### [SI-LIST]: Differential Signal S Parameters

Hi All,

I was extracting 4 port S parameters for differential signal on FR4 PCB

trace referenced to ground with connector and vias in the trace path.

S21 is having resonance at 10GHz where the S parameter file is single ended

version, while same S21 does not have resonance at 10GHz when the single

ended version is converted to mixed mode S-parameters using simulation

tools.

The resonance is expected due to discontinuity in the path but both P & N

of the differential signal is having same discontinuity, hence both P & N

are seeing same amount of insertion loss. Single ended impedance is 50 ohms

while differential impedance is 90 ohms.

So was wondering what is causing a differential signal not have resonance

compared to single ended signal leaving apart the mathematics used in

conversion to mixed mode :-)

Regards

Vinod A H

ah.vinod
5 years 2 months 22 days

### The best answer

You can select the best answer for current question!

Answered byah.vinod
5 years 2 months 20 days

Thanks David, Ken & Jia Gongxian.

As mentioned rightly by David, It was due to mode conversion & fact that

this behavior is related to the even and odd modes propagate at different

speeds due to the non-homogenous nature of the microstrip interconnect.

Also connector too was adding up to the effect.

Regards

Vinod A H

On Wed, Jul 1, 2015 at 8:37 AM, vinod ahwrote:

The trace configuration is as below.

Top layer Microstrip 3 inch --> Blind Via to Inner layer 3 --> Stripline

3rd layer 3 inch --> Blind via to Top layer --> Top layer Microstrip 1

inch --> Connector.

Layer 2 is gnd while layer 4 is signal layer. Via separation between P & N

is 75 mils. Via differential impedance is 48 ohms.

Regards

Vinod A H

On Wed, Jul 1, 2015 at 3:49 AM, David Banaswrote:

Hi Ken,

I assume you meant to address me and not Vinod.

When I used the term "mode dispersion", I was referring to the different

propagation velocities of the even and odd modes of a differential

micro-strip pair, which is the physical structure I'm assuming Vinod is

discussing.

For an excellent treatment on how this dispersion results in far end

cross-talk, please, refer to section 11.11 of Eric Bogatin's book, "Signal

Integrity - Simplified".

Thanks,

-db

On Tue, Jun 30, 2015 at 1:07 PM, Ken Cantrell

wrote:

Vinod,

There are only two dispersive modes for a Microstrip. The first is when

v_g

(group velocity), is less than the v_p (phase velocity), called normal

dispersion. The second is when v_g is greater than v_p, called

anomalous

dispersion.

Under either scenario, I am not clear how this effects (enhances)

crosstalk.

Can you elaborate?

Thanks,

Ken

-----

Answered byah.vinod
5 years 2 months 21 days

The trace configuration is as below.

Top layer Microstrip 3 inch --> Blind Via to Inner layer 3 --> Stripline

3rd layer 3 inch --> Blind via to Top layer --> Top layer Microstrip 1

inch --> Connector.

Layer 2 is gnd while layer 4 is signal layer. Via separation between P & N

is 75 mils. Via differential impedance is 48 ohms.

Regards

Vinod A H

On Wed, Jul 1, 2015 at 3:49 AM, David Banaswrote:

Hi Ken,

I assume you meant to address me and not Vinod.

When I used the term "mode dispersion", I was referring to the different

propagation velocities of the even and odd modes of a differential

micro-strip pair, which is the physical structure I'm assuming Vinod is

discussing.

For an excellent treatment on how this dispersion results in far end

cross-talk, please, refer to section 11.11 of Eric Bogatin's book, "Signal

Integrity - Simplified".

Thanks,

-db

On Tue, Jun 30, 2015 at 1:07 PM, Ken Cantrell

wrote:

Vinod,

There are only two dispersive modes for a Microstrip. The first is when

v_g

(group velocity), is less than the v_p (phase velocity), called normal

dispersion. The second is when v_g is greater than v_p, called anomalous

dispersion.

Under either scenario, I am not clear how this effects (enhances)

crosstalk.

Can you elaborate?

Thanks,

Ken

-----

Answered bycapn.freako
5 years 2 months 22 days

Hi Ken,

I assume you meant to address me and not Vinod.

When I used the term "mode dispersion", I was referring to the different

propagation velocities of the even and odd modes of a differential

micro-strip pair, which is the physical structure I'm assuming Vinod is

discussing.

For an excellent treatment on how this dispersion results in far end

cross-talk, please, refer to section 11.11 of Eric Bogatin's book, "Signal

Integrity - Simplified".

Thanks,

-db

On Tue, Jun 30, 2015 at 1:07 PM, Ken Cantrellwrote:

Vinod,

There are only two dispersive modes for a Microstrip. The first is when

v_g

(group velocity), is less than the v_p (phase velocity), called normal

dispersion. The second is when v_g is greater than v_p, called anomalous

dispersion.

Under either scenario, I am not clear how this effects (enhances)

crosstalk.

Can you elaborate?

Thanks,

Ken

-----

Answered bycapn.freako
5 years 2 months 22 days

If you’re not seeing it in the differential S-params. then it’s probably not

resonance, but rather cross-talk due to mode dispersion.

(I’m guessing you’re using micro-strips, as opposed to strip-lines; is that

correct?)

-db

On Jun 30, 2015, at 4:36 AM, vinod ahwrote:

Hi All,

I was extracting 4 port S parameters for differential signal on FR4 PCB

trace referenced to ground with connector and vias in the trace path.

S21 is having resonance at 10GHz where the S parameter file is single ended

version, while same S21 does not have resonance at 10GHz when the single

ended version is converted to mixed mode S-parameters using simulation

tools.

The resonance is expected due to discontinuity in the path but both P & N

of the differential signal is having same discontinuity, hence both P & N

are seeing same amount of insertion loss. Single ended impedance is 50 ohms

while differential impedance is 90 ohms.

So was wondering what is causing a differential signal not have resonance

compared to single ended signal leaving apart the mathematics used in

conversion to mixed mode :-)

Regards

Vinod A H