Negative Resistance (Positive S parameter)

I'm simulating some simple antennas (such as dipoles and loops) using CST MW Studio, but in some frequencies I'm getting positive values (in dB) of S11 and negative values of resistance.

I'm wondering if this have a physical meaning, or it is just a simulation issue. Does anybody have an explanation?

That's a simulation issue. Antennas(dipoles and loops) should not have a positive value for S11. Check the field distribution of the port (input port used to exite the antenna). Are there any requirements, like minimum width and height required for your input port in CST MW studio.

I didn't expect to have such requirements, since I'm using a lumped port...

Is your radiation boundary defined correctly ?

I'm quite sure it's correct, because the boundary is automaticaly defined based upon the lowest frequency of the range. But I suspect that the MW Studio does not behave so well at low frequencies. I am experience these problems specially below 100 MHz.

mateusbatera said:

But I suspect that the MW Studio does not behave so well at low frequencies. I am experience these problems specially below 100 MHz.

Click to expand...

Did you analyze enough time steps to get reliable low frequency data from the FFT?

Hello,

If you simulate small structures (w.r.t. wavelength), |S11| can be very close to 1. Small accuracy issues may lead to |S11|>1.

Other point may be related to Volker_muehlhaus's issue. If you have data interpolation enabled and you use few simulation points, the interpolation scheme may cause a significant error. Can you turn that of so that you see your real simulation results (connected by straight lines for example)?

WimRFP said:

Hello,

If you simulate small structures (w.r.t. wavelength), |S11| can be very close to 1. Small accuracy issues may lead to |S11|>1.

Other point may be related to Volker_muehlhaus's issue. If you have data interpolation enabled and you use few simulation points, the interpolation scheme may cause a significant error. Can you turn that of so that you see your real simulation results (connected by straight lines for example)?

Click to expand...

Thanks for the tips! I will try to disable data interpolation if this option is available. Also I figured out that using the frequency domain solver (I was using the transient solver previously) gave more acurate results.

hello,

When you simulate a physically half wave dipole in free space, you should find a real Z for a frequency somewhat below the half wave frequency (due to capacitive effects). You should expect between 50 and 70 Ohms.

When you put that dipole close to a ground plane (for example 0.05*lambda above it), you will see that the S11 curve travels close to the left edge of the Smith Chart. The ground plane cancels part of the radiation, hence given a certain feed current, the dipole over ground plane radiates less. This results in less real part in the impedance, hence larger S11.

When simulating full wave dipoles, you will notice high resonance impedance just below the full wave frequency. When you place it close above a ground plane, you will find even higher impedance (in the many kOhm range). So the curve travels close to the edge of right side of the Smith chart (around the full wave frequency).

Generally spoken, antennas close and parallel to a ground plane result in strong increase in S11 (or strong increase in return loss) when directly fed from 50 Ohms. You will also notice that the equivalent Q factor of the resonance increases when you reduce the distance to the ground plane, so this makes narrow band antennas.

Try to simulate a thick strip half wave dipole also and note the difference in d(Im(Z)/dfreq with respect to the thin dipole.

In all cases, |S11| < 1. Antennas in air over ground planes simulate well (and fast) when using MoM method. You may find strange results for the radiation pattern when you use infinite dielectric sheets (for example to simulate a patch antenna inclusive the substrate).

mateusbatera said:

Also I figured out that using the frequency domain solver (I was using the transient solver previously) gave more acurate results.

Click to expand...

This happens if the time domain solver is not used properly.

Hello,
In my opinion, the other users gave you the most feasible causes. However, you should take into account the impedance of the source and check if the "impedance normalization" is properly defined. If the source impedance is different from 50 ohm and then the fields are renormalized to 50 ohm, some signals can appear as being |s11|>1, which is phisically not true. So, it is another simulation particularity.

volker_muehlhaus said:

This happens if the time domain solver is not used properly.

Click to expand...

Which solver should I use for electrically small antennas? One of the issues of these cases is the boundary box size. I'm considering using FEKO, for it uses MoM, which doesn't rely in a boundary box.

WimRFP said:

Hello,

If you simulate small structures (w.r.t. wavelength), |S11| can be very close to 1. Small accuracy issues may lead to |S11|>1.

Other point may be related to Volker_muehlhaus's issue. If you have data interpolation enabled and you use few simulation points, the interpolation scheme may cause a significant error. Can you turn that of so that you see your real simulation results (connected by straight lines for example)?

Click to expand...

How can I disable this option in CST ?

You could also play with the mesh size, e.g. use a higher mesh.