FEDEVEL
Platform forum

suleymancskn , 06-14-2022, 01:07 AM

Hi Everyone,

I want work multilayer and high speed pcb design at master thesis.
Does anyone have any suggestions on this subject?
If there are some of you who are working on this subject, can you give suggestions on which subjects I will work on?

Thank you!

qdrives , 06-14-2022, 01:03 PM

This is actually more a low speed topic, but in this video https://youtu.be/nPx2iqmVAHY?t=374 Robert has a board with multiple Gnd planes.
Why is there current flowing even on the third Gnd plane?
My question, as is of @robertferanec (as far as I understood), is: does this relate to skin depth? In other words, if the copper were thicker (70um or 105um) would there be no current?
This is not so much a problem for high speed, but switch mode regulators are in the frequency range where that shielding may be expected.

robertferanec , 06-25-2022, 12:58 AM

In my opinion, decoupling may be even a better topic than high speed. If I would be doing something about high speed, than it would be probably about return currents or a visualization how the fields are travelling.

suleymancskn , 08-02-2022, 12:54 AM

Hi everyone,

I decided to study crosstalk in high frequency circuits.
I am trying to develop a method like the attached image.
What is your opinion about this method?
Thank you!

qdrives , 08-02-2022, 03:08 PM

Eric Bogatin has made several video's on cross talk.

What I would find interesting is the difference in cross talk between:
- Two traces where the spacing between them is 3W
- Adding a Gnd in between the two traces (like your board shows, but without the capacitors?)
- Adding stitching vias to the Gnd.
In all there cases the distance between the source and victim traces are identical (although the via's may be problematic or ignore the distance there).
This on a typical 4 layer board with 0.36mm between top and inner layer.

In other words: does a guard ring help reduce or increase the crosstalk.

suleymancskn , 08-03-2022, 07:15 AM

Thank you for your answer @qdrives.

Since it will be a thesis work, I want to add something to the existing methods.For this reason, I want to add capacity on the line between the guard tracks.
I don't know if this method will work or not.
However, if this method works, I believe it will work in reducing the distance between the differential lines.
In cards where high-speed lines are not dense, it can cause the design to be completed reliably in 2 layers and reduce the cost.

I need your comments and ideas.

qdrives , 08-04-2022, 01:07 PM

Question is what would be better: zero ohm resistors or capacitors?
2 layers may be low cost, but for most of my designs it was to few layers.
But, yes, perhaps interesting too as it makes it somewhat more extreme.

suleymancskn , 08-04-2022, 10:21 PM

In fact, the components there can be replaced.
My goal is to make various experiments by opening pads in those locations.
Does this topic seem interesting to you too?

At the end of my work, I want to offer something new and see people benefit from it.

qdrives , 08-05-2022, 06:00 PM

A benefit for me would be the tests I mentioned (guard "ring" on 4 layer board).
I did place a guard ring several times in the past, but after seeing some newer presentations, it may have made it worse than without.

suleymancskn , 08-08-2022, 12:38 AM

Hello, I rearranged my design as I sent in the attachment.

All signals are set to 50 ohms in Altium Designer to provide the required impedances in 2 layers.
Calculated as coplanar impedance.
As you mentioned, in the next step, I can do this method on 4 layer boards.


As seen in the images, I will apply an input from one Sma connector and look at the signal on the other.

But I'm not sure how to do it in differentials.
I do not have the necessary equipment. Do you know someone who can help with this?
Simulation can also be effective. Can someone help me with this too?

qdrives , 08-08-2022, 01:50 PM

Eric Bogatin made some videos where he did cross talk measurements (perhaps it can be found here: https://www.bethesignal.com/bogatin/)
One of the methods he used was with a simple hex driver and an oscilloscope - like this video: https://go.teledynelecroy.com/e/4839...t2nuWJG0S-sxLI
I do not know if I can help much. Neither with simulation (I do not have the tools) nor with equipment. I do have a 100MHz scope, but that is it.

suleymancskn , 08-09-2022, 02:33 AM

Thanks for your help @qdrives
I am not sure if video can be used as a crosstalk measurement.

I have a Siglent SSA3021X Plus Spectrum Analyzer. I can guess measure single ended trace crosstalk.

I ordered the boards shown above.
I'll at least test the single ended ones when they arrive.

But I still need help with differentials and simulation.


Do you know anyone who can help with this?

qdrives , 08-09-2022, 02:03 PM

Have you watched the video?

Another source of information: https://www.linkedin.com/posts/berts...edium=member_d esktop_web
Link to the article (if LinkedIn does not work) https://www.signalintegrityjournal.c...-and-crosstalk

suleymancskn , 08-26-2022, 07:42 AM

Hi @qdrives

There is progress in my work.

So far, there has been no application that leads to a result.
But there has been progress in how the measurements are made and how they are applied.

I will inform you about my work as soon as possible.

qdrives , 08-29-2022, 01:59 PM

Not entirely sure what you meant with that sentence.

I am curious what you will find.