Is it a good practice to use a smaller dumpbox than actual pcb ?

[sprhawk]

Hello, I'm new to openEMS here.

I have issues to the RF traces on my PCB board, I think I can use a smaller dumpbox instead that hold the all PCB board. Am I correct about it?

Here is my FreeCAD showing:

[VolkerMuehlhaus]

It depends how you want to evaluate your data. The dump box is for visualization, so if you only need to visualize a part of the model, sure that will work. It avoids excessive file sizes of the field dump.

That said, in my models I usually don't use any dump box because I don't need to look at the fields, and instead look at S-parameters are reflected pulses. It really depends what you want to investigate.

Or maybe you mean that your model boundaries include only a part of your PCB? That is possible as well, if you want to explore only that PCB section.

Regarding your model screenshot: You have extended the center conductor at the coax pads to the boundary. I would not model it that way, and instead use a lumped port between the center conductor pad and ground pad.

Your entire ground pad layout looks a bit strange. Also, there are too few vias to connect the top gound with the inner ground layer along the RF path, and also at the coax ground pads. This is not good RF routing.

[sprhawk]

Thanks for reply.
I'm really newbie to openEMS and RF design, and I found little tutorial or manual about openEMS. I learned it from a series about openEMS and KiCad from youtube videos.

That said, in my models I usually don't use any dump box because I don't need to look at the fields

I thought I need a dumpbox to collect the data. I used FreeCAD-export-openEMS plugin to do so, I can see dumpbox or probe to collect data.

and instead look at S-parameters are reflected pulses

I think I need it too. If I didn't set a dumpbox, how can I get the data ?

You have extended the center conductor at the coax pads to the boundary

Yes, I learn it from the youtube videos. In the video, he taught that make a trace extension to the boundary and set the boundary as PML to remove the reflection from the end of the trace. So I did it.

Your entire ground pad layout looks a bit strange

Could you please tell me more about it? thanks

there are too few vias to connect the top gound with the inner ground layer along the RF path

Yeah, I know, but there are some components on the other side, I cannot set more vias at the moment. I'll update it in the next revision.

That is the reason I really want to use simulation to tell me more about my layout. I know the simulation itself will not be accurate if I don't know what is right, but for me it is what I can get right now. I really need it !

[sprhawk]

instead look at S-parameters are reflected pulses. It really depends what you want to investigate.

So, to my understandings, you can just put a port on one end of the trace, then run simulation, then get the S-parameter of the single port ?

[VolkerMuehlhaus]

In the video, he taught that make a trace extension to the boundary and set the boundary as PML to remove the reflection from the end of the trace. So I did it.

That is one possibility. You would need to make sure that your extended line is aligned with the top and bottom grounds. In your screenshot, it looks like the signal line extends a bit more, that would cause issues. They all need to be aligned.

You can then look at the input reflection at that port. The dump box (field plot data) is not required, you only need that to visulize the fields.

However, in your case there are some flaws the transition from the line to the coax connector, and you would not see that if you simply extend the line to an absorbing PML. You will also get some extra reflection at the coax pads. Overall, I would not do that, and instead create a lumped port at the coax pads, so that we have one input port and one output port. This might require some more manual work with your model, I am not familar with FreeCAD interface and create all my models from code.

I cannot set more vias at the moment.

I understand, but that might cause major issues, because it insufficient connection between your ground paths can create funny resonances. Running a simulation is indeed useful to check this, but make sure your termination at the end of the line (coax connector) agrees with your physical setup.

[sprhawk]

Do you suggest me to remove unrelated part of the PCB to simulate the RF trace alone like this?

[VolkerMuehlhaus]

I can't comment because I don't know the FreeCAD export workflow. Hopefully someone else can guide you on this topic.