Group Actions
Engagement
  • Author Author: kmikemoo
  • Date Created: Date Created
  • Views 8406 views
  • Likes 10 likes
  • Comments 18 comments
Related
Recommended

Antenna Radiation - Phase 3

kmikemoo

Phase 3 of testing antenna radiation patterns.  The intro to this journey is Antenna Radiation.

Phase 3 will be to test a 2-element cardioid pattern.  This is formed when two 1/4 wave antennas are placed 1/4 wavelength apart and one is fed 90 degrees out of phase. I want to verify the single null to the rear.

 

The Amateur Extra amateur radio exam pool contains a question about this antenna pair and its unique radiation pattern - so, of course, I have to see if it's for real.  If it's what it's supposed to be, why isn't it common in Radio Direction Finding?

image

Based on the description in the test question, I built a two element, 1/4 wave dipole pair.  Based upon the testing done in Antenna Radiation - Phase 1, I do know that the antenna feed line will influence the radiation pattern.

I was able to use the automated transmitter (fox) and record the readings with my cellphone.  Three 30 second segments for each of the eight directions (0, 45, 90, 135, 180, 225, 270, 315) and an additional 30 seconds back at 0/360 - just to be sure.

I did not get the pattern that I expected.  The graph is a top down view.

imageimage

The in-phase feed point is at the center, both of the dipole mounted in line with the vertical support and of the graph.  The element to the right in the picture is 1/4 wavelength (+/-) away, resulting in a 90 degree phase shift of the transmitted signal at (again) 1/4 wavelength spacing.  This is what creates the large lobe to the right on the graph as 0 / 360 is the direction of the second element.

This is an awesome directional pattern and the nulls were clean (zero) and very distinct.  Too bad this wasn't the pattern I was looking for.  So... where did I go wrong?

 

DI-pole.  Two.  While the elements are a quarter wavelength long, there are two.  This makes my dipole a half wavelength antenna.  My error was assuming that the lower elements would act as a ground plane - which they did not.  I will have to add more counterpoise elements to try to create a ground plane.  Maybe I can put something together with whip antennas...

I'm starting to see why this antenna array might not be the system of choice for radio direction finding.

 

After thinking about it, I decided to turn the dipole pair into two 1/4 wavelength ground plane antennas.  Surely, that will give me the pattern that the Amateur Radio exam question references.

image

Yeah... no.  Wrong again.  I got the same radiation pattern with the ground plane antennas as I did with the dipoles.  The magnitude of the readings did not change if the lower dipole elements were connected to the ground plane elements or not.

 

Given that this pattern is accepted as a universal constant in antenna propagation, I must be doing something wrong.  It must be that the phase shift is more than feeding the radiator element 90 degrees later.  If the arrangement I have is feeding the radiating elements in phase, I should have symmetrical lobes - at least according to my antenna books and all of the web sites that I've visited in the last few weeks.  I do not have symmetrical lobes.  Time to restore the antenna back to the twin dipole configuration and see about creating a 90 degree shift in the feed lines to the forward radiating element.

 

After some Internet research, it doesn't look like I'll be able to create that 90 degree phase shift without spending a bunch of money - and nothing would arrive in time anyway.  Argh!

 

So I shifted to "What CAN I do?"  Well... I can try simple.  A capacitor.  I only have one that can handle the voltages.  It does nothing.  Not surprised.

 

What if I shift the feed line of the second radiating element?  Sure, if I connect the top radiating element at the first feed point to the lower radiating element on the second feed point it is a 180 degree shift, but I should see SOME change in radiation pattern.  The pattern is a decent approximation.  I did three laps around the antenna confirming the peaks and valleys and rough readings.  At least the pattern changed.

image

Again, the main vertical support is also the center of the graph.  The second dipole is to the right along the zero axis.  The measurements are made at two meters from the vertical support.  The measured values are doubled to illustrate the pattern.  At least I know that the radiation pattern can be changed.

 

One of my Yagi antennas has a 1/2 wavelength balun.  What if I try something like that?  I returned the dual dipoles to their original configuration and connected a 1/4 wavelength piece of coax to the second radiating dipole such the the center conductor (connected to both upper radiating elements) terminated on the feed point to the lower element of the second dipole.  The coax shield went from the lower element on the second dipole to the upper element.  Think: Feed point, feed point, shorting loop.

image

I would say that the 1/4 wavelength "balun" effectively eliminated the second dipole because the resulting pattern looks a lot like the original single dipole from Phase 1.  Again, the measured values were doubled for illustrative purposes.  Neither variant was pegging the meter.

 

Unless I stumble across a way to give me a 90 degree phase shift in the next week, I will not be able to see the cardioid pattern in real life.  This exercise does demonstrate why this antenna setup is not used more often in Radio Direction Finding.  Time Difference Of Arrival is a simple enough circuit and uses the venerable 555 timer.  I was not able to find something quite so elegant for the 90 degree phase shift.  I'm not done looking... just bitter. Ha! Ha!

 

This was an awesome experience and I thank Element14 for selecting RF.  I would not have been motivated enough without the challenge.  I appreciate the opportunity to actually see phenomenon that I've only read about.  I will definitely play with this more once the weather improves.  I originally planned to scrap the dual dipoles after this, but after seeing how clean their radiating pattern is... I'm going to find a use for them.

73, Mike, WI9MMM

 

UPDATE:  8 JAN 2020

Huge thanks to Jon Clift (  ) and Shabaz (  ) for their wisdom.  I ran some simulations using EZNEC, the free version of an antenna modelling program.  Given some allowance for the feed line within the radiation pattern, I would say that this is similar to what I measured.

   imageimage

The speed of the signal through the coax - despite the short physical length - caused my second dipole to radiate at about 135 degrees out of phase.  I am using RG8x coax with a velocity factor of about 0.8.  If I go for a 270 degree shift and account for the velocity factor, I should still get the cardioid pattern - only on the opposite side of the pair.  I am curious to see how much the radiation pattern is affected by the extra feed line within the field.

 

Running a few more simulations, I discovered that a 270 degree shift would result in high SWR but a 450 degree shift (1-1/4 wavelengths) should be fine.  BUT... the velocity factor range of my RG8x is 0.79 to 0.84 - according to the internet.  Being off by a few inches can skew the pattern.  Window line, similar to the antenna wire on our old televisions, has a velocity factor of 0.92 to 0.95 and just wire is supposed to have a velocity factor of 0.95.  0.05 difference in the RG8x.  0.05 difference of wire to open air.  I'll try just wire.

image

Yes, that is just two wires connecting the dipoles.  It would be the same thing as window line - only with a bit more spacing.  I did need to change out the radiating elements for longer elements to get the antenna's SWR to be its lowest in the 144MHz to 148MHz range.  I fully expected the unshielded connector to severely mess with the radiation pattern would it be that much more than the coax?  We'll find out.

image

WOW!  Is that an ugly pattern, or what?!  BUT... the null is definitely there and it is where it should be.  I have not figured out how to model how the feed line affects the pattern but from previous experiments, I know that it decreases the radiated signal strength around it.  On this experiment, the feed line only extended away from the feed point.  I needed to use it to "anchor" the antenna to keep it from thrashing about in the wind.  I also added some simple reference points.  The resulting null is fairly shallow due to the feed point phase difference being past 90 degrees (I estimate around 105 degrees) and, more significantly, RF being radiated from the feed line and the 1/4 wave connector.  I'm confident that I could get a deeper null if I shifted to coax, but not sufficient enough a null to be discernible for Radio Direction Finding.

 

Mission complete.  Return to base.  Clean up shop. image

  • HUGE THANKS to and !!!  I think I can make the cardioid happen.  I'm also thankful that it will be much warmer tomorrow.  It started out at 5 F (-15 C) this morning.

     

      I took your lead and used EZNEC.  I was going to see if an oscilloscope would show me my feed point phase angle difference, but why?  The objective is the pattern, not the measurement method (at this point).

     

      You are right.  The issue is the coax and the velocity factor.

    To get a good null, everything will need to be right. The distance betweeen the dipoles, the 90 degree shift by delay in the coax for the second dipole [which will be the hardest to get right], and the alignment (the dipoles need to be accurately in the same plane).

    "Everything will need to be right" really kicked the wheels in motion.  In RF, sometimes the "bandwidth" of the solution is very narrow.

     

    I will try a 270 degree shift accounting for velocity factor.  We shall see...

  • in reply to shabaz

      The two plots with 3d are AWESOME.  I did have plans to try 1/2 wavelength spacing.  It will be interesting to see what I measure.

  • in reply to kmikemoo

    I'm not an RF engineer, so if your Radio Amateur friends say one thing and I say another, go with what they say (they're the ones with real experience). I can understand some of this, partly from general principles and theory, but also from working with EMC (where you realise, after you've been doing it for a while, that you're reversing into rf engineering and gradually understanding what were once regarded as the 'black arts').

     

    An important thing to remember with high frequencies is that currents flow close to the surface [skin effect]. So the return current, for the signal that goes down the centre conductor of the coax, flows near the surface of the inside of the shield. It's on the inside because it wants to be as close as possible to the signal it's associated with. Although it seems counter-intuitive to what we're used to at DC, where a current can only flow in one direction down a conductor, in this case a second rf current can flow on the outer surface of the screen without interacting with the one inside. It can even flow in the opposite direction to the one inside. Essentially, the outside can then behave just as a passive director or reflector would, receiving a signal and then re-radiating it. [Real life is, of course, more messy and there will often be some small currents on the outside of the coax. In situations where people are doing critical measurements, like EMC testing, you'll often see them clip ferrites to both ends of the antenna connecting cable to absorb any outer-surface currents.]

     

    What your friends are saying about the diagonal is that if you have the choice between parallel to the dipole or on the slant, go for the slant. That's sensible if your feed has to be close to the dipole as it disrupts any director/reflector effect as much as possible. But, that said, it still should be better to get it as far back as possible [because the further away, the less the signal received and reradiated, and the less the effect it then has when it gets back to the dipole].

     

    This is how I'd explain the cardioid pattern.

     

    image

     

    When the wave radiated by dipole A arrives at B (a quarterwave away), it's 90 degrees on. If B is fed with a signal that's also 90 degrees on, then the two add in the forward direction (constructive interference). However, the signal radiated by B in the reverse direction, when it gets to A, will be 180 degrees on from the signal emitted by A in the reverse direction and the two will subtract (destructive interference, if you want to think of it in wave terms). So that's why the strong null in the reverse direction. But it only applies to the reverse direction along the axis joining the dipoles, as soon as you move away from that axis the cancellation starts to fall and the signals start to progressively add more and more as you come round.

     

    To get a good null, everything will need to be right. The distance betweeen the dipoles, the 90 degree shift by delay in the coax for the second dipole [which will be the hardest to get right], and the alignment (the dipoles need to be accurately in the same plane).

  • in reply to kmikemoo

    Hi Mike,

     

    I think the model shows the radiation pattern centered between the two dipoles. The diagram orientation is with the dipoles sticking out of the page so-to-speak, i.e. the drawn radiation pattern is in directions perpendicular to the dipoles, i.e. what would be measured walking around the dipoles if they are vertical.

    There is a sharp, narrow null skyward, and a sharp narrow null toward the ground, but I didn't capture the plot for that.

    If the dipoles are spaced half-wavelength away instead of quarter-wavelength, then the result is different:

    image

    And this is the 3D plot:

    image

    It's quite neat.. blasting out a lot of radiation in two quite focussed directions : )

    I got curious and tried six dipoles (spaced half wavelength apart):

    image

    image

    That's even more highly focussed : ) The 3D plot looks like some bizarre dystopian mind control ray..

  • in reply to jc2048

    Hi Jon,

     

    It only allows variation within certain ranges, i.e. it generates an error message if I try to move the feed too much from the location it picked. I'm fairly sure it must be formula-based, and using the rules to block massive errors if the formula is attempted to be used outside bounds. Also, it's way too speedy hehe (near-instantaneous).

    I'm guessing (not sure) the commercial use-case is to approximately (yet rapidly) design it with the formulas, and then bring out the expensive tools.

    I think we by chance ended up trying the same bits of software, because what you describe sounds similar to what I experienced too.

    Also, I tried free NEC a year or two ago, for a few hours. I don't recall much but it had a weird entry system for defining individual elements, and I think it kind-of worked, but I didn't get very far. It is a shame in this area the great software is locked down and out of reach of individuals : ( It was clear that the free version of AN-SOF was using the same underlying engine, because its system had similarities, but with a front-end.