# Why is Nec2 predicted feed impedance so far from measured?

Discussion in 'Antennas, Feedlines, Towers & Rotors' started by WA7ARK, Mar 11, 2018.

1. ### WA7ARKHam MemberQRZ Page

I recently made a non-resonant vertical antenna, intending to use it on 60m on up. I put it up, measured it on various bands, matched it, and have used it on the air. It works as expected, with the exception that the feed-point impedance is much different than what is predicted by modeling it using Nec2.

The antenna consists of a 40' long vertical pipe standing upright on the steel roof of an aircraft hangar. It is a 2”o.d., thin-wall Aluminum irrigation pipe. It stands on a Delrin insulator about 1.5” high. The bottom of the pipe is 6” above the ridge-line of the steel roof. The roof ridge is 20ft above the ground. This puts the tip of the pipe at 40'6” above the earth.

The steel hangar roof is 65ft East-to-West and 85ft South-to-North; the side and end walls are about 16' high and go all the way to the earth. The pipe stands about 1/3 of the way from South to North, and is centered on the roof-ridge. The roof slopes down with a 1:12 pitch to the East and West. There is a 70' high tower with other antennas at the North end of the hangar.

I was interested in using this antenna on 20m, so I measured the feed-point impedance right at the base of the antenna (actually just below the roof line, standing on a tall ladder inside the building). There is a 8” high-voltage wire that passes through the roof, so I connected a RigExpert AA-600 analyzer between the hangar-end of this wire and the structural steel just below the antenna base.

Shown below is a plot of R and X across the 20m band. Note that the measured R=52 Ohm, and X = -152 Ohms @ 14.175 MHz. The 52 Ohms resistive is surprising; the -152 Ohms is not so much because the antenna is just a bit too short to be 3/2 wavelengths at 20m.

Based on the measured values, I made a matching network which I inserted between the 50 Ohm coax feed-line and the base of the antenna. The online calculator said I needed a series 1.7uH inductor, and 2pF (yes, 2pF) of shunt capacitance across the feed-point. I ignored the 2pF, and put a series 1.71 uH coil made from #12AWG solid Cu house wire on a PVC form. To bleed static charge off the vertical, I also put a shunt 1mH pie-wound RF choke between between the vertical and the hangar steel.

With the matching network in place, the measured SWR from the shack-end of ~65ft of foam RG-8 cable is less than 1:1.05, and it rises to about 1:1.1 at the band edges...

So what is not to like?

I modeled this structure in EzNec 6.0+ using Nec2 (I don't have Nec4). The feed-point impedance predicted by Nec2 is not even close to what I measured above. All of the Nec2 simulations I have tried always shows something close to this:

Thinking that somehow the nearly solid ground plane is effecting the results, I tried modeling the elevated ground plane (the roof) using everything from two simple ¼ wavelength elevated radials 20' above ground to 32 such radials, with very little effect on the EzNec source numbers.

I have tried modeling the roof and hangar side-walls as a grid of wires (670 wires, see next figure, below), with and without the tower on the end of the hangar with virtually no change in the Nec2 predicted numbers. I have tried earthing the base of the hangar side-walls using a Mininec ground scheme; I have tried floating them above the earth using a “high-accuracy” ground, all without hardly any effect on the predicted feed-point impedance.

So, I would appreciate suggestions as to why the Eznec predicted feed-point impedance is so different from the measured results. I would like someone that has Nec4 to try modeling this to see if Nec4 comes closer to predicting the measured values...

btw-I wrote a Free-Basic program to “build” the wire-frame model. Do you think I put all of those wires into EzNec by hand? I can post the Eznec model if someone is willing to see if they get the same results, or so that they could see if I did something stupid in my model.

Mike- WA7ARK

KB4QAA likes this.
2. ### W5DXPHam MemberQRZ Page

"Ours is not to reason why. Ours is to do and die." - don't die but do read the help file.

Direct measurement is better yet. If you have an antenna analyzer ... follow this procedure:
1. Measure the antenna's feedpoint resistance. ...
2. Add a load to the model at the feedpoint and adjust its value until the feedpoint resistance matches the value you measured.

3. ### WA7ARKHam MemberQRZ Page

Cecil,

I have read the help file and it wasn't very helpful. I measured the real antenna with a calibrated AA-600.
Without adding any loads to the model, Eznec predicts that the feed-point impedance is 114 - j351. The measured values are 52-j153.

Without building the physical antenna, and actually measuring the impedance, how would you know what resistance to add?

I am assuming that the added resistance you are describing is supposed to represent the loss in the earth under the antenna? There should be very little loss in my vertical because of a near-perfect, elevated ground-plane out to more than 1/4 wave-length in every direction.

Adding a series load to the antenna can only raise the feedpoint impedance, which will make the error between measured and simulated values even worse!

4. ### WB3BELHam MemberQRZ Page

You might want to post some of the model. Although the way you constructed it sounds like you have separate wires for each side of a box in the ground grid rather than a wire that connects at segments so the model is likely very large.
Here are a few questions I would have.
How many segments are used in the 2 inch dia 40 ft vertical section ?
Where is the source placed?
How is the source connected to the ground grid?

5. ### WA7ARKHam MemberQRZ Page

One minor correction to the antenna description. Should have said:
"This puts the tip of the pipe at 60'6” above the earth."

6. ### W1VTHam MemberQRZ Page

Are you modeling what looks to be a Yagi antenna? It tops loads the tower. It may be useful to just vary the resonant frequency of the tower, as opposed to accurately modeling the Yagi, which may be hard.

7. ### WA7ARKHam MemberQRZ Page

Wire 1 is the vertical. 101 seg per 40 ft so as to put the source close to the bottom.

Here is the only source, second segment up from the bottom.

The wire frame is sparse on segments, so the current step under the antenna seems large. However, I have run a simplified model with only a few radials which had dozens of segments each, and the simulated feed impedance is just about the same...

Note that bottom seg of wire 1 (the vertical) connects to four wires (179, 180, 475, and 494) of the roof grid. The current divides equally and spreads across the roof.

Possibly anticipating your next suggestion, I increased the number of segment in those four wire to 5 segments each. There was only a tiny effect on Z.

Not having uploaded an .ez file before, I am trying a .zip as well as the .ez

ps. The forum software will not allow the upload of "post.ez"

#### Attached Files:

• ###### post.zip
File size:
4.7 KB
Views:
36
Last edited: Mar 11, 2018
8. ### G8HULHam MemberQRZ Page

could you upload the file as plain text rather than an .ez file for those of us who don't use EZnec

Jeff

9. ### N0TZUPlatinum SubscriberPlatinum SubscriberQRZ Page

I've not tried to go through the wire list yet but from the diagrams it appears that there are horizontal wires resting on ground or nearly so. Horizontal wires within a few thousandths of a wavelength can cause strange results; has this been explored?

10. ### K4SAVHam MemberQRZ Page

I suspect your 1 mH choke does not look like a high impedance on 14 MHz. Try disconnecting it and remeasuring with the AA-600.

Jerry, K4SAV

N0TZU likes this.