As an expirience I tested one the same model of patch antenna in HFSS on AMD and Intel Core I7 processor and have different results. How can trust HFSS results after this example?
Any had the same problems?
How different? I would be surprised if it significantly different. Codes like HFSS are now very good. You need to compare power, not dB, and beware of changes of angle between 0 and 180 which can happen for very small differences.. Looking at the difference on a polar plot or Smith Chart is helpful.
If it is very different then Ansys will want to know.
Thank you, Malcolm White ! I researched VSWR. Of course, this difference is small, but it is enough to jump over the VSWR = 2.00 limit (see attached files).
The differences are a bit larger than I would have expected, but that is based on my expectations of modern codes rather than experience with HFSS on different processors. It could be due to different precision or rounding rules in the two processors. The two results are closer than many or even most comparisons I see between simulated and measured.
Out of my own interest, I am confused that the ripple on both sets of results changes in period in a way that seems to be inversely proportional to frequency. Mostly, ripple like that is due to 2 reflections and has the same period across the plot. Maybe your structure is log periodic or similar.
I agree with your comment, dear Malcolm White , that It could be due to different precision or rounding rules in the two processors. And you are all right regarding the ripple on both sets, which can be the result of not-optimal matching of antenna. Thank you for your participation in this discussion.
- Badges
- Science topic