radiation pattern issues in wideband antenna design

[whirlwindz]

I'm in the process of developing a wideband antenna for EM monitoring applications. It is to be omnidirectional, and have as even a possible radiation pattern both vertically and in azimuth.

I'm currently using CST to design the antenna, and am experimenting with different profiles, such as a biconical antenna, and other variations.

The phenomena concerting azimuthal and vertical radiation patterns confuse me somewhat.

I've simulated perfectly symmetrical, closed shapes, and at certain frequencies the azimuthal radiation pattern will begin to appear irregular, and sometimes regular and 'petal-like' (I don't understand how there can be an irregular radiation pattern in azimuth for a perfectly symmetrical antenna). The the vertical radiation pattern will stay close to the 90/270 degree point for a large portion of the frequency range, but during a portion of it, it will diverge, with lobes pointing up towards 0 degrees and downwards towards 180 degrees.

I understand this is likely due to constructive/destructive interference from the relative phases of the EM waves, but i'm still struggling to get my head around the cause of the phenomena, and the process of rectifying it.


Thanks in advance!

Michael.

 

[Higgler]

Perfectly symmetrical antennas like a bicone should have perfect omni patterns if your input connector is centered. If they don't, then your computer meshing is off in CST. Try to let it remesh and compare to HFSS too.

Is the erroneous frequency the highest?

Domed Bicone with dielectric on the surface to help the upper end maintain a wide elevation pattern seems like a good option. Easy to get 40:1 bandwidth or more.

 

[VEBill]

Is this a consideration?

Modeling an ultra wideband antenna might require more than one model to span the range. Same overall antenna shape of course, but several mesh sizes to stay in the computational sweet spot.

I don't know if this applies to this question or not.

 

[Higgler]

I know CST does great on very wideband antennas because of their Time Domain math approach. HFSS is much slower for ultra wideband.

The meshing on both programs can be driven by the user. If on it's own the CST or HFSS meshes in the high current area's of the feed in a non symmetrical way or not fine enough, the VSWR will typically be more accurate than the patterns, especially at the highest frequencies.

 

[whirlwindz]

Yes, i'm using CST's transient domain solver, which apparently is very robust for wideband simulation purposes.

Erronous frequency is not the highest. Petals are evident at even the lower end of the frequency range when i'm using an exponentially tapered conical shape (think of back-to-back engine valve shapes with the apexes getting stimulated).