Eznec Question

Discussion in 'Antennas, Feedlines, Towers & Rotors' started by K5WZI, Jul 23, 2019.

ad: L-HROutlet
ad: l-rl
ad: Left-3
ad: Left-2
ad: QSOToday-1
ad: Subscribe
ad: abrind-2
ad: L-MFJ
  1. UA3TW

    UA3TW Ham Member QRZ Page

    Many thanks, Dan. I think it is shere psychology - one prints something into a box and one expects that the printed content pops up in the box immediately. If not - one gets exited and reloads the model. It is just what happened to me.
    And this has led me to the second blunder, which I afore described. I put it in other words now. If we have a number of variables but by some mistake excluded one of the variables. For example we had a variable 470 pF as =A
    Then we have gone to the list of loads and replaced 470 by 400. It means that the variable is no longer active isn't it. Of course it was a mistake - you should change it in the list of variables, not loads. But you did not noticed it.
    Then you returned to optimization and started it having all the variables ON. And the variable =A (that stand for the cap 470pF) will look as if it takes active part in tuning, it changes somehow in the table of active variables.
    In my case it changed down to 150 pF, whereas it really remained 400 pF in the model. I was unaware that it does not change and it is strange that the program does not give any warning in such a case, isn't it?
  2. UA3TW

    UA3TW Ham Member QRZ Page

    I was asked to develop a 4 el Yagi for 40m and I've done that in MMANA. When I opened this MMANA file in AutoEZ I got a different result. Particularly SWR at 7200 in MMANA iz close to unity whereas in AutoEZ it is about 1.8. For checking I developed 2 more files - one is out of wire elements the other is out of constant diameter elements. They coinside with AutoEZ 100%. Therefor I think that the reason of the difference is somewhere in stepped diameter correction. But I have no idea what stepped dia correction is more correct, MMANA or EZNEC. Would appreciate any opinions. The three models are attached, stepped dia, constant dia and wire elements.

    Attached Files:

  3. AC6LA

    AC6LA Ham Member QRZ Page

    The elements in Nick's 40m 4L Yagi have a relatively large number of diameter steps with each diameter section being relatively short. For example, here are the taper sections for the driven element (10 sections per half-length):


    Compare that to a 40m 3L Yagi from The ARRL Antenna Book (6 sections per half-length, including an equivalent diameter "boom clamp"):


    It seems that the relatively short sections are causing some problems with both NEC-2 (using the Leeson stepped diameter correction algorithm) and NEC-4 (no Leeson correction, both per the EZNEC Help recommendation). Here's a comparison of the Free Space SWR(50) curves for five different scenarios:

    1) (Red) MMANA (modified MININEC engine) with "tapering segmentation" per Nick's original model. Shorter segment lengths close to each change in diameter and longer segment lengths in the middle of each section. Total pulses 660.

    2) (Blue) EZNEC NEC-2D engine with the Leeson stepped diameter correction in use. Segmentation is set at 20 segments per wavelength, the EZNEC "conservative" guideline, with three segments in the center section of each element to allow centering of the source. Total segments 82.

    3) (Green) EZNEC NEC-4D engine, no Leeson correction. Total segments 82.

    4) (Magenta) NEC-5 from LLNL. Segmentation density at 20 segments per wavelength with an even number of segments at each element center since with NEC-5 sources are placed at segment junctions rather that at segment centers. (Paraphrasing the EZNEC Help: Technically with NEC-2/4 a source is distributed over the entire segment at which it's connected but it's usually envisioned as being located at the segment center.)

    5) (Cyan) NEC-5 at 400 segments per wavelength. Total segments 854.


    It's pretty clear that the NEC-5 engine with a high segmentation density (cyan) comes closest to matching the MMANA results (red). So why was it necessary to use such a large number of segments? As the NEC-5 Validation Manual points out, for dipoles and Yagis NEC-5 converges to the correct solution more slowly than NEC-4 or NEC-2.

    Using NEC-5 at 7.150 MHz, here's a comparison of various metrics as the segmentation density is increased from 10 to 1000 Segs/WL in logarithmic steps.


    A density of 400 Segs/WL was somewhat arbitrarily chosen as the point at which all of the metrics had "flattened out" (converged) to a relatively stable value.

    As to Nick's question of which scenario is "more correct", probably the only way to know for sure is to measure the SWR response (with the measuring instrument reference plane at the antenna feedpoint) and compare that to the modeled results. My guess is that in this case either MMANA with "tapering segmentation" or NEC-5 with a relatively high segmentation density would be closest to reality.

    Dan, AC6LA
    EA1DDO and UA3TW like this.
  4. UA3TW

    UA3TW Ham Member QRZ Page

    Many thanks, Dan. It must have positive practical consequences. The issue was also discussed at qrz.ru forum. I dare quat the GOOGLE translated post of one of the participants by the avatar Peter99
    And here is what is obtained from 4el_40m_18m_alt_tapering.maa located on the wanderer indicated in P.S.
    MMANA-GAL: Z = 50.91 + j0.60 Ohm
    NEC-2: Z = 16.29 + j24.76 ohm
    NEC 4.2: Z = 46.05 + j2.08 ohm
    NEC 5: Z = 51.51 - j2.62 Ohm

    As you can see, the differences in the calculation in MMANA, NEC 4.2 and NEC 5, in contrast to the calculation in NEC-2 (used in EZNEC if there is no license to use 4.2), are not very large. It has long been known that the NEC 2 does not behave very well with tapering conductors.
    Last edited: Sep 4, 2020
  5. UA3TW

    UA3TW Ham Member QRZ Page

  6. UA3TW

    UA3TW Ham Member QRZ Page

    I was making a model of insulated wire in EZNEC. Each wire has it's own description for insulation or can be bare. When I converted the model to AutoEZ I found out that all wires are alike in this respect, all bare or equally coated. Is that really so?
  7. EA1DDO

    EA1DDO Ham Member QRZ Page

    Dan, would be good to add NEC4 high segmentation density to the tests.

    Great job, as usual.
    Thank you

    73, Maximo
  8. AC6LA

    AC6LA Ham Member QRZ Page

    It is possible, although slightly inconvenient, to use different insulation settings for each wire of an AutoEZ model. See "Wire Insulation - Multiple Settings" in the AutoEZ documentation.

    Dan, AC6LA
    UA3TW likes this.
  9. UA9LIF

    UA9LIF Ham Member QRZ Page

    Here are 40m 4L Yagi in free space modeled in FEKO
    UA3TW likes this.

Share This Page