Question about s-parameter causality changes on the same PCB

[yefj]

Hello , I am measuring two differential lines one is 3 inch the other is 6 inch over a span of 70 GHz.
one differential line shows 92% percent causality CQMi while the other is 66% CQMicausality.The PNA is well calibrated.
why is that ? causality means the quality of the s-params is not good .
Why one trace give good quality s-params while the other in not?
Thanks.

 

[Rive]

FR4?
Weave effect can do that.

 

[Baluncore]

3 inch : 92% percent causality.
6 inch : 66% percent causality.
Is what you would expect if the line was not correctly terminated.
How do you terminate the line?
What is the impedance step at the driven end, and at the termination end of the line?

 

[yefj]

Hello Baluncore the 6 inch line is the one giving better causality.
both differential lines are 80 ohms and have two SMA on each end as shown bellow.
bad termination will just create reflections ,bad insertion loss.
How bad termination can cause causality issue in CQMi?
Hello Rive, Why weave effect can do causality issue?
Thanks.

 

[yefj]

FR4?
Weave effect can do that.

How exactly can non stable dielectric constant make the s-params non casual?
same thing says Balun core with bad derminations.
why impedance discontinuety can do such a thing.
impeedance discontinuty makes bad insertion loss.
How its linked to CQMI causality?
Thanks.

 

[Baluncore]

How bad termination can cause causality issue in CQMi?

Because internal reflections destroy causality by diluting the reflected signal.

 

[yefj]

Hello Balun core, this is very intersting.if I understand correctly then CQMI is doing error vector magnitude on the samples of the s-params as shown below.
when the dielectric coefficient is not constant then I agree the reflection coeffient get bad so I expect the SDD21 to go bad because of micro reflections.
why these micro reflections make the causality test go bad?


https://www.signalintegrityjournal....andard-for-iinterconnects-up-to-50-ghz?page=2

 

[Baluncore]

why these micro reflections make the causality test go bad?

Maximum entropy detection involves removing recognisable waves from the signal. Each time you subtract a wave from the signal, there will be rounding errors, and the noise floor will rise. At some point, there will only be noise remaining. When should you stop believing you still have valid data. Maybe the causality score, indicates the validity of the data.

 

[yefj]

Hello Balucore, Very intresting what you said could you please help me with the following so I will see the situation in more detail?
1.Do you have some good CQMI manual which mathematicakly shows how we process the s-params?
2.Could you recommend me basic EM simulation physical of uneven dielectric constant so i will simulate and see the S-params are getting non casual?
Thanks.