[SI-LIST] Re: Understanding S-Parameters
- From: "Eric Bogatin" <eric@xxxxxxxxxxxxxxx>
- To: <wilson@xxxxxxxxxxxxxxxxxx>, "'si-list'" <si-list@xxxxxxxxxxxxx>
- Date: Wed, 2 May 2007 11:31:27 -0500
Ralph-
You may gain a little insight in single ended and differential via design by
looking at the talk I am giving at the Agilent ADS roadshow, on "Designing
transparent vias". I cover both single ended and differential via design.
The schedule for the tour is posted at:
http://www.home.agilent.com/agilent/eventDetail.jspx?cc=US
<http://www.home.agilent.com/agilent/eventDetail.jspx?cc=US&lc=eng&ckey=8737
87&nid=-536900532.536905430.08&id=873787>
&lc=eng&ckey=873787&nid=-536900532.536905430.08&id=873787
If this link doesn't go through, there is another link posted on my web
site, www.bethesignal.com <http://www.bethesignal.com/> , under the "where's
eric" tab at the top of the splash page.
I also posted a copy of the handouts for this talk on my web site under the
features articles tab.
The bottom line is that to make a transparent differential via, you want to
make the return loss, SDD11, as large a negative dB value as you can, and
make the insertion loss, SDD21 as close to 0 dB as you can.
As a quick estimate, and to provide guidance on which knobs to tweak, and in
which direction, you can fit the via to an ideal differential pair model,
with a diff stub attached, as needed, to extract the differential impedance
and the degree of coupling.
In looking at a number of stack up configurations, I typically find that if
nothing special is done in the design, the coupling, Z21, is small and the
diagonal element, Z11, is very low. To match the via to a 100 ohm diff
impedance usually requires increasing the single ended impedance by such
things as removing the NFPs and increasing the antipads clearance holes. As
I show in my talk, since there still is considerable coupling, you can't
just take two 50 Ohms single ended vias and have them be optimized for 100
Ohms differential vias. You have to include the impact for the coupling.
Feel free to drop me a note off line if you have any questions about the
method I used. I hope to have this talk posted as an online lecture in the
next month.
--eric
**************************************
Dr. Eric Bogatin, President
Bogatin Enterprises, LLC
Setting the Standard for Signal Integrity Training
26235 w 110th terr
Olathe, KS 66061
v: 913-393-1305
f: 913-393-0929
c:913-424-4333
e:eric@xxxxxxxxxxxxxxx
www.BeTheSignal.com
****************************************
-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] On
Behalf Of Ralph A Wilson III
Sent: Wednesday, May 02, 2007 8:02 AM
To: si-list
Subject: [SI-LIST] Understanding S-Parameters
The question was asked, "which of these differential
vias is better?" Ignoring all the philosophical issues
of the meaning "better", I put together a simple 3D-model
of each of the vias in Microwave Studio and took a
look at S11 and S21 (among other things). I have
several questions regarding the setup of the simulation
and the interpretation of the results:
1) Is a differential via (pair) a two-port or four-port
model? The model I put together has two vias, with
pads/connections (a "P" and an "N" pair) on the top layer
and on an inner layer. The model includes a number of
reference planes tied together with additional "ground
vias". In modeling this, do I attach a single port
between the "P" and "N" leads at the micro-strip, and
a second port between the "P" and "N" striplines (hence
a two-port model)? Or, should I attach a port from
a reference plane to each "P", and from a reference
plane to each "N", hence having a four-port model?
Exciting the two-port model in MWS is straight
forward - I haven't figured out how to stimulate the
four port simulation to give me "differential"
operation - any hints?
2) Assuming I get #1 figured out, do I want to normalize
the resulting S-parameter to 50 Ohms? If a 4-port model
is the right one, I'm assuming normalizing to 50 Ohms is
the "right thing to do". However, if the answer to #1 is
a 2-port model, shouldn't the normalization be done to
100 Ohms?
3) Finally, regarding S11 - I'm assuming I can use this to
determine how "transparent" my vias are - that is, how
close each via is to matching the characteristic
impedance of the rest of my transmission line
(normalized to 50 or 100 Ohms). True or not true?
Can I use S11 to calculate the "Z0" of the via?
I'm intuitively drawn to wanting to equate S11 to the
reflection coefficient, and therefore use that to calculate
the Z0 of the via (given that S11 is normalized to 50 or
100 Ohms). However, since S11 is always positive,
this implies I could never have a via whose impedance is
less than that of the transmission line - and I don't
believe that. Or, do I need to look at the phase information
as well as S11 to determine positive vs. negative reflection?
Thanks in advance for the help!
Ralph Wilson
Alcatel-Lucent
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
or to administer your membership from a web page, go to:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List technical documents are available at:
http://www.si-list.net
List archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
or to administer your membership from a web page, go to:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List technical documents are available at:
http://www.si-list.net
List archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
