Antenna modelling help

Discussion in 'Antennas, Feedlines, Towers & Rotors' started by M7DJK, Jul 31, 2020.

ad: L-HROutlet
ad: l-rl
ad: Left-2
ad: FBNews-1
ad: Subscribe
ad: L-MFJ
ad: Left-3
  1. M7DJK

    M7DJK XML Subscriber QRZ Page

    Hi all,

    Attempting to educate myself on antenna modelling :)

    Have got a copy of 4NEC2 running, seems ok, a little complex but some good videos on youtube.

    I jumped straight in to the deep end and modelled my existing aerial, a G7FEK which is a nested Marconi style, nothing like biting off a big chunk eh :rolleyes:

    It looks good on the model view but i get and error that two wires are crossed or not joined which is a bit baffling as they are on the picture (below) wires 5 & 6
    [​IMG]
    Wire 5 is the shorting link between two wires of a ladder line vertical section, wire 6 is my feed point wire, both parts of the ladder are fed in parallel.

    Anyone have any views ??
     
  2. M7DJK

    M7DJK XML Subscriber QRZ Page

    Sorted :)

    Wires can only join at a segment junction, moving on......
     
  3. G8HUL

    G8HUL Ham Member QRZ Page

    Hi

    Wires can only touch at either wire ends or at a segment junction.
    If you turn on the segment display you will see the problem.

    It would be much easier to break wire 5 into 2 parts and have 3 wires join at the same spot, it is much easier than trying to align segments.

    73
    Jeff
     
    M7DJK likes this.
  4. K6CLS

    K6CLS Ham Member QRZ Page

    Thanks for posting the resolution. Good reference for future readers.
     
    M7DJK likes this.
  5. WA7ARK

    WA7ARK Ham Member QRZ Page

    This is a problematic configuration for NEC2 (doesn't matter if it is being called via EzNec or 4Nec2).

    1. A source must be placed in a wire with at least one equal length segment on either side of the seg where the source is placed. This usually means creating a short, 3 seg wire, and putting the source on the middle seg of three segs. The source acts as though it is in the geometric center of the middle wire seg. Another way is to place the source on the second seg in from the end of a longer wire with many segs. There is a min seg length restriction.

    2. Modeling two closely-spaced, parallel conductors (your ladder line) requires that the parallel wires be segmented such that the segs are the same lengths, and that the seg dots are aligned.

    3. Where you have a situation where current divides (as shown in the figure in post #1), that is tough to model in a way to get a close to unity Average Gain Test. L.B.Cebik came up with a way of solving this using a single source and a very short transmission line, which is shown in Dan's AC6LA's write up here: https://ac6la.com/aecollection4.html#15a Look for the two illustrations labeled "...leads to good AGT results" and "...leads to poor AGT results". Your posting unfortunately is done the "poor" way.

    4. Your model shows that you are effectively electrically connecting the bottom end your source to earth ground. That connection performs two separate functions. Electrically, it provides an "electrical ground" to complete the electrical circuit for the current path to flow in. From an "antenna" standpoint, the EM radiation from the vertical monopole interacts with the earth surrounding the base out to a distance of many wavelengths to form the far field radiation pattern of the antenna.

    EzNec has three different "ground" models, Perfect, Mininec, and HighAccuracy. I think that 4nec2 has these.

    Perfect Ground is a zero Ohms electrical connection to earth so it produces an unrealistic (too low) feed-point impedance for your modeled antenna.
    Perfect Ground is a lossless, perfectly reflecting, surface of infinite extent that produces a totally unrealistic far-field antenna pattern. Its only use is to compare to "textbook" illustrations.

    Mininec Ground is a zero Ohms electrical connection to earth so it produces an unrealistic (too low) feed-point impedance for your modeled antenna (same as Perfect Ground). This is what Mmana has... If modeling a vertical antenna, it is necessary to add an explict resistor (load) to account for ground loss.
    Mininec Ground is a first-cut, simplistic method at modeling losses in earth near an antenna and how that effects the radiation pattern. Unfortunately, it works poorly for any antenna with near-earth horizontal conductors (like radials). It works fairly well for horizontal antennas if they are at least ~0.2 wavelength above earth... You can specify user-defined conductivity and relative permittivity (dielectric constant).

    HighAccuracy Ground in NEC2 does not allow a direct, Ohmic connection to earth, which means you have to do something else... There is an extensive coverage of how to do this in the EzNec help file. The methods will apply to 4nec2.
    HighAccuracy Ground in NEC2 uses the same Sommerfield method as NEC4, so it does a good job of computing the radiation pattern and how much of the transmitter power is absorbed in the earth during the pattern formation.

    There is a learning curve with antenna modeling. Unfortunately, you dove right in and chose a difficult model to learn on...
     
    K0UO and M7DJK like this.
  6. M7DJK

    M7DJK XML Subscriber QRZ Page

    Thanks for the tips.
    Yes i know its a tough model, but its what i'm using so i thought it best to be able to correlate the real with the theoretical worlds :)

    In answer to WA7ARK, i will look at the feed point, thanks. The parallel section is segmentally aligned, looks ok. I will follow that link and look at the info.

    The original pic was only a beginning the aerial actually has two tuned radials, i have now lifted the whole above ground, added the radials, disconnected the whole from "Real Ground" and run auto-optimisation on the two main frequencies.

    I "think" i have some sensible data in the pictures below, i now need to look at the data and learn what it all means WRT real-world usage. I think the patterns look correct for what they are but obviously have to learn if they are saying good or bad yet.

    It appears that the vertical section does not radiate - it shows no current waveform???

    At least i have got further than i did with MMANA ;)


    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]
     
  7. WA7ARK

    WA7ARK Ham Member QRZ Page

    It looks like you are trying to model a dual-band, 80m/40m, back-to-back inverted-Ls where a piece of balanced feed-line is used as the vertical. I had previously created a parametric model with the following parameters:


    B: 0.3 Base Height (ft)

    P: 32.242 40m radial (ft)
    Q: 35.232 80m radial (ft)
    S: 4 vertical wire spacing (in)

    U: 1 Feed wire (ft)
    V: 25 Vertical Height (ft)
    X: 9.289 40m horizontal leg (ft)
    Y: 50.158 80m horizontal leg (ft)

    upload_2020-7-31_17-29-58.png

    upload_2020-7-31_17-38-50.png

    I used the optimizer in AutoEz to simultaneously make a low Swr50 at 3.65 and 7.15MHz. I allowed it to vary P, Q, X and Y. The optimizer found ratios between the horizontal wires and their respective radial which makes the feepoint on both bands near 50 Ohms resistive.

    Here is what I came up with using average dirt and uninsulated #14awg copper wire:

    upload_2020-7-31_17-45-27.png

    Notice that the optimizer found lengths that make for a low Swr50 at both frequencies.

    The "Average Gain Test" column shows that my model is pretty good. Getting this number as close to unity is a necessary (but not sufficient condition) of validating the model (see Eznec's Help file for an explanation).

    The maximum gains are not great (-1.5 and -1dBi at 40m and 80m, respectively) . This due to effectively having a total of only two radials. This is also shown in the "Average Gain Factor" column. If you feed 100W to this antenna, only 19.4W gets radiated on 80m, and 23W on 40m. Most of the rest is lost in the earth as the pattern is formed and a tiny bit is lost heating the wires.

    I would encourage you to find the "Average Gain Test" number for your model. If it is not as close to unity as mine, then study what I did near the feedpoint to get mine as close to unity as I show... The trick is the segmentation of the vertical wires, the segmentation around the feedpoint, and using the lossless transmission line to feed the second vertical wire...

    It is interesting to plot the current distribution on the antenna wires at the two design frequencies. Ideally, on 40m all of the current should be in the 40m side, with no current in the 80m side, and vice-versa. I am driving the antenna with 100W at each freq. This shows that doesn't happen as expected, but you can see the trend:
    upload_2020-7-31_18-12-53.png

    I am a bit curious where you came up with 3mS for ground conductivity and a dielectric constant of 4?
     
    Last edited: Aug 1, 2020
    M7DJK likes this.
  8. AC6LA

    AC6LA Ham Member QRZ Page

    M7DJK, for some help on that see this old QRZ post and the one after that.

    And an unrelated suggestion: When requesting modeling help it's always nice to include your model so others don't have to guess what you did. With 4nec2 and MMANA you can just copy/paste the ASCII text of the *.nec or *.maa file directly into a post. With EZNEC you can either a) put the *.ez file inside a *.zip file and attach the zip file using the "Upload a File" button, or b) use Mike's trick of appending .txt to your model name and attach the txt file. That is, if your EZNEC model file is MyModel.ez rename it to MyModel.ez.txt and upload that.

    Looks like that is one of the "generic" ground type choices in 4nec2; Poor, Moderate, Average, Good.

    moderate.png

    Dan, AC6LA
     
    M7DJK likes this.
  9. M7DJK

    M7DJK XML Subscriber QRZ Page

    Yes its a G7FEK nested Marconi 40/80 multiband thingy :)

    I did try the transmission line bit but have not yet found out enough as to how its setup/implemented to make it work, will keep playing with that.

    My efficiency is poor so far, something like 10-15% so i guess my 10W on 40m is only 1.5W radiated, maybe, plus looking at my orientation and max signal it seems i am firing most of my 1.5W at Norway/Sweden :)

    I was wrong in my earlier post, the current is there on the vertical but it was at 90deg to the horizontals so my picture view did not show it :) I am guessing this is why my max output is rotated to one side?

    [​IMG]
    Its certainly fascinating but i need to figure this zero length transmission line bit some more.

    Model is here...
     

    Attached Files:

  10. W9XMT

    W9XMT Ham Member QRZ Page

    Sorry, but that belief is invalid (see the graphic below).

    In the case of the antenna system configuration in the Dave G7 2 Optimised.nec file posted by M7DJK here in Reply 9, the 40m radiation pattern is fully formed at a horizontal distance of about 10.2 meters (less than
    λ/4 from the radiator).

    [​IMG]
     

Share This Page