printed dipole feko
Or using HFSS and NEC?
Yes I have for a printed dipole and a rectangular patch antenna. There was a significant difference. I observed differences in resonant frequency upto 50 - 100 MHz. Though I am not claiming that I setup both the models in an identical way. And the HFSS model seemed to yield closer results to simulation when I fabricated these antennas.
Thus HFSS is based on different method compared to FEKO or NEC it will require preparation of models to be done different ways, with own specifics for both methods. Comparing the results you should be sure, that you have both models really created correctly.
Also it depends on the problem I think - when you can specify problem, which is good to both approaches - it is the best trial to get similar results. But when one program is more suited to the problem than another the probability of different results is higher.
Sincerely, I doubt that the discrepancy between FEKO and HFSS exists because HFSS is more accurate. To the best of my knowledge and experience, if models are properly built (and equivalent), both tools should give very similar results.
With the simple stuff I have been doing,
in general HFSS and FEKO tend to give similar results in far field patterns.
My problem is when generating S-parameters. Generally shape or outline of curve is the same (ie parallel), but it bothers me that I can get S11 = 0.96 for HFSS, and get S11 = 0.91 for FEKO.
Also, this is all simulation based, i have not done any of the tests in reality.
Actually the difference is not so high, probably you need to test the model converegence. If you could attach the model I would try to simulate it with EMC Studio and obtain one more results for comparison.