Usually, "waveport excitations" calculates a possible modal transverse field distribution,

so the incident field may be different when the size of the wave port is varied.

In theory, the s11 result can be really differ between lump ports and wave port since the incident field can be different.

Lumped port some times can better mimic a coaxial line feed excitation,

which resembles to the measured results in many case.

However, what mentioned above does not hold every time. Using both of them comparing with the measured result is recommended.

Discrete port feeds the signal to the line at a discrete point, thus introduce some discontinuity. Not only that, but the virtual source connected to the port has some length (e.g.substrate height for microstrip line); which has an additional parasitic series inductance.

Waveport feeds the line with the natural modes of fields and current distributions associated with the port cross section, as if there was no discontinuity at the port.

It is only natural that the matching (i.e. S11) will not be exactly the same for both cases.

The difference shouldn't be that great particularly since you are working in the low GHz. A discrete port has properties of its own (i.e. inductance as mentioned before) which can sometimes interfere with the simulation. Waveguide ports don't normally do this except in rare cases of specifying the wrong boundary conditions. Normally discrete ports are faster to simulate than waveguide ports particularly if you have a very fine mesh. Is this your reason for wanting to compare the two, or is there something more interesting than this?

Things I can suggest are:

1. Check your boundary positions and boundary conditions. I have found sometimes that discrete ports can resonate with certain boundary types to create extra resonances of their own. Easy way to check this is to extend a critical boundary by about 1/4 wavelength and see if results change much. (may cause more simulation time)

2. Check the length of your discrete ports. Longer discrete ports create more problems.

3. I am not sure what type of feed you are using for the antenna. Maybe it is possible in your structure to extend the feed a bit more so that the local effects of a discrete port die down and it is only the propagating wave that enters your structure. You will of course need to ensure that extending the feed changes your S11-dB by less than the error limit. Your S11 phase can be easily adjusted.

4. And finally, it may be worth checking your mesh. I have found waveguide ports sometimes depend on the meshing in its vicinity. This is mostly if you have Auto Meshing turned off, if it is ON this is probably not the cause.

5. Might also be worth extending your accuracy limit a bit lower particularly if it is set to -20dB or higher. Not sure if this will help but may give you a clue as to what is going on.

Hope this was of some help. I would be interested to know the root cause of this when you find it.

Following Chris comments about meshing and accuracy, I would like to add that a common mistake is to attempt getting high accuracy and fine mesh by setting the convergance frequency at a very high frequency (i.e. short wavelength) which is in fact out of operating frequency range. When the structure analyzed is frequency selective and the field energy distribution across it has significant variance, the final mesh can be insufficient for accurate analysis in the desired frequency range.

As a general rule, the convergence frequency shold be set within the pass band.

I agree that you should not expect the same results for the two cases but they should be close if implemented properly such that the discrete port approximates a physical excitation. It may be instructive to examine the modal field and current patterns as that will often reveal mistaken physical assumptions.

The discrete port may not have the impedance you expect.  What does it look like at one end of a transmission line fed from the other with a waveguide port?

If it is not the same impedance as the transmission line, the transmission line will reactively load your patch, and possibly introduce an extra resonance. 

can any body tell me how to make exact equivalent of discrete port like micro strip or coax in bow_tie antenna. how we match impedance to get same results of discrete port?

Microstrip or coax are not discrete ports when feeding a bowtie antenna. They have a finite size and will excite the port with "spread-out" current and voltage distributions. A discrete line feed in cst would be equivalent to something joined to the bowtie by thin wires. You can also have a cst discrete port that has uniform current density between two edges (and possibly two faces) but this is not the same as the distribution from a microstrip or coaxial feed. The difference in current distribution results in different capacitance and inductance around the feed point when a discrete port is used compared to when a waveguide port is used.

Model a short length of microstrip or coax joined to the antenna as you intend it to be and feed that with a waveguide port.

Dear White , Bow tie antenna impedance varies from 150 ohm to 250 ohm. we use feed of 50 ohm with waveguide port excitation in CST. so how to match that using microstrip?

You will need to use a transformer. A simple balun will be suitable. If you look up baluns on the internet it will give lots of examples. A split coax balun will get you 200 ohms which will be ok for a reasonable band. There are broadband stripline baluns such as the Marchand balun. All of these can have most of the parts on the back of the bowtie.

You need a balun of some kind anyway otherwise the feed won't be balanced and will radiate itself and mess up the radiation pattern.

Thanks

The bow tie antenna simulated is simulated at 200 ohm with any feed. Will it give similar results when i successfully match 200 ohm with 50 ohm feed.?

And what are tge ways to model baluns in cst?

You model the structure that you have to build to make a balun, like a microstrip balun or a coaxial balun. Many baluns have one end at 50 ohm, unbalanced, which is a match for coax or microstrip, and the other end 200 ohms balanced, which is a match (nearly) for your bowtie. If your antenna is close to 200 ohms then any feed close to 200 ohms will be reasonable. The connections you need to make will add capacitance and/or inductance but if you are careful these will be small and won't affect the match much.

Hi Majid,

Differences in results with respect to wave guide and discrete port arise sometimes, but the difference should not be to a greater extent. Steps are to be taken to minimise this difference. All the answers above provided useful insight into ur problem. As suggested please check the boundary conditions and the mesh settings also. Sometimes, correcting these two itself would give u the results u are expecting.

Wave guide and discrete port are the two main ports used in CST for the absorption and simulation of power for any device under test. The results cannot be the same due to these reasons:

a. Wave guide port require more energy to simulate the infinitely long wave guide connected to the device under test thus resulting to more lower level of reflections in the computational domain as compared to the discrete port that requires less energy for the computation of the boundary domain.

b. Discrete port perform better for the simulation of lumped element in the structure , so you do not expect to see the same results when you used wave guide as an alternative.

c. Discrete port is more faster in simulation than wave guide port.

d. Wave guide are more accurate for s parameter calculation while discrete port are better for far field calculations.

Good luck.

Excuse me sir @Isah Musa

I have design a microstrip patch antenna to calculate s- parameter , which port is suitable (discrete or wave guide) ?

When I change the port I have a different results , even an impedance mismatch has occurred.

If you use a discrete port for microstrip you will introduce unwanted modes around the port, which will cause a mismatch. You need to use a waveguide port. Read the help and look at examples to find out how big it should be.

@Malcolm white, Thanks alot for your information. Actually I used macros to calculate the wave guide port dimension. It is easy way.

Now I have a good matching (z_real=50, z_imag=0) at a frequency less than resonance freq.

May you kindly suggest to me witch parameter should I change to obtain a required freq.

Thank you again..

Make the frequency you want the centre of the CST frequency range and CST will calculate the port impedance at that frequency.

Yes. Sure.. I have used frequency range(25-31) to get resonant frequency 28GHz

Patch width=5.44

Feed length=2.23

Feed width=1.1

But the result : minimum s11 at resonant freq=27 not 28

!!

Your opinion please, doctor.

I don't know what question you are asking. Do you want the port to be 50 ohms at 28 GHz? Do you want your patch to be matched best at 28 GHz? Something else?

Ok, I'll send you more details to your research gate. So, may you kindly response to my question.

And accept my appreciation.