Evaluation of crosstalk

[Tomsiik]

Hello,
thanks to Robert's episode about crosstalk, I've decided to try an experiment on my own.

I've designed PCB for testing crosstalk, there are some traces like aggressor/victim with multiple gap cases.
Stackup up of PCB is standard JLC7628 from JLCPCB manufacturer.

The first idea is to compare against software calculator "PCB Toolkit" with my test PCB.
I've got calculated coupled voltage 1.65V, I wanted to measure this voltage on my test PCB.

There is a test connection. All oscilloscope channels have 50Ohm impedance. Test signal has 10Mhz and approximately 5ns edges.

On scope, I've got 116mV pk-pk as NEXT and just 15mV pk-pk as FEXT.
I just want to see a correlation of my measured result with the calculated coupled voltage. I can't imagine what coupled voltage means, how to measure it that I can get the notion.

Is there anyone who can explain what I calculated and what I measured?
Because calculated 1.65V looks very horrible but measured 116mV pk-pk is correct in my opinion...

btw. I've tried to change impedance in the oscilloscope channels but the measured pk-pk voltage is still under the 200mV.

[robertferanec]

The calculated value is a very theoretical value. This is where I talked about it in my video with Ken (the creator of Saturn PCB): https://www.youtube.com/watch?v=fJCRrEf_IH8&t=2192s

Basically, the calculator is super simple and doesn't consider other factors what are important (e.g drivers, PCB properties, etc). When I did simulations, I found out exactly the same thing as you did. I was getting better results than what Saturn PCB was showing.

PS: To see a bigger difference, maybe try to use 2 layer PCB, a 4 layer PCB may be too good (?)

[Tomsiik]

Thank you Robert for your explanation.
I think calculating crosstalk is very difficult due to many parasitic parameters, but it is good to have imagination about the level of crosstalk in specific situation in real case.