably is still some slight radiation occurring in the phasing loop but the omni pattern is restored. Even more gain was found in stretching the insulator length to gain more spacing between sections.

Do you have the extended antenna up and functional?  If so, can you send me
the dimensions for it?  It shouldn't be difficult to build a scale model for 2m using
#12 solid copper wire and try it out.

I had a 2 meter collinear version up for a while. It's down now but I have some measurements for you at 146 Mhz. The main radiator is 63.1 inches from the connector with the mounts for the 4 radials to the top where the insulator goes. Each of the 4 radials are 19.75 inches long. The loop is 17.75 inches around the top of the radials.

The insulator between collinear sections was 12 inches but longer probably would continue to increase gain. The phasing section was just a 1/4 wavelength (90 degree long) piece of wire wrapped around the insulator in a single turn with one end connected at the bottom of the insulator and the other end connected to feed the top radiator at 39.75 inches.

The antenna should be matched with a gamma and tuned to give a flat VSWR before attaching the phasing section and upper element. Then when you add the collinear section, you adjust the phasing loop length to give a good VSWR to drive the 1/2 wave above it.

I'm hosting an antenna workshop this Saturday where other hams can come over
and build antennas, and that might be a good time to try it.  We'll have the ability
to measure antenna gain, and I think I can find my 2m current sniffer.  
Unfortunately the barn isn't heated, and the weather is supposed to be stormy,
so I suspect everyone will be in a hurry to finish up and go home.  After that I'm
packing up my equipment in preparation for a move to a new house, so it may be
months before I can get back to it (and the rest of the pile of projects that have
been stacking up.)

The collinear version of this Coaxial J-Pole I've outlined is under patent pending protection and I ask that people respect that by not coping the intellectual property for resale. At the same time, anyone who is interested in building one for their personal use is strongly encouraged to do so. That's what the hobby is about. Learning and experimenting.




But as I understand it, your observation is this:  with a standard J-pole, or any
other common half-wave radiator, one can add a quarter wave shorted phasing
stub and a second half wave element and the radiation will be in phase, giving
a gain increase at the horizon.  But if the antenna is matched with a tapered
cone in place of the tapped matching stub of a J-pole, then the shorted phasing
stub has to be 1/8 wavelength long to achieve the same effect.  Is that correct?

Unless I've managed to misinterpret many indicators in the field, that is my experience. One slight difference is the antenna may not rely so much on the cone for matching since it uses a gamma. If you omit the gamma it gets very hard to match the antenna by simply trying to find some shunt feed tap point on the radiator that provides a good match.


If so, it shouldn't be difficult to build a common radiator structure so we can just
vary the shape of the matching stub at the base between a single parallel wire
and a 4-wire cone to see how much difference it makes.  One approach might be
to build the antenna on an SO-238 socket with the center radiator attached to a
banana plug in the center and the cone build on the outer shell of a PL-259 that
screws onto the outside.

Or we could build a phasing stub and extension that works when attached to the
top of a J-pole, and then try it on the cone version and see if it acts the same
way.

If I made an open cylinder out of some tin cans, would you expect it to act more
like the conical stub or parallel wires?  That would be an easy test to run once the
basic structure was in place, as would the "open sleeve" version with 4 parallel
wires.

Maybe I can get some of the other hams at work interested in the problem and
we can run some tests in the lab using the network analyzer, but that probably
would need to be scaled for 440, as the 2m version is getting a bit tall and
the ceiling may detune it.


All of these seem like good things to test. If you have the time to try any of them I'd be very interested to hear your results.
Logged
WB6BYU
Member

Posts: 13172




Ignore
« Reply #73 on: February 21, 2013, 05:36:25 PM »




Let's see if I can untangle the quotes to get the important part...


Quote from: KE1IZ
...The phasing section was just a 1/4 wavelength (90 degree long) piece of wire wrapped around the insulator in a single turn with one end connected at the bottom of the insulator and the other end connected to feed the top radiator...


Which might be part of the problem.  You are wrapping ~19" of wire
around a 12" insulator and apparently expecting:
     1) it will give a phase shift relative to the length of the wire, and
     2) it won't radiate.
Neither of these is likely to be the case in the real world.

You can't just wind a length of wire onto a coil form and expect it
to replace the same length of radiator in an antenna.  As an example,
consider a loading coil for a mobile whip.  Let's say that, for a given
whip length, I need a 10uH coil for 20m.  If I wind a coil that is long
and skinny it takes more wire to get the same inductance than if I
make it short and fat.

The Hamwaves inductor calculator may come in handy to see why:
http://hamwaves.com/antennas/inductance.html

Say I make the coil 25mm diameter and 100mm long (about 1" and 4").
I need around 42 turns to get 10uH, with a total wire length of
3.3m (using 1mm wire).  Now if I made a long helical coil, say 10mm
x 300mm, the same conditions would require 175 turns and 5.5m of
wire for 10uH.  But by making it 100mm diameter and 25mm tall it
only takes 10 turns and 3.1m of wire.  All of these would provide
about the same loading effect (though the physical length of the
long skinny coil would affect the total antenna height.)

The point of this is that the length of the wire in a coil doesn't
have any bearing on the actual coil inductance or the phase
shift through the coil.

Where you do see a coil used for phase reversal in an antenna (such
as the traditional 900 MHz cell antenna, or some 2m/440 dual band
antennas) the coil is designed to be parallel resonant.  This
requires a specific set of coil dimensions so that the parasitic
capacitance across the coil resonates with the inductance.  In this
case it is really acting like a trap.  This will give a 180 degree phase
shift, but varying the wire length will NOT vary the phase shift
in any sort of linear manner.

I'll have to do some modeling when I get home to see if I can
understand exactly what is happening.  I'd suggest that you try
replacing the long insulator and 1/4 wave wire with an actual
shorted stub and an insulator of not more than about 1" long (to
match the stub spacing) and see if that works more as expected.

What you are finding about the gain increasing as you increase the
length of the insulator appears to be why we often use an extended
double Zepp rather than a standard one:  you are moving the two
radiating half wavelengths further apart.  With 12" spacing you are
very close to the dimensions for an EDZ, and with proper winding of
a stub or coil you might get there.

But what you have described is not going to give you the phase
shift you expect.
Logged
KE1IZ
Member

Posts: 30




Ignore
« Reply #74 on: February 21, 2013, 05:55:47 PM »




Let's see if I can untangle the quotes to get the important part...


Quote from: KE1IZ
...The phasing section was just a 1/4 wavelength (90 degree long) piece of wire wrapped around the insulator in a single turn with one end connected at the bottom of the insulator and the other end connected to feed the top radiator...


Which might be part of the problem.  You are wrapping ~19" of wire
around a 12" insulator and apparently expecting:
     1) it will give a phase shift relative to the length of the wire, and
     2) it won't radiate.
Neither of these is likely to be the case in the real world.

You can't just wind a length of wire onto a coil form and expect it
to replace the same length of radiator in an antenna.  As an example,
consider a loading coil for a mobile whip.  Let's say that, for a given
whip length, I need a 10uH coil for 20m.  If I wind a coil that is long
and skinny it takes more wire to get the same inductance than if I
make it short and fat.

The Hamwaves inductor calculator may come in handy to see why:
http://hamwaves.com/antennas/inductance.html

Say I make the coil 25mm diameter and 100mm long (about 1" and 4").
I need around 42 turns to get 10uH, with a total wire length of
3.3m (using 1mm wire).  Now if I made a long helical coil, say 10mm
x 300mm, the same conditions would require 175 turns and 5.5m of
wire for 10uH.  But by making it 100mm diameter and 25mm tall it
only takes 10 turns and 3.1m of wire.  All of these would provide
about the same loading effect (though the physical length of the
long skinny coil would affect the total antenna height.)

The point of this is that the length of the wire in a coil doesn't
have any bearing on the actual coil inductance or the phase
shift through the coil.

Where you do see a coil used for phase reversal in an antenna (such
as the traditional 900 MHz cell antenna, or some 2m/440 dual band
antennas) the coil is designed to be parallel resonant.  This
requires a specific set of coil dimensions so that the parasitic
capacitance across the coil resonates with the inductance.  In this
case it is really acting like a trap.  This will give a 180 degree phase
shift, but varying the wire length will NOT vary the phase shift
in any sort of linear manner.

I'll have to do some modeling when I get home to see if I can
understand exactly what is happening.  I'd suggest that you try
replacing the long insulator and 1/4 wave wire with an actual
shorted stub and an insulator of not more than about 1" long (to
match the stub spacing) and see if that works more as expected.

What you are finding about the gain increasing as you increase the
length of the insulator appears to be why we often use an extended
double Zepp rather than a standard one:  you are moving the two
radiating half wavelengths further apart.  With 12" spacing you are
very close to the dimensions for an EDZ, and with proper winding of
a stub or coil you might get there.

But what you have described is not going to give you the phase
shift you expect.


I think you're under the impression I have wound the phasing section right over the insulator. I tried to describe how this is a one turn "curved stub" with no part overlapping itself so as not to form inductance that would significantly alter the electrical length. The physical length of the curved stub tuned up to within an inch of the folded sections length.

Another way to approach this would be to peak the phasing section in EZNEC with whatever type of delay you wish to form. Whatever length EZNEC says will peak gain on the horizon, cut it in half for it to work in the field.
« Last Edit: February 21, 2013, 06:07:33 PM by KE1IZ » 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!