Interconnect sim problem in HFSS9

hi,all
When i simulate the structure based on the attachment,i can't get the result i need.
When i define the excitations,it's said in the manual of HFSS9 that the LUMP port is the proper one,however,When i use 30,20,10 um overlap in my model,the results are not gradual.
Then i use the wave port which i can do deembedding,i enlarge the length of the model,about 10*(length of the metal),but the result is even bad,how can i?
i define no boundary,i wonder if i'm right?
In which situation should i use PML boundary conditions?
thanks a lot! God save me!

1. Don't use the FFS for wideband cases (such Your).
2. You don't need so long structure to obtain the good results.
3. For such structures the 2.5D EM-simulator (Sonnet, MWO, Genesys) more useful.
4. Set the adaptation freq. at the center of the frq. range or slightly shift it up.

The example of corrected structure and results in attached files. If You have another questions - post it here.

Best regards,
Kit-the-great