Ansoft Designer calculation problem with patch array

EDAboard.com Forum Index -> RF, Microwave, Antennas and Optics -> Ansoft Designer calculation problem with patch array

Author

Message

Mantas

Joined: 16 Jan 2004
Posts: 15
Location: Lithuania

29 Oct 2009 19:43 Ansoft Designer calculation problem with patch array

Hello,
I am having difficulties for simulating patch antenna array with ansoft designer. We designed 4x4 array and tried to do simulation, but we didn't succeeded to reach convergence until out of memory. So I split in half our array and got 4x2 but again the same problem with out of memory. The simulation stops after 5 adaptive passes with 90k unknowns and 62k triangles. This is weird because I just simulated another design 4x4 with no problem at all. I have to mention that this design have additional layer of parasitic patches.
I am not surprised that I am out of memory, but the main concern is that when I analyze convergence chart I see that after each re-meshing the DeltaS is not improving (attached plot). So the main question is how much memory it requires to converge and will it converge at all? Or maybe the simulation is in some way not correctly setup?

Sorry, but you need login in to view this attachment

dongxl

Joined: 31 Oct 2006
Posts: 9
Helped: 1

02 Nov 2009 7:19 Ansoft Designer calculation problem with patch array

The design of this 4x4_patch may be related to your memory does not matter. The convergence of data with your settings, matching networks have great relationships.

Google
AdSense
Google Adsense

02 Nov 2009 7:19 Ads

Mantas

Joined: 16 Jan 2004
Posts: 15
Location: Lithuania

02 Nov 2009 12:42 Re: Ansoft Designer calculation problem with patch array

dongxl wrote:

The design of this 4x4_patch may be related to your memory does not matter. The convergence of data with your settings, matching networks have great relationships.

Hi,
it is hard to understand what you mean from the replay above. So I understand that feeding lines (micro-strip) have large relationships, but what can be done, just upgrading the system? Or maybe there is some other explanation? Because in near future we are going to purchase new PC for calculation purposes in mind, maybe then someone could recommend some price/performance of system requirements that best suitable for EM simulations?