[SI-LIST] Re: s parameters and transient simulation

  • From: Steve Corey <steven.corey@xxxxxxxxxxxxxx>
  • To: "'si-list@xxxxxxxxxxxxx'" <si-list@xxxxxxxxxxxxx>
  • Date: Thu, 11 Dec 2003 09:58:10 -0800

Geoff -- your statements are true about network theory.  However, I 
don't see where common ground nodes become a "bad recipe".

When you take a measurement, you connect a reference conductor and a 
signal conductor from your measurement system to the device.  By 
connecting the reference conductor to a specific point, you're stating 
that you don't care what the voltage is at that point -- you're only 
interested in the difference between the reference conductor and the 
signal conductor.  If you're taking two-port measurements, you can't 
truly know how the voltages at the two signal nodes are varying with 
respect to one another, since you don't know how the voltages at the two 
ground reference nodes are varying with respect to each other.

When you connect a receiver, be it digital or analog, to a port, you are 
roughly making the same statement.  You don't care what any of the 
voltages is with respect to some global ground, or some faraway port, 
you care about the difference between the receiver circuit's nodes and 
the local reference.   All these voltages are local -- theoretically, 
the power rail, vin+, and vin- at the receiver could all be measured 
against a single local "ground", and those are the same voltage 
differences the device will see.

When you drop a multiport network into SPICE, you have to be aware that 
you're making the same assumptions.  Your simulation needs to be set up 
-- the same way your circuit design is -- so that the behavior doesn't 
depend on the difference between reference voltages of distant ports. 
This is the same partitioning into "groups" that is outlined in the MTT 
paper you referenced.  The paper does little more than to cast the logic 
outlined above into modified nodal analysis (MNA) matrix notation which 
casts all local references to zero volts by zeroing out their 
rows/columns in the MNA matrix.

The bottom line is, if you can't know the voltage between two faraway 
points, you need to make sure that you don't care what it is -- by good 
design, good measurement setup, good simulation setup.  It should be 
reassuring that they all have to comply with the same set of limitations.

   -- Steve

-------------------------------------------
Steven D. Corey, Ph.D.
Time Domain Analysis Systems, Inc.
"The Interconnect Analysis Company."
http://www.tdasystems.com

email: steven.corey@xxxxxxxxxxxxxx
phone: (503) 246-2272
fax:   (503) 246-2282
-------------------------------------------


Geoff Stokes wrote:
> Hi Ray
>  
> With reference to your posting earlier this year regarding n-ports etc.,
> here is a thought on simulation of interconnects at high frequencies where
> the concept of common voltage reference nodes seems to become a bad recipe,
> thinking particularly of RF modelling of IC packages.
>  
> As Khalil and Steer (paper cited below) have pointed out, the voltage
> between two points is undefined in general.  This is an aspect of field
> theory which becomes relevant when the frequency is high enough that the
> phase delay between two points in a structure is a significant proportion of
> the wavelength.  The significant proportion of course depends upon the
> application, so we can't define a specific threshold frequency even for a
> specific mechanical dimension.  In analog or mixed-mode circuit designs,
> relatively small values of couplings or impedance may be significant, but
> such values might be ignored in a purely digital circuit.  In addition, for
> a correct DC simulation of the operating point and power supply currents,
> together with broad band accuracy, the effect of internal inductance and
> frequency dependent resistance 
> (both arising from skin effect and providing several percent effects) will
> need to be included.
>  
> In an earlier posting, Ege Engin wrote this helpful comment:
>  
> "If an S parameter matrix is implemented in a circuit simulator, it
> actually divides the rest of the circuit (all the other linear and
> non-linear elements) into groups, that are only coupled to each other by
> means of this S parameter matrix (due to the fact that an S parameter
> matrix represents a distributed circuit). Since the voltage drops
> between the local reference nodes in various groups are undefined, they
> can be connected to each other in an arbitrary manner."
>  
> I would just add that in practice, from the s-parameters obtained by
> electromagnetic simulation or measurement, we have to formulate a polynomial
> or lumped-element solution to feed into the nodal transient circuit
> simulator (Spice or Spectre).  Ege Engin's final sentence would then apply
> to the interconnection of the extracted n-port model with the chip
> schematic.
> In Khalil and Steer, "Circuit Theory for Spatially Distributed Microwave
> Circuits" (IEEE Transactions on Microwave Theory and Techniques, Vol. 46 No.
> 10, October 1998), we find:
> 
> "The essence of the problem is that a global reference node cannot
> reasonably be defined for two spatially separated nodes when the
> electromagnetic field is transient or alternating.  In this situation, the
> electric field is nonconservative and the voltage between any two points is
> dependent on the path of integration and, hence, voltage is undefined.  This
> includes the situation of two separated points on an ideal conductor."
> 
> So we see that each port requires its own separate local return pin in order
> to describe the distributed structure with sufficient accuracy over the
> required frequency range.  Two or more ports can only use a common ground if
> they are physically close enough to one another (for the specific case).
> 
> Finally, we make the arbitrary (?) decision to join the local ground(s) to
> the common ground and hope it's OK.  From the network theory it seems OK,
> but is a little hard to swallow.
> 
> Any comments?
> 
> Best wishes,
> 
> Geoff
> 
> ______________________________________________ 
> 
> Geoff Stokes
> Applications Engineer, Signal Management Group
> 
> Zetex plc 
> Lansdowne Road, Chadderton, Oldham, OL9 9TY,  UK 
> Tel direct:  +44-161-622-4857   Switchboard: +44-161-622-4444
> Fax:  +44-161-622-4469 
> http://www.zetex.com <http://www.zetex.com/>  
> e-mail:  gstokes@xxxxxxxxx 
> 
>  
> 
> 
> _________________________________________________________
> 
> Zetex Semiconductors - Solutions for an analog world
> 
> EID Award Winners for  'Best Use of Technology' 2003 for the 
> AcoustarTM ZXCW8100 End-to-End Digital Audio Amplifier Controller
> 
> http://www.zetex.com
> _________________________________________________________
> 
> ######################################################################
> E-MAILS are susceptible to interference. You should not assume that
> the contents originated from the sender or the Zetex Group or that they 
> have been accurately reproduced from their original form.
> Zetex accepts no responsibility for information, errors or omissions in
> this e-mail nor for its use or misuse nor for any act committed or
> omitted in connection with this communication.
> If in doubt, please verify the authenticity with the sender.
> ######################################################################
> 
> ------------------------------------------------------------------
> 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.org
> 
> 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.org

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: