# In electromagnetic simulation, how much does the feed model impact an antenna's directivity performance and E-field phase readings?

Question: How much would a simplified feed model in an EM simulation alter an antenna's directivity and the E-field phase reading when compared to using a more complicated/realistic feed model?

I am simulating a set of custom metal pieces using MATLAB antenna toolbox (MATLAB uses Method of Moments). Originally, I wish to study the 3 behaviors of these metal pieces when they are excited by a source: 1)the directivity, 2)the E-field phase readings at a far field distance, 3)the s-parameter(S11).

In brief, I construct a metal piece, save it as a .stl file, import such file, mesh it, assign the feed, and run the simulation for the three behaviors mentioned above. In MATLAB, it seems that delta-gap source feed is the only option for simulating a custom input .stl shape.

For example, I generate and mesh the following metal: I compare the directivity when I assign 1 edge, 3 edges and 4 edges near the [0,0,0] point: Their radiation pattern/directivity respectively: It seems to me that 3 edges and 4 edges polygon source feed yields similar result when compare to 1 edge source feed.

Would be nice to have an answer to following questions:

1. The 1 edge source I am understanding it as the vanilla-ish delta-gap source feed input model. Then how does the polygon input feed work? Is it a derivative of the 1 edge delta-gap source feed which approaches a more realistic result? I could not find enough documentation on the MATLAB site. If someone from stack knows it would be great to hear it.

2. At this point I am not expecting to get a very accurate s-parameter result from any of the feed-cases. I still wish to study the directivity and far-field E-field phase reading. Which brings back to the question: How much does a simplified feed model impact the accuracy of directivity and E-field phase reading? I have quickly skim through a few papers around EFIE source model, and these investigations only delved into input impedance accuracy, does it mean that it does not impact the overall radiation pattern much? In other words, if I continue on using a 3-edge polygon or a 4-edge polygon feed as demonstrated in the example, would the results be close to realistic? If not, any way to ensure I would?