Two Microstrip Lines

How to use openEMS. Discussion on examples, tutorials etc

Moderator: thorsten

Post Reply
w00t
Posts: 4
Joined: Fri 14 Jul 2017, 01:42

Two Microstrip Lines

Post by w00t » Fri 14 Jul 2017, 02:17

Hello,

I'm trying to model a two-layer PCB with two independent MSLs. The bottom layer is a ground plane. The top layer contains the two MSLs surrounded by three grounded strips.

I based my code off of this tutorial - http://openems.de/index.php/Tutorial:_M ... tch_Filter. I have been able to model the geometry correctly and can view it in AppCSXCAD.

I am encountering two problems at the moment:

I tried applying a Gauss excitation to both MSLs through ports, but the energy remains at 0 dB throughout the entire simulation. Am I not applying the excitation correctly, or have I set up the simulation improperly?

I am unable to design the mesh so that the thirds rule applies to both MSLs. Lines 57-64 define the mesh in the y direction. Line 57 creates mesh lines 2/3 outside the metal. Lines 60-61 create mesh lines inside the metal, bounded to 1/3 inside the metal. Line 64 creates the substrate mesh lines. If I uncomment lines 60-61, the mesh is created abnormally and does not follow the thirds rule as I intended. What am I doing wrong? I did notice that all of the tutorials also use at max two mesh initialization statements in any direction, and I'm using four.

I would really appreciate your help. Thanks.
Attachments
MSL.m
(5.51 KiB) Downloaded 196 times

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

Re: Two Microstrip Lines

Post by thorsten » Sun 16 Jul 2017, 22:56

I see quite a few mistakes.
First, you cannot add manual mesh lines to mesh.y after you have already smoothed it (everywhere). The smoothing already adds a lot of lines. If you add more manual lines, you have no idea how they are located in relation to the others and may create very tiny mesh cells that gets smoothed by the next smoothing call...
Create all manual mesh lines first, then do a single smoothing run.

Secondly you have a edge resolution of 1/5000 of the wavelength? That is insane. Make the metal (egde) resolution dependent on how many cells you want to have across the metal strip. E.g. you decide you want 3 cells across the metal strip, then you can calculate the cells width (resolution) you need....

I hope that helps.

regards
Thorsten

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

Re: Two Microstrip Lines

Post by thorsten » Sun 16 Jul 2017, 23:04

Am I not applying the excitation correctly, or have I set up the simulation improperly?
Yes, your MSL port stop z-coordinate must be the bottom of your substrate not the top of your strip thickness...

w00t
Posts: 4
Joined: Fri 14 Jul 2017, 01:42

Re: Two Microstrip Lines

Post by w00t » Tue 18 Jul 2017, 22:59

Thanks for your help, I got it working!

w00t
Posts: 4
Joined: Fri 14 Jul 2017, 01:42

Re: Two Microstrip Lines

Post by w00t » Sat 29 Jul 2017, 00:12

Hi thorsten,

I'm now trying to measure the coupling between two MSLs. I found a forum post where jockeosth is achieves something similar. - http://openems.de/forum/viewtopic.php?f=3&t=466. The difference is that my two MSLs are now facing each other on two PCBs, whereas his MSLs are side-by-side on one PCB.

I'm interested in the s-parameters and impedance around 1GHz. I based my code off of his simulation. With only one PCB and MSL, openEMS outputs 50 ohms and a similar S11 to what I physically measured on a VNA (~-40dB) at 1GHz.

When I introduce a second PCB and MSL, the impedance and s-parameters don't behave as expected. The impedance is 40 ohms instead of 50 at 1GHz. I physically measured S11 and S41 to be around -40 and -30dB respectively at 1GHz. Instead, the s-parameters graph show them to be -15 and -5dB respectively at 1GHz.

My guess is that my setup is incorrect and this may be a meshing problem. The mesh difference between the one MSL simulation and the two MSLs simulation is the addition of new z-mesh lines for the second PCB (line 46). Do you have a sense of what I'm doing incorrectly to get these results?

I've attached the simulation file for the two MSLs and the corresponding graphs. Thank you for your help.
Attachments
two-MSLs.png
two-MSLs.png (95.99 KiB) Viewed 4598 times
one-MSL.png
one-MSL.png (76.91 KiB) Viewed 4598 times
msl.m
(5.9 KiB) Downloaded 190 times

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

Re: Two Microstrip Lines

Post by jockeosth » Thu 17 Aug 2017, 11:54

Hi woot.

I think your mesh resolution is too big? If I do a diff(mesh.y) I see about 4200-4400 um. Your MSL is only 720um wide. I like to use the 1/3 2/3 rule and have at least 2 mesh lines inside the MSL and at least 1 line between the two MSL in the z-direction.
It also seems your substrate is not completely filled from bottom GND to top GND plane. There is an air gap between the two MSL, this will change the coupling between the two lines compared to if the substrate is used between them.
I would try to use something like edge_mesh = [-1/3 2/3]*strip_width/10 or increase f_max to get a better resolution.

Hope this helps
Regards
Joakim

w00t
Posts: 4
Joined: Fri 14 Jul 2017, 01:42

Re: Two Microstrip Lines

Post by w00t » Thu 17 Aug 2017, 20:39

Hey jockeosth,

Thanks for your suggestions! I went through a ton of trial and error in the past few weeks and ended up implementing just what you suggested. One thing to mention is that I'm modelling a system of two PCBs that are facing each other, which is why there is an air gap in between the two ground planes. The results look very similar to what I saw on the VNA! I've attached them in case you're interested.
Attachments
results.png
results.png (70.63 KiB) Viewed 4509 times
msl.m
(5.82 KiB) Downloaded 203 times

Post Reply