Pls review my model

Discussion in 'Antennas, Feedlines, Towers & Rotors' started by SV1IYF, May 19, 2017.

ad: L-HROutlet
ad: l-rl
ad: L-rfparts
ad: l-gcopper
ad: Subscribe
ad: K3QNTad-1
  1. SV1IYF

    SV1IYF Ham Member QRZ Page

    The attached MMANA file contains the model of a 4 element 20, 17, 15m Yagi-Uda.
    Three elements comprise an open sleeve bundle of driven elements for the 3 bands.
    The fourth element doubles as a (compromise) passive director for 17m and a reflector for 15m.
    A bend-ends dipole, carrying the common feed-point, is provided for 20m.
    At 20 meters AGL my optimization gives the following results:
    Frequency MHz / SWR / Gain (dB) / F/B (dB) / Elevation (deg)
    14.077 / 1.37:1 / 7.72 / (-)1.13 / 15
    18.103 / 1.14:1 / 11.75 / 11.4 / 12
    21.077 / 1.09:1/ 10.85 / (-)10.9 / 9.9
    (-) sign means 180 deg rotation as in the 17 and 15 m that "fire" towards opposite directions.

    From a mechanical point of view, the tube diameters may need some increase (used what I have at hand) and the close spacing of the open sleeve trio looks quite sensitive.
    From an electrical point of view, I am aware of the relevant article by Sevick (SK) regarding the options in arranging elements but went this way in order to keep the boom the shortest possible.
    I am only working JT modes, so bandwidth is irrelevant to me.

    Finally, I would welcome your comments/opinions and mostly a model verification by experienced modellers.

    Attached Files:

  2. AC6LA

    AC6LA Ham Member QRZ Page

    Hello Nicos,

    Here are the results using EZNEC Pro/4 with the NEC-4D engine. Variable "E" is used to set the elevation angle of the azimuth slice for each test case.


    The Average Gain Test values (column U) are not very good for the 17m and 15m bands. That may be due to the close spaced wires. (Ideally the AGT |dB| values should be below 0.2 dB.)

    Results are also very sensitive to the segmentation. I experimented with different segmentation densities from 10 segs/wl up to 50 segs/wl. Best AGT seemed to be when using 20 segs/wl, dynamically reset at each calculated frequency. I'm afraid I can't remember what the MMANA "-1" (automatic segmentation) setting translates to in actual segments per wavelength.

    For anyone else who would like to experiment with this model, the free demo version of AutoEZ (link) can be used to create .ez files from the original .maa model. Open the .maa file with AutoEZ then save in .ez format at 3 different frequencies to get 3 different segmentation densities. Note that when using a NEC-2 engine the results won't be very accurate since the "Stepped Diameter Correction" cannot be used with all wires of this model, as shown in this EZNEC screen grab.


    Dan, AC6LA
  3. K4SAV

    K4SAV Ham Member QRZ Page

    I got ready to post this and see that Dan has already jumped on it. I will post it anyway. My data was manually produced using EZNEC and Dan used AutoEZ in conjunction with NEC4.

    I can't help you much with a MMANA model since I don't use that, but I did convert it to an EZNEC model and made some plots. I should say, AutoEZ converted it for me.

    I noticed the program error was higher than I liked so I increased the number of segments on some of the wires. The error originally varied between 0.9 and 1.8 dB depending on the band. After making the model changes the error dropped to 0.55 dB (on all bands), which is still higher than I like but it's as low as I could get it.

    There was also a couple of wires where their lengths were shorter than their diameters. Most programs don't like that. I deleted the short wire and added their length to the adjacent wire. That unbalanced the segment length a little between adjacent parallel wires but it did reduce the error.

    My resonant frequency plots did not agree with yours. The changes I made to the model didn't move the resonant frequencies much either. Here is what I got using EZNEC.

    Freq ___ SWR
    13.6 ___ 1.03
    17.6 ___ 1.34 (lowest SWR)
    17.4 ___ 1.45 (resonant point)
    20.8 ___ 1.22

    I don't know how much of that is due to using different programs. It is a significant difference from what you got with MMANA. Don't assume my answers are any more accurate than yours. I think the MININEC based MMANA does handle varying wire diameters differently from NEC2. The variation in wire diameters in this model is fairly small, but the separation between the elements and element diameter is very critical relative to resonant frequency.

    Jerry, K4SAV
    Last edited: May 19, 2017
  4. SV1IYF

    SV1IYF Ham Member QRZ Page

    Thank you very much Dan and Jerry,

    I started this in 4NEC2, landing constantly into the red regarding AGT.
    Tried manual "coherent" segmentation of varying lengths, thing that sometimes improved -others worsened- things, but none of them produced any descent AGT value.
    The problem lies with the open sleeve coupled triplet.
    This is why I switched to MMANA, but was cautious regarding the outcome, mainly due to the high volatility of the latter as related to dipole spacing.
    That's why I called for specialists' help.
    It seems now that the only way to know better is to try the real thing.
    Thanks again.

    73 Nikos SV1IYF
  5. AC6LA

    AC6LA Ham Member QRZ Page

    Actually I think MMANA (or any other MiniNEC-based program) is the better choice for this antenna. MiniNEC does not have the problem with stepped diameters that NEC-2 and (to a much smaller extent) NEC-4 have. If you know of anyone who still has the old "Antenna Model" program by Teri Software you might ask them to run your model. "Antenna Model" (what a strange name for a program!) uses a MiniNEC engine and can read .ez format files. Another choice would be the old "AO" program by Brian Beezley, K6STI. That one would require manually rebuilding the model file.

    As an alternative, if you temporarily "straighten out" the two ends of the 20m dipole then EZNEC and 4nec2 could apply the stepped diameter correction to the wires. That could give you a check on other results.

    And one final point: Please accept my apology for misspelling your name in my earlier post! :)

    Dan, AC6LA
  6. SV1IYF

    SV1IYF Ham Member QRZ Page

    Thank you Dan for the additional info.
    No need to apologize regarding spelling, as I was using this form, until about mid 80s, when official transliteration rules for names and geographical places came into effect.
    The up to then Nicos Michaelides became today's Nikos Michailidis.
    Took me some time to adjust though...

    73 Nikos SV1IYF

Share This Page

ad: vanity