Radial length and angle radiation patterns... Long...

[The DB]

I was modeling different sized and angles for radials for a 5/8 wavelength antenna and opted to put this information here to make it easier for people who are interested to find...

Some info on the models first. The base of the 5/8 wavelength vertical section is 1 wavelength above average ground in all cases. Obviously the radials do go below this point as in most cases they are angled down. Changing the height and the ground quality, as well as other obstacle that may exist around the antenna can change the results you will see. Also, the angle of the radials is always in relationship to vertical.

The plots are always in the same order. First is the 1/4 wavelength radials, then the 1/2 wavelength radials, and the third and last are 5/8 wavelength radials.

Starting with the 90* or horizontal radial set.

The radiation pattern varies some here, as the 1/4 and 1/2 wavelength radials both have good low angle gain, with the 1/2 wavelength radials the best gain at 0.8 DBi gain over the 1/4 wavelength radial set. The 5/8 wavelength radial set is causing a high major lobe at 45* above the horizon. This high angle lobe gains dominance when the radials are not much longer than 1/2 wavelength in length. It is like flipping a switch, not a gradual change. If you are designing an antenna with such a horizontal radial length you may consider using slightly shorter radials than 1/2 wavelength to avoid the quick changeover to this high angle lobe.

Next the 75* angle set of radial set.

With the 1/4 wavelength antenna angling the radials down even this small amount pushed the strongest signal to a higher angle lobe. As we continue to angle the radials down more and more of the signal is put in this higher angle lobe, part of it being robbed from the lower angle lobe.

With the 1/2 wavelength radials gain appears to have shot way up, but so has the radiation pattern. I was surprised at how quick this change over happened.

With the 5/8 wavelength radials gain seems to have dropped, but so has the dominance in the higher angle lobe.

The 60* angle radial set.

Both the 1/2 and 5/8 wavelength radial sets have moved their strongest radiation to a lower lobe. In neither case is it the lowest lobe, but both of the antennas would be useable.

The 45* angle radial set.

At this point the 1/2 wavelength radial set is lowering its main lobe some and increasing that lobes gain at the expense of all other lobes, including the low angle lobe.

The 5/8 wavelength radial set for the first time has the low angle lobe as the dominant lobe. As of yet this is the strongest low angle lobe seen. There are also no significant high angle lobes, allowing for all of the significant output to be within 30* of the horizon.

The 30* angle radial set.

Here we see more of the same as above.

The 15* angle radial set.

For the second time we see the half wavelength radials with a dominant low angle lobe, however, the low angle lobe with the 5/8 wavelength radial set has even more low angle gain with a pretty significant 6 DBi gain not quite doubling the gain from the typical 4 1/4 wavelength horizontal radial setup (3.17 DBi)

Just for kicks, a single radial of said lengths at 0* off of vertical.

In all cases we see gain drop. There appears to be a point in this range that is the limit of the beneficial effects from the 1/2 and 5/8 wavelength radial lengths. From past experience I think this point is closer to the 15* point than the 0* point.

Notes:
1) I did not model a feedline or mast in any of these cases. Especially with the low angle of radial sets you would have to assume there will be some coupling to a mast and feedline as they are not far away.
2) The naturally higher impedance of the 1/2 and 5/8 wavelength radials will allow more RF to flow on the coax outer shielding and mast, compounding the problems from number 1) above.

Because of 1) and 2) above if experimenting with any of these setups I would recommend a non-conductive mast and a good RF choke for the coax.

I'm curious for your thoughts, or did I miss anything?


The DB

 

[Unit 194]

Thanks for sharing.

 

[wa1knx]

radials

Db,
Very interesting work. I have always been under the impression that
the currents in the radials of 5/8 antennas are out of phase with the upper
1/2 wave of the 5/8 radiator. so drooping radials makes it behave more like a
long wire, pushing the radiation to the higher lobes. (unlike drooping
a 1/4 wave radiators radials) the surprise result you have unearthed
is the 45 degree drooping 5/8 radials! Marconi had some models around
of some of this. (speaking of which I have not seen him in here of late.?)
thanks for the models! and happy new year.

 

[The DB]

The currents in 1/4 wavelength radials are out of phase with the upper part of the 5/8 wavelength vertical. Therefore it makes sense that as you angle the radials down there is more and more vertically polarized out of phase radiation, thus shooting the RF skyward.

With the 5/8 wavelength radials the phase at the tips of the radials is always in phase with the phase at the tip of the vertical. There is also always a phase inversion that you would expect with a 5/8 wavelength length as you approach the feedpoint.

With the 1/2 wavelength radial set the currents were also always out of phase with the upper part of the except for the 15* and 0* sets.

I'm guessing that at those angles the radial sets appear to be somewhat longer than 1/2 electrical wavelength in the models, thus causing a phase switch similar to that of the 5/8 wavelength radials. To test this I shortened the radial length in one of the models... The before and after are below.

A small shortening now shows the radial as being out of phase with the tip. The thought seems to be confirmed. So how would this affect the radiation pattern of the antenna? Again a before and after set.

What a difference a small change can make. We are talking a change of less than 0.05 of a wavelength difference in overall length of the radial, and the radiation pattern changed dramatically. When dealing with half wavelength radials fairly big changes apparently happen quickly with a minor change in length as you pass over that electrical half wavelength length boundary of the radial. Not only has overall gain gone down but the dominant lobe is now at a higher angle...


The DB

 

[hotrod]

thanks very nice write up. now whats all this mean in simple terms,lol

 

[HomerBB]

thanks very nice write up. now whats all this mean in simple terms,lol

 

[The DB]

thanks very nice write up. now whats all this mean in simple terms,lol

It was an experiment to see what would happen under varying conditions. It shows that the standard commercial layout isn't necessarily the best layout for gain, and what you might do to achieve such gain. But some points to take note of...

1) If your sticking with standard near 1/4 wavelength radials, horizontal is better than angling them down, even slightly. As you angle said radials down the more gain will shift to a higher lobe at the expense of a lower angled lobe.

2) When dealing with horizontal radials the standard near 1/4 wavelength radials don't necessarily provide the most gain. That being said, making said radials longer will increase gain until they reach an electrical half wavelength point, at which point the major RF lobe switches to a much higher lobe. This happens drastically and quickly once you pass a certain length so being slightly shorter isn't necessarily a bad thing.

3) If your antenna is long enough to have full length 5/8 wavelength radials at a 45* or lower angle there is low angle gain to be had. This gain peaks when the radials are 15* off of vertical (almost straight down), and at this point can almost double the gain of the antenna with standard 1/4 wavelength horizontal radials.


The DB

 

[Shockwave]

Now I wonder if there is not some easy way to get this almost double gain? What happens if we take the antenna and put it on a 5/8 wavelength mast and choke the coax off at the bottom of the mast?

 

[The DB]

Now I wonder if there is not some easy way to get this almost double gain? What happens if we take the antenna and put it on a 5/8 wavelength mast and choke the coax off at the bottom of the mast?

I added a wire length that went from the base of the 5/8 vertical element through the middle of the radials to the ground, and had 4nec2 connect that wire to the ground. This was to simulate a mast or coax feed. I took the following from that model, first is the plot, then the current distribution, and the last one is the phase.

NOTE: 4nec2 did throw up an error when running the far field data telling me that the angles between the central vertical element and the radials was to sharp, this is the data after it ran it anyway. I'm not sure how much of an effect this known issue with the NEC2 engine will have.

The magnitude of the out of phase portion of the wire below the radials seems to be low, and seems to have a minimal effect on the pattern as a whole. I think the improved gain shown on the plot is from 4nec2 seeing a higher radiation efficiency (about 10% higher) for this model over the previous one with no mast simulation. I should play with this model at different heights to see how much all of this changes. That will be a task for tomorrow...

I do have a free copy of the an-soft 100 modeling software. Perhaps I should put the antenna data in that program and see if it has the same issues. They do advertise that several known issues with nec2 has been corrected in their software. If the issue has been fixed it could help confirm/counter the 4nec2 data shown here. That will be another task for tomorrow...


The DB

 

[Shockwave]

DB, I was trying to come up with a way to get the same gain without having to use a set of 5/8 wavelength radials and should have been more specific in the earlier post. Since you see highest gains when the 5/8 wavelength radials are nearly vertical, what happens if you remove them and replace them with a single mast that is exactly 5/8 wavelength?

If this works, it's easy to choke the coax off right at the bottom of the mast so it doesn't upset the match. If it produces a stronger high angle lobe without any radials decoupling the feedpoint, adding a few short radials might fix that. Something like four horizontal radials between 2 and 5 feet.

Marconi had some models around
of some of this. (speaking of which I have not seen him in here of late.?)
thanks for the models! and happy new year.

WA1KNX, I've been wondering the same thing about Marconi and am hoping he's just spending time with family around the holidays. If anyone has heard from him, please keep us posted.

 

[ghz24]

4nec2 did throw up an error when running the far field data telling me that the angles between the central vertical element and the radials was to sharp, this is the data after it ran it anyway

To avoid the sharp angle problem make a short wire stub horizontal to the mast and attach the angleed radials to the end of the stub. Also be sure toi use an EK (extended kernel) "card"
The AGT should tell you how compromized the model is or isn;t.

 

[wa1knx]

edz

I see where your heading shockwave. 5/8 under the 5/8 would be a
extended double zepp. choked off, it could produce a few db maybe 3
over a half wave. perhaps that is how the 15 degree radial model performs well. Though most readings I've seen says its a difficult task to actually get
3db vertically this way. Marconi, hope hes enjoying the holidays to.

 

[Marconi]

I'm still here. I'm just trying to better understand how 4nec2 works. My old models of the Imax with 5/8 wave radials showed similar results as DB I think.

There is a local ham guy here in the Houston area that sometime back published his Eznec Demo version of an Imax model and his real world experiences using 5/8 wave slanted down radials among other configurations. He claimed his real world Imax showed maximum gain using 5/8 wave radials as guy wires and the gain was significant.

I posted his work and discussed this idea before here on WWDX. I saw similar results in my models too.

I never tried the idea, and I can't recall if Homer ever tried such long radials or not. You would think if there was a chance that the Imax gain could be doubled this way...then folks would be trying the idea.

In my modeling I tend to find as models get worse and exceed the capabilities and limitations...the gain tends to increase, and within reason as the model gets to be a better model I see the gain decrease.

DB, GHZ24 is right, but try adjusting the segment numbers and tubing diameters closer to the same values and maybe you can fix the angle error problem without building a hub to connect the radials to the radiator.

Check your Average Gain results too.

What are the dimensions and segments for the radiator and radials for this 5/8 wave? I could fix my model and maybe it will be closer to your results.

 

[The DB]

DB, I was trying to come up with a way to get the same gain without having to use a set of 5/8 wavelength radials and should have been more specific in the earlier post. Since you see highest gains when the 5/8 wavelength radials are nearly vertical, what happens if you remove them and replace them with a single mast that is exactly 5/8 wavelength?

I modeled that in the ton of models shown initially. Here it is again:

The gain drops from 6 dBi to 4.77 dBi. The bottom tip is only 3/8 of a wavelength off of ground and I did have the thought that that might skew things a bit. When I get a chance I'll raise this and the 15* radial set model higher up above ground and see what happens.

If this works, it's easy to choke the coax off right at the bottom of the mast so it doesn't upset the match. If it produces a stronger high angle lobe without any radials decoupling the feedpoint, adding a few short radials might fix that. Something like four horizontal radials between 2 and 5 feet.

I'm curious as to where you are looking to put said 4 or 5 radials?


The DB

 

[The DB]

To avoid the sharp angle problem make a short wire stub horizontal to the mast and attach the angleed radials to the end of the stub. Also be sure toi use an EK (extended kernel) "card"
The AGT should tell you how compromized the model is or isn;t.

I should have thought of that (the short horizontal segments before the radials droop), I did the same thing when talking to some recent person about some vector model stuffs, although that may have been because the vectors actually come that way...

When I run the AGT on said model it matched up exactly to putting in a perfect ground plane underneath. Both of those showed 10 plus dBi gain (the same figure, I just don't remember specifically what, 10 point something). I also put that model in freespace to see what would happen as well, that time it was 5 point something dBi gain...


The DB