I have been playing with various antenna modelers (EZNEC, MMANA-gal, 4nec2) and I keep finding the same (counter-intutive) thing. Imagine a simple 5 wire 2m Ground Plane antenna in free-space where the four resonant radials are horizontal.

Wouldn't you expect that the maximum radiation would be a few (maybe 30) degrees above the plane of the radials, kind of like a 1/4λ monopole against a real ground.

All of the NEC2 based engines show max radiation at zero degrees, and the 3-d pattern shows a perfect donut with symmetry along the Z-axis, just like a free-space vertical 1/2λ dipole would have.

Seems counter-intuitive to me. I would expect that the four orthogonal radials should cause the peak radiation point to move upwards from 0 degrees. Is this a limitation of the NEC2 algorithm, or is this a real behavior?

 

Using MMANA I see the same result - provided I have interpreted your meaning correctly. Also, tilting the ground wires symmetrically has little effect on far field and indicates a greater impact on the driving point impedance rather than on other factors.

I built one of the latter type for 2.4GHz work some years ago based on this ...
http://wireless.ictp.trieste.it/handbook/C5.pdf

 

what impedance are you modeling at? the ground plane with the radials at 90 degrees to the vertical is actually a 70 ohm vertical, you have to angle the radials down at a 45 degree angle to get it to 50 ohms.

 

I believe the feedpoint impedance predicted by NEC2. It predicts the textbook result for the following three cases:

1. four ~1/4λ radials in the horizontal plane (no droop). The feed point impedance is ~37+j0 Ω at resonance.

2. four ~1/4λ radials drooped down about 45 degrees. The feed point impedance is ~50+j0 Ω at resonance.

3. four ~1/4λ radials drooped down 90 degrees. The feed point impedance is ~75+j0 Ω at resonance. This approximates the coaxial sleeve vertical antenna:

All three of these cases have the same free-space radiation pattern. That is what I am questioning.

 

I have been playing with various antenna modelers (EZNEC, MMANA-gal, 4nec2) and I keep finding the same (counter-intutive) thing. Imagine a simple 5 wire 2m Ground Plane antenna in free-space where the four resonant radials are horizontal.

Wouldn't you expect that the maximum radiation would be a few (maybe 30) degrees above the plane of the radials, kind of like a 1/4λ monopole against a real ground.

All of the NEC2 based engines show max radiation at zero degrees, and the 3-d pattern shows a perfect donut with symmetry along the Z-axis, just like a free-space vertical 1/2λ dipole would have.

Seems counter-intuitive to me. I would expect that the four orthogonal radials should cause the peak radiation point to move upwards from 0 degrees. Is this a limitation of the NEC2 algorithm, or is this a real behavior?

If the four radials are perfectly balanced, they will contribute nothing to radiation, but simply form the "negative terminal" of the antenna. The only radiating element is the vertical member, and it can ONLY radiate at right angles to itself. I know it seems counter intuitive...but you can easily measure this in the real world.

 

With a 'perfect (elevated resonant ground plane) ' ground the program calculates the radiation pattern using an image antenna formula.
http://en.wikipedia.org/wiki/Image_antenna

 

With a 'perfect (elevated resonant ground plane) ' ground the program calculates the radiation pattern using an image antenna formula.
http://en.wikipedia.org/wiki/Image_antenna

The wiki article implies that for the image antenna to be a mirror image of the driven part, the ground plane is horizontal, and infinite in extent (at least for several wavelengths, anyway). In the antenna I am considering (CubeSat), the ground plane is neither horizontal, nor infinite, or even "resonant".

My question still is: does the NEC code simplify its calculation based on an assumption that the driven monopole has a perfect image? If so, how good does the ground have to be for the image to be perfect? What happens to the pattern if the image is less than perfect?

The ground plane configuration I cannot change (it is the body of the satellite). It is less than 1/4λ on an edge. I tune the driven monopole to bring down the SWR at the operating frequency. I am modelling the satellite body explicitly as wire frame patches (not relying at all on any concept of a "ground" in the modeler.