How to see if microstrip line is radiating or not?

[margarida]

When simulating a microstrip line, with two lumped ports, I have found a 4.5dBi total gain, and almost 10dB of radiating eletrical field.
S11 is -25dB and S21 is -0.2dB. Instead of two lumped ports, I have replaced one with impedance boundary.However total gain is 3.6dBi. How can this happen?

 

[volker@muehlhaus]

I have found a 4.5dBi total gain

I am quite sure that this is directivity, not gain.

Reference for gain is the power into by the device. This should be really small for a transmission line.
Reference for directivity is the total radiated power, so that you can see large numbers even if the amount of radiaton is very small.

 

[vfone]

Gain = k*Directivity, where k (dimensionless) is the efficiency factor (0 ≤ k ≤ 1)
Due to symmetry unbalance in microstrip line, all discontinuity elements make the line to radiate somehow, nevertheless its dimensions.
When the microstrip line was invented in 1952, radiation was the main issue of this RF component.
If compare a microstrip line to a patch antenna you will not see much differences, other than dimensions.

 

[volker@muehlhaus]

all discontinuity elements make the line to radiate somehow

That is true, and we can calculate the maximum total loss (dissipation plus radiation) from the S-parameters given in the initial post.
S11 -25dB means S11=0.0562
S21 -0.2dB means S21=0.9772

Total loss (dissipation + radiation) is then 1 - |S11|^2 -|S21|^2 = 0.0419
So for this model, sum of radiated and dissipated power is 4% of the input power. This already shows that the model has only little radiation.

To get the radiated percentage (excluding dissipation loss) we can simulate a lossless model and do the math above. Or as you mentioned, we can look at the "radiation efficiency" parameter.

 

[margarida]

That is true, and we can calculate the maximum total loss (dissipation plus radiation) from the S-parameters given in the initial post.
S11 -25dB means S11=0.0562
S21 -0.2dB means S21=0.9772

Why do you use 20*log10 instead of 10*log10?

 

[volker@muehlhaus]

Why do you use 20*log10 instead of 10*log10?

S-parameters are voltage, not power