[SI-LIST] Re: Choosing extraction frequency for package RLGC
- From: Geoff Stokes <gstokes@xxxxxxxxx>
- To: si-list@xxxxxxxxxxxxx
- Date: Wed, 7 Jan 2004 09:22:37 -0000
The answer to this might be obvious, but l'll tell you what I found.
Whatever the choice of extraction frequency, if you do an AC analysis of the
extracted model, it will not match the original s-parameters exactly over a
frequency range of interest. The frequency should be based on band limits
and Ata's good advice seems to address this. Often pulse response is the
target and the frequency range relates to the bulk of the power spectral
density (Fourier transform). Otherwise specific frequency performance is
required. So there will be a compromise. You may need to extract for a few
different frequencies and choose the best fit over the frequency range.
An alternative, as discussed before, depending on your software and
application, is to feed s-parameters directly into the Cadence n-port
element. That would probably give a better fit because it can to some
extent include effects of delayed couplings. A simple series of RLCG
networks does not do that, even if M's are added.
The major difficulty seems to be getting the s-parameters to begin with.
You can't measure at internal connections of the package because probes will
disturb the fields. For direct measurements, special test chips need to be
constructed and also some assumptions are made about impedances of the chip:
if characterised by CPW wafer probing, that situation requires analysis also
to extract a chip model. However it's quicker to use a full-wave 3D
simulator, but you need to do some testing to confirm the method.
> -----Original Message-----
> From: Zadehgol, Ata [mailto:ata.zadehgol@xxxxxxxxx]
> Sent: 03 January 2004 05:09
> To: si-list@xxxxxxxxxxxxx
> Cc: Zadehgol, Ata
> Subject: [SI-LIST] Re: Choosing extraction frequency for package RLGC
>
>
> Hi Frank,
>
> That is a good question, personally the approach I'd like to take in
> choosing the bandwidth for system simulations, model
> generation, and lab
> data collection is to consider three main factors, these are
> as follows:
> 1. The power spectral density of the transmitter device.
> 2. The Transmit and Receive compliance requirements such as
> slew rate,
> voltage level, etc..
> 3. The protocol spec such as trise10%-90% vs. trise20%-80%, etc...
> It is important to keep in mind the derivation of the 0.35/trise found
> in Hspice manual or some other books as a quick
> approximation, which is
> basically based on a 1st order approximation of passive
> system response
> and 10%-90% assumptions; however, this may not necessarily
> apply exactly
> to every design/optimization problem. I personally prefer to
> take a few
> moments up front and hand calculate the bandwidth
> requirements based on
> the three items I outlined above, this will not only give the
> customized
> bandwidth requirements for the specific design, but will also help
> define the optimization goals more intuitively as well. Hope this
> helps.
>
> Ata Zadehgol
>
> -----Original Message-----
> From: si-list-bounce@xxxxxxxxxxxxx
> [mailto:si-list-bounce@xxxxxxxxxxxxx]
> On
> Behalf Of Frank Dunlap
> Sent: Friday, January 02, 2004 1:08 PM
> To: si-list@xxxxxxxxxxxxx
> Subject: [SI-LIST] Choosing extraction frequency for package RLGC
>
>
> I apologize if my question has been answered in previous postings. I
> searched the archives, yet did not find an answer.
>
> From my understanding, an "extraction" frequency must be specified for
> tools
> that generate RLGC from package data. My question is, "how does one
> choose
> an appropriate frequency?"
>
> Note the excerpt below and take note of the implications of the last
> sentence:
>
> ******************************************
> Pages_5-21_5-22_from hspice_signal_integrity_guide.pdf
>
> Some designers use 0.35/trise to estimate the skin effect frequency.
> This
> estimate is good for the bandwidth occupied by a transient,
> but not for
> the
> clock frequency, at which most of the energy is transferred.
> In fact, a
> frequency of 0.35/trise is far too high and results in excessive loss
> for
> almost all applications. Simulation computes the skin effect frequency
> from
> 1/(15*trise). If you use precomputed model parameters (ELEV =3D3D 2),
> compute
> the resistance matrix at the skin effect frequency.
>
> ******************************************
>
> I've heard various opinions regarding frequency selection,
> yet wanted to
> gather more opinions from you SI folks.
>
> One other question: when working with a limited budget while dealing
> with a
> package that supports multiple interfaces that operate at different
> frequencies, are there recommendations as to how to determine the best
> frequency for RLGC extraction that yields reasonable results for all
> interfaces? Given only one frequency to use for the entire RLGC
> extraction,
> my choice would be to use an extraction frequency based upon the
> requirements of the highest-frequency interface.
>
> Thanks,
>
> Frank
>
> Some more search terms to aid others in finding this e-mail later:
>
> RLC LCR RLGC=3D20
> ------------------------------------------------------------------
> 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: =20
> //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
> =20
>
> ------------------------------------------------------------------
> 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
>
>
_________________________________________________________
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
Other related posts:
