CST or HFSS， when simulating printed antenna

I just began to study printed UWB antenna.
To my experience (I was used to work on waveguide component), hfss is more accurate than CST.
But I found most researchers prefer to use CST to simulate printed antenna.
Is CST MORE ACCURATE than hfss when simulating printed antenna?
please help.

Accuracy of your model depends on how many parameters you define in which case, HFSS is better as there are lots of options to play with. However if there's any time constraint in your project, its better to go with CST as its easy to learn as well as easy to design structure in there. I learned CST first n then switched to HFSS n now am woking on ANSys..

One major limitation in EM accuracy is the user.
But even if you have a very experienced user, some solver methods are more efficient that other for certain problem types. That is why companies like CST offer many different solver engines, for different applications. For antenna analysis, where problems tend to be large in terms of wavelength, time domain solvers can have some advantages.

ANSYS also offers several solver engines and hybridization between them so the user can have the most efficient path to a solution, particularly electrically large structures. HFSS now currently has: FEM frequency domain, FEM transient, IE solver, FEM/IE hybrid solver (very efficient), and up and coming PO solver. So in reality, for electrically very large problems, time domain analysis cannot be more efficient and maintain the accuracy of a hybrid FEM/IE approach as the mesh is dramatically reduced via the usage of the IE solver for regions of isotropic space.
However, it is true that the limit of any tool is the users familiarity with it. An uber user of HFSS will blow away an average user of CST for the same simulation in terms of efficiency and accuracy.
I prefer HFSS myself...and I have used it for years and have enjoyed correlating measurements ;) However, full disclosure, I have never used CST..never needed to.
Have Fun

Correct.
Not sure about the hybrid approach. What I was trying to point out is that (pure) time domain has advantages over (pure) finite elements for a large number of cells, because the solution time grows "only" linear with the number of mesh cells.

I would imagine that you would be correct in that statement. If pure FEM could be made more efficient, I doubt ANSYS would have pursued the difficulty of creating new IE code and then efforts to hybridize. Also, I have seen that very large simulations can solve quickly in FDTD with relatively small memory footprint, however my colleagues were never quite so confident in the results. But they were not uber users of FDTD, so that could limit their capabilities. HFSS FEM did take a bit longer, but the time it took was actually less than the colleagues who had to modify the mesh repeatedly and manually to converge their results to their satisfaction. HFSS seems to do this automatically and pretty well. But again, I can only speak for HFSS as my experience with other codes is very minimal.
Have Fun!

Ansys might have learned from what FEKO is doing for many years. If you look at automotive antennas, for example, you will find FEKO all over the place.
I think it's what people are trained to use. Over here in Germany (=home of CST) you will find CST MWS mostly, and users will praise the results that they get from CST, and how much easier it is to get accurate results with CST MWS compared to good old HFSS. In the US (=home of Ansoft) it might be the opposite.

I would have to agree with this...HFSS seems to have much more use here than CST in industry and seems to be growing significantly in Academia (more students graduate knowing how to use it). I also would agree that the user ability to use a tool will dictate the performance of the tool...much like a violinist and a violin.
Viel Spatz!