ad: ProAudio-1

Woud a disguised Cobweb work the same?

Discussion in 'Antennas, Feedlines, Towers & Rotors' started by G1YBB, Sep 8, 2015.

ad: L-HROutlet
ad: l-rl
ad: abrind-2
ad: HRDLLC-2
ad: Left-2
ad: L-MFJ
ad: L-Geochron
ad: Left-3
ad: MessiPaoloni-1
  1. K6FNI

    K6FNI Premium Subscriber QRZ Page

    What's the scale of the diagram?

    Robert, K6FNI
  2. SP3L

    SP3L Ham Member QRZ Page

    Large grid is 0.25m, small grid is 0.05m.
  3. SP3L

    SP3L Ham Member QRZ Page

    Here are the details of the models I simulated:
    SP3L Mk.1
    SP3L Mk.2

    1) Sp3L Mk.2 was my original design after the simulations. However, when constructing the real antenna, I simplified the design and finished it as SP3L Mk.1. I did not add the "wings" in the 10m band radiator. It was not my best decision and I think I will migrate to Mk.2 next summer.
    2) 4nec2 simulations predicted rather low resonant resistance of all cobweb antennas. The SWR results I gathered on the Internet prove that the actual resistance is higher - closer to the optimal 12.5 ohm. However, I hope that the simulator shows the TREND correctly. So, my versions should have bigger resistance than G3TXQ and MP0PZT but not necessarily the values shown in the table.
    3) My goal was to increase the bandwidth on 15m and 10 m bands
    4) When calculating the BW, I assumed SWR=5 at the usable band ends because such SWR will be seen by a TRX as something around 3 due to the losses in the coax; SWR=5 at the antenna end of the coax translates into about 1.5 dB additional loss because of the standing wave if only you have an A.T.U. in your TRX that will find a match; such additional loss was quite acceptable for me
    5) widening gaps between the radiator ends except for increasing the bandwitch, also make the antenna less omnidirectional; this is a trade-off; if you want to have full 15m band coverage and large part of 10m band coverage, you have to accept that such version of cobweb will have radaition pattern more directional (although not as much as a dipole of course); you can clearly see that in the table; it is especially visible in 10m band; I will somewhat reduce this undesired effect in my antenna when I migrate to MK.2
    6) in my simulations, I assumed copper wire with insulation; copper wire diameter 1mm and overall diameter 2mm; this was simulated as distributed inductance but agreed very well with my real antenna;
    7) of course, you should initially cut the radiators with some margin - e.g. 15 cm
    8) I also calculated in the simulator how big frequency shift will happen when 1 cm is cut off from from every radiator end; this prediction agreed with the real world results very well; here are the numbers: 20m band = 35kHz/cm, 17m band = 40kHz/cm, 15m band = 60kHz/cm, 12m band =80kHz/cm, 10m band = 60kHz/cm.

    Now, make your own decisions. :)
    G1YBB likes this.
  4. G3TXQ

    G3TXQ Ham Member QRZ Page

    Any form of cobweb is tricky to model; the segments on adjacent wires need to be equal length and aligned with each other - no easy task, especially where the wires approach each other at the feedpoint.

    As Sp3L notes, his modelling results for Rres and SWR are significantly different from measured results. For comparison, here are the measured SWR results for my single-wire version with 1:4 balun, referenced to the feedpoint:


    Steve G3TXQ
  5. G1YBB

    G1YBB XML Subscriber QRZ Page

    Thanks a lot both. The trim dimensions particularly useful, but this is one of the areas of radio I really love, experimenting with and sharing info about antennas.

    Can't wait to get mine going.
  6. W7CJD

    W7CJD Ham Member QRZ Page

    I would use pop-rivits.

    It involves a hand tool and aluminum or stainless steel pop-rivits.

    That is, unless you notch the spreaders and lash everything in place, for example, with waxed cord for the purpose.
  7. K3LI

    K3LI Ham Member QRZ Page

    it should work if the arms are non conductive. No idea what the radiation pattern would be but I darn sure would try it. It will be fun.

    My current project is extending the arm on my hex beam and try to build a 40 meter radiator for it. It will need to be
    trapped to keep the diameter down. May be it will work, may be not. It is fun to try.
  8. AC6LA

    AC6LA Ham Member QRZ Page

    That sounded like an interesting challenge so I took the bait. I created a general purpose "G3TXQ-like" cobweb model in which everything is controlled via variables. Dimensions below are in inches per Steve's cobweb page.


    The variables for the element half lengths (D-I), for the center to corner distances (L-P), and for the center to balun box distance (S) are all inter-connected. That is, if you increase D the 20m element tips will move closer to the X axis. If you increase L the entire 20m square will get larger and the element tips will move away from the X axis. If you change S the tips will move closer or farther away depending on what it takes to keep D the same.

    To improve the accuracy of the model I used a trick developed several years ago by LB Cebik. The feedpoints for the elements are physically separated by a small amount at the centers but are electrically joined by "zero" (almost) length and zero loss NEC transmission lines. Making the connection between elements this way, as opposed to having the elements physically connected to a common center wire, results in much better Average Gain Test results, hence more reliable impedance and gain values.

    Another trick to improve model accuracy is to make sure that segment junctions are aligned, particularly important at high current regions. Here's an exploded view of the feedpoints.


    For more info on the use of "zero" length transmission lines see here and about half-way down the first section here.

    With all that out of the way it was time to tune the elements for resonance.


    Finally, I ran a sweep from 14 to 30 MHz to compare with Steve's measurements.


    Not bad. I couldn't quite get the minimum SWRs on 12m and 10m that Steve got but otherwise the model and the measurement are remarkable close.

    Next I changed the units from inches to meters, modified the variable values, and added one more variable to model Jacek's "SP3L Mk.1" cobweb.


    Here's how the SWR curves on 20m, 15m, and 10m compare for these two models. Note that reflection coefficient is shown on the left scale with the corresponding SWR on the right scale. With large swings in SWR values this is a more representative way of showing SWR since the difference between SWR 3:1 and 2:1 is much more important than the difference between 10:1 and 9:1. That fact is lost when showing SWR on a linear scale.


    AutoEZ and EZNEC format models are in the attached zip. For those who want to use the EZNEC models with 4nec2 be aware that 4nec2 will not recognize a) EZNEC virtual wires, b) EZNEC transformers as used for the 4:1 balun, or c) EZNEC insulated wires. So with 4nec2 you'll have to a) add a source to the model, b) change the SWR reference to 12.5 ohms, and c) add the appropriate 4nec2 LD7 card to simulate wire insulation.

    This was fun. Thank you Steve and thank you Jacek for the inspiration.

    Dan, AC6LA

    Attached Files:

    SP3L likes this.
  9. KC8VWM

    KC8VWM Ham Member QRZ Page

    There's no computer modeled substitute for sticking it up in the air and actually using one.
  10. G3TXQ

    G3TXQ Ham Member QRZ Page


    I can see the day coming when I shall just have to get a copy of MS Excel ;)

    Steve G3TXQ

Share This Page

ad: Alphaant-1