Problems with counter-propagating plane waves in HFSS 11
时间：03-31
整理：3721RD
点击：
Hi everybody,
I am running simulations on a slab symmetric dielectric/pec based accelerator. The structure is periodic in the z direction (I've used master & slave bd's on the faces normal to the z axis to account for this) and invariant and effectively infinite in the x direction. Since the latter fact means that the x component of the E field should be 0, I have employed perfect H boundaries at the two faces of the structure normal to the x axis. If you were to travel from the origin along the y-axis in the positive direction you would see:
1)bottom face of the structure. There is an incident plane wave with k=+1 in the y direction and E=+1 in the z direction on this face. Accordingly, there is a radiation boundary here that I have classified as an incident field and use as a reference for S11 measurements. The face is on a vacuum rectangular prism which extends to y=.75 microns. At this point, you will encounter a thin (50nm) PEC followed by a 59nm layer of dielectric. After the dielectric there is a vacuum gap of varying thickness (but on the order of 100 nm) followed by another dielectric layer of 59nm, another PEC of 50nm thickness, and another vacuum of 750nm. At the top face of the final vacuum is a radiation boundary that I have been simply classifying as a radiating only boundary.
However, Since I now need to hit the structure with a plane wave from both the positive and negative y directions, this set up is no longer adequate. I created a second plane wave which is incident on the top face of the top vacuum prism (with k=-1 in the y direction and E=+1 in the z direction) and classified the top face as an incident field boundary.
When I validate the setup, HFSS has no problems with it. Furthermore, the simulation consistently runs without crashing. My problem is that I get the exact same results with 2 counterpropagating plane waves as I did when I only had 1 plane wave hitting the structure from the bottom. It is as if HFSS is ignoring the second wave. When I delete the first incident wave and make the suitable adjustments, the resulting fields flip with respect to the xz plane at the center of the structure, as one would expect from symmetry considerations. So when the second plane wave is by its lonesome, HFSS recognizes it. When I then try to re-add the 1st plane wave, I now get the flipped fields again! Clearly, HFSS is not taking into account the second incident wave.
My guess is that HFSS doesn't like having plane waves propagating in different directions when master/slave boundaries are involved, but I'm not 100% sure of that. However, I would expect the setup to not pass validation or simulation if that were the case.
Does anyone have any ideas as to what's going on here? Even better, could you assist me with finding a solution to this problem? If it is indeed impossible to use both periodic BC and counterpropagating incident waves, would using ports be of any help? I've tried this in the past but have ran into even more problems. Any help would be greatly appreciated.
Thanks,
Josh
I am running simulations on a slab symmetric dielectric/pec based accelerator. The structure is periodic in the z direction (I've used master & slave bd's on the faces normal to the z axis to account for this) and invariant and effectively infinite in the x direction. Since the latter fact means that the x component of the E field should be 0, I have employed perfect H boundaries at the two faces of the structure normal to the x axis. If you were to travel from the origin along the y-axis in the positive direction you would see:
1)bottom face of the structure. There is an incident plane wave with k=+1 in the y direction and E=+1 in the z direction on this face. Accordingly, there is a radiation boundary here that I have classified as an incident field and use as a reference for S11 measurements. The face is on a vacuum rectangular prism which extends to y=.75 microns. At this point, you will encounter a thin (50nm) PEC followed by a 59nm layer of dielectric. After the dielectric there is a vacuum gap of varying thickness (but on the order of 100 nm) followed by another dielectric layer of 59nm, another PEC of 50nm thickness, and another vacuum of 750nm. At the top face of the final vacuum is a radiation boundary that I have been simply classifying as a radiating only boundary.
However, Since I now need to hit the structure with a plane wave from both the positive and negative y directions, this set up is no longer adequate. I created a second plane wave which is incident on the top face of the top vacuum prism (with k=-1 in the y direction and E=+1 in the z direction) and classified the top face as an incident field boundary.
When I validate the setup, HFSS has no problems with it. Furthermore, the simulation consistently runs without crashing. My problem is that I get the exact same results with 2 counterpropagating plane waves as I did when I only had 1 plane wave hitting the structure from the bottom. It is as if HFSS is ignoring the second wave. When I delete the first incident wave and make the suitable adjustments, the resulting fields flip with respect to the xz plane at the center of the structure, as one would expect from symmetry considerations. So when the second plane wave is by its lonesome, HFSS recognizes it. When I then try to re-add the 1st plane wave, I now get the flipped fields again! Clearly, HFSS is not taking into account the second incident wave.
My guess is that HFSS doesn't like having plane waves propagating in different directions when master/slave boundaries are involved, but I'm not 100% sure of that. However, I would expect the setup to not pass validation or simulation if that were the case.
Does anyone have any ideas as to what's going on here? Even better, could you assist me with finding a solution to this problem? If it is indeed impossible to use both periodic BC and counterpropagating incident waves, would using ports be of any help? I've tried this in the past but have ran into even more problems. Any help would be greatly appreciated.
Thanks,
Josh
Problem Solved:
For those interested, the problem was incompatibility with the Master/Slave BC I had set up. I was able to achieve the same physical structure using Perfect E surfaces instead, and wala (sp?), the two counter propagating laser simulation worked as it was supposed to.