signal integrity ringing proper interpretation of VIA Z-param

[yefj]

Hello , i have build a structure of two traces and via going threw a plane as shown below.
Its supposed to be purely inductive .
I know there is also place capacitance from the plane .I got Z11 imaginary part very negatice which means that its capacitative trace.
i didnt even put decoupling capacitors. where did i go wrong seeing the inductive behavior of the trace?
Thanks.

 

[psach27]

Hi!

Microstrip lines also have a capacitance with respect to ground plane. Could you give us more details about the geometry of your problem?
I noticed that you via is almost touching ground plane. What clearing space do you have? Proximity between via and gorund plane increase your capacitance.


Regards

 

[volker@muehlhaus]

I got Z11 imaginary part very negatice which means that its capacitative trace.

Z11 is the input impedance into the line with all other ports OPEN.
This means you see the capacitance of the open ended line.

 

[D.A.(Tony)Stewart]

Looks like you measured between the trace and the floating ground, but poorly defined where the source and destination were placed and 2D geometry with a 3D problem. What would you expect from L/R here or RC ? What does Z21 mean? How is it measured?

 

[volker@muehlhaus]

What does Z21 mean?

In the text he wrote Z11 but in the plot we see Z21. Anyway, both parameters are defined with an OPEN at port 2, so there is no current flow at the load and thus we don't see the inductance that he expects.

To see the series inductance from port 1 to port 2, we would use
Lseries = -imag(1/Y21)/omega

The result includes the lines between ports and via, and the via itself. Proper de-embedding of feedlines is required to calculate the via inductance without lines.

 

[D.A.(Tony)Stewart]

That's what I hoped he would realize. With no R in L/R measure Z21 where the current is thru the line capacitance. The response below 100 kHz appears to approach Z21=0 at f=0 may be a measurement/computational limitation. Understanding how Z21 is computed with a current source and plotting phase may help. If plotted on a log f scale, the capacitance ought to yield a straight-line slope. Showing a higher f then would result in some 1/4 wave effects .

 

[yefj]

Hello volker, I want to see my transmission line as a series RLC circuit. i have all Z-params and Y-params is there a good manual where i could learn and find more transitions as you shown in the formula bellow?
Lseries = -imag(1/Y21)/omega
Thanks.

 

[volker@muehlhaus]

I want to see my transmission line as a series RLC circuit

I understand your need, but the solution is not so simple.

There are tools of the "broad band SPICE" type which can exctract an equivalent circuit model from S-parameters.

But let's see how to do that manually:
For your case, I would split the EM model into a "line" piece and a "via" piece.
For the line, we can calculate the corresponding R'L'G'C' elements, or simplify that using Zline = sqrt (L'/C') and extract that from simulation. To get the C', you can simulate an open ended piece of line. To get the L' you can simulate a short (less than lambda/20) length of line that is shorted at the end.

For the via (excluding the line!), depending on frequency we can use a pi model (series L and two shunt C) or see if a series L is good enough. To get the via alone (excluding the line!) from your EM results, you would need reference plane shift which your EM solver might not provide. In that case, you would do the de-embedding yourself, for example in ADS using the S2P de-embedding element.

This is not trivial, but certainly possible.

 

[yefj]

Hello Volker ,is there some manual i could use for the formulas you presented?
Lseries = -imag(1/Y21)/omega
Thanks.