Confused Over Lack of Consistency of Original HFSS Version 14 Performance

I am using a metamaterial as superstate to enhance gain of antenna.When i use 10 pass with center frequency 2.5 GHz for the frequency range of 1-3 GHz using fast sweep i got good result.But when i use 10 pass with center frequency 2 GHz for same frequency range of 1-3 GHz
i got poor result.In both the cases HFSS shows that convergence criteria is satisfied.I am using HFSS 14 academic licence.My system has 4 GB RAM with 64 bit Windows 7.I am confused.Please help me.

take a look at your mesh and see if there's something blatantly different between the two sims. another place to check is your ports. when you say "center frequency 2 GHz for same frequency range of 1-3 GHz" are you talking about solutiuon frequency in solution setup? That is for determining the mesh size so you should generally set it to the highest frequeny in your range, in your case 3 Ghz.

Metamaterial is very narrow band, might be something for you to check as well.

Correction to my previous post, for fast sweeps, you lose accuracy as you move farther away from the solution frequency fast sweeps are good for resonances or plots with fast changes in frequency performance, you want the solution frequency centered on the resonance/big change in performance. You can force a minimum of 2 converged passes as a tradeoff of repeatability vs. speed.