Stripline simulation - setup problem?

How to use openEMS. Discussion on examples, tutorials etc

Moderator: thorsten

jockeosth
Posts: 13
Joined: Wed 11 Jan 2017, 20:24

Stripline simulation - setup problem?

Post by jockeosth » Wed 11 Jan 2017, 21:54

Dear all

I try to simulate a simple stripline between two ground planes. The simulated S11 seems to be too high compared to what I expect.
As a starting point I use a online calculator tool. http://wcalc.sourceforge.net/cgi-bin/co ... ipline.cgi
See attached stripline.m file for my setup. I am not sure what is wrong but I suspect S11 to be much lower.

Any help or advice?

Best regards
Joakim
Attachments
stripline.m
(5.09 KiB) Downloaded 188 times

thorsten
Posts: 1388
Joined: Mon 27 Jun 2011, 12:26

Re: Stripline simulation - setup problem?

Post by thorsten » Wed 11 Jan 2017, 22:50

Hi Joakim,

I think the problem is that you use a lumped port that only connects your stripline to the bottom ground. The propagating mode will surely adept after some short distance, but your port will still record false current and voltages. That will lead to wrong S-parameter. My suggestion ist that you use the AddStripLinePort. As the name implies, that might be the better choice here :D
This port will create an excitation (and termination) to the top AND bottom ground and especially record multiple voltages (against top and bottom) and currents at some distance to the excitation.

Furthermore your mesh is a bit overkill in y-direction. Do not use 20 lines to resolve the width of the strip. Instead try to use a 1/3-2/3 rule for the stripline edges (have a look at the MSL tutorial) and maybe 4-5 lines for the width.

Additionally add some mesh lines to the edges of the outer metal box.
Or even better, use a PEC BC instead of the outer ground-plane boxes. They are far enough away to make no difference...

regards
Thorsten

thorsten
Posts: 1388
Joined: Mon 27 Jun 2011, 12:26

Re: Stripline simulation - setup problem?

Post by thorsten » Wed 11 Jan 2017, 23:31

I have created my own simple stripline example to test if the port still works. And it does to my surprise ;). See attached.

The boundary conditions in z and y are the metal box surrounding the strip. In x-dir I use PML to terminate the line.

Notice the excitations and multiple voltage and current probes along the line:
stripline.png
stripline.png (87.25 KiB) Viewed 6124 times
Attachments
stripline.m
(2.72 KiB) Downloaded 206 times

thorsten
Posts: 1388
Joined: Mon 27 Jun 2011, 12:26

Re: Stripline simulation - setup problem?

Post by thorsten » Thu 12 Jan 2017, 00:17

I have create a (maybe) new tutorial. A StripLine to MSL transition. I just have not yet optimized the transition at all. But it works quite ok already.
StripLine2MSL.png
StripLine2MSL.png (60.41 KiB) Viewed 6124 times
Let me know what you think.

regards
Thorsten
Attachments
StripLine2MSL.m
(4.2 KiB) Downloaded 216 times

jockeosth
Posts: 13
Joined: Wed 11 Jan 2017, 20:24

Re: Stripline simulation - setup problem?

Post by jockeosth » Thu 12 Jan 2017, 22:21

Hi Thorsten!

Thanks for your reply. Now results looks good. I like your new tutorial about stripline to MSL transition and I think it is very useful. Some other questions:

For the feed shift, what is the rule I need to follow? Should the feed shift be a value that is higher than the number of cells for the boundary condition, like in your example a PML_8 is used and feed shift is 10*resolution = 10*250.
If I want the stripline to change shape, can I just create a box that overlap the stripline port with the same width and later (after some length) only the metal continues, for example if the stripline have some turn or get wider/thinner.

Best regards
Joakim

thorsten
Posts: 1388
Joined: Mon 27 Jun 2011, 12:26

Re: Stripline simulation - setup problem?

Post by thorsten » Fri 13 Jan 2017, 00:02

For the feed shift, what is the rule I need to follow? Should the feed shift be a value that is higher than the number of cells for the boundary condition, like in your example a PML_8 is used and feed shift is 10*resolution = 10*250.
Yes, if you use a PML to terminate the line, you need to make sure your feed is not inside the PML. By default it is zero (aka feeding at the start). Additionally you need to specify a 'Feed_R' to terminate the line with a lumped resistor instead of the pml.
If I want the stripline to change shape, can I just create a box that overlap the stripline port with the same width and later (after some length) only the metal continues, for example if the stripline have some turn or get wider/thinner.
I just used the stripline port all the way for convenience. No need to do it like this. You can continue after e.g. a short port however you like. Maybe you try to create one port in x-dir and another in y-dir and make a curve conection inbetween? ;)
But the transition tutorial shows for example that you can do afterwards whatever you want...

regards
Thorsten

jockeosth
Posts: 13
Joined: Wed 11 Jan 2017, 20:24

Re: Stripline simulation - setup problem?

Post by jockeosth » Sun 05 Feb 2017, 21:11

Hi Thorsten!

I try simulate a coupler in stripline, basically two lines close to each other. It seems to work but I have enter a problem that the simulator reporting "nan" for the energy when simulating for some specific values of the spacing between the two lines. For example with a spacing of 70um and 150um work but not 100um. How do I avoid this kind of problem?

Another question regarding the AddStriplinePort. If I have an asymmetrical stripline, how about the "height" parameter? It seems to work in the case with only one plane above and below the stripline if I make sure the hight is large enough to touch both sides of the bottom/top planes in case they have different distances to the stripline.

Best regards
Joakim
Attachments
stripline_CouplerAsymetric.m
(5.6 KiB) Downloaded 174 times

thorsten
Posts: 1388
Joined: Mon 27 Jun 2011, 12:26

Re: Stripline simulation - setup problem?

Post by thorsten » Sun 05 Feb 2017, 23:13

Hi,

I have no idea why it is getting unstable for 100. But I fixed some issues with your mesh in y-direction:
- missing 1/3-2/3 rule on both inner sides of the striplines
- commenting out the odd additional 5 lines in y-direction which must be a left over and were just messing up the mesh
Unfortunately that does not explain why it is unstable. But with that all fixed, it runs fine...

Furthermore the substrate thickness parameter "height" in the ports is meant both on top and bottom, your port was sticking out half the way on the bottom and top.
If I have an asymmetrical stripline, how about the "height" parameter?
Unfortunately that is not possible at the moment with that port.

regards
Thorsten

thorsten
Posts: 1388
Joined: Mon 27 Jun 2011, 12:26

Re: Stripline simulation - setup problem?

Post by thorsten » Sun 05 Feb 2017, 23:22

Well I found the problem.

Code: Select all

mesh.y = SmoothMeshLines([-substrate_width/2 -offset-(strip_width/2+edge_mesh) offset+strip_width/2+edge_mesh substrate_width/2], resolution/4 , 1.5 ,0);
mesh.y = [mesh.y 0 linspace(-offset,offset,5)];
The first smoothing would create (by chance) a mesh line very close to 0 ( 5.6843e-14). In the second line you would add another "real" zero. That killed the FDTD coefficients...
Rules for you to remember: Do not first smooth the mesh and than add random lines in the same area again! That will only result in trouble always!

This is by the way the corrected mesh lines:

Code: Select all

mesh.y = SmoothMeshLines([-substrate_width/2 -offset-(strip_width/2+edge_mesh) -offset+(strip_width/2+edge_mesh) offset+(strip_width/2+edge_mesh) offset-(strip_width/2+edge_mesh) substrate_width/2], resolution/4 , 1.5 ,0);
Nothing more needed!

One more hint. Let the strip-lines run into the PML for termination (and do not use a lumped "Feed_R") ... Use the "FeedShift" option to make sure to not excite inside the PML...

regards
Thorsten

thorsten
Posts: 1388
Joined: Mon 27 Jun 2011, 12:26

Re: Stripline simulation - setup problem?

Post by thorsten » Sun 05 Feb 2017, 23:36

Hi,

I quickly played around allowing the port to have different substrates bottom/top.

What would be the expected outcome? That the voltage running on top and bottom has the same amplitude? (I think it must have as it is in both cases the line vs. ground).
But that would mean that I have to excite different E-fields on top and bottom because I have to take the different height into account...

Your thoughts?

regards
Thorsten

Post Reply