[SI-LIST] Re: Decoupling capacitors

Ross - You are quite correct in identifying the need for chip
capacitance to be accounted for when doing a power distribution
analysis.  Not only is the chip capacitance important, but the package
inductance is also extremely important.  These reactances form a
resonant tank circuit that often dominates the power impedance seen by
the chip circuits in the 50MHz to 200MHz band.  We call this
chip/package resonance.  Many packaged chips also include discrete
decoupling capacitors somewhere between the chip and the PCB.  The
chip-capacitance, package-inductance and on-package capacitors are
extremely important components that must be accounted for when doing
power distribution system analysis.

The Cadence SQPI tool is excellent for designing a broad band low
impedance at the PCB level.  But to make the tool really useful, we
must include the effects of chip/package resonance.  I know that
Cadence has this enhancement on their roadmap.  Perhaps a little more
pressure from SI-list will help motivate them to put the resource on
it.  :)  I have seen some scripts from Intel that essentially embed the
SQPI noise source inside a package.  It would be nice to get a fully
supported version from Cadence.

regards,
Larry Smith
Sun Microsystems

> Delivered-To: si-list@xxxxxxxxxxxxx
> From: Ross Jatou <RJatou@xxxxxxxxxx>
> To: "'raymond.anderson@xxxxxxx'" <raymond.anderson@xxxxxxx>, si-list 
<si-list@xxxxxxxxxxxxx>
> Subject: [SI-LIST] Re: Decoupling capacitors
> Date: Thu, 23 May 2002 11:54:49 -0700
> MIME-Version: 1.0
> Content-Transfer-Encoding: 8bit
> X-archive-position: 2948
> X-ecartis-version: Ecartis v1.0.0
> X-original-sender: RJatou@xxxxxxxxxx
> X-list: si-list
> 
> 
> Ray,
> 
> We have SQPI as well and have been trying to correlate it to measurements.
> My question is: how do you incorporate chip capacitance ?
> The capacitance is in two forms:
> 1) package and die capacitance between power and ground
> 2) integrated caps that are placed in the package.
> 
> 
> 
> Ross
> 
> -----Original Message-----
> From: Ray Anderson [mailto:raymond.anderson@xxxxxxx]
> Sent: Thursday, May 23, 2002 11:08 AM
> To: si-list
> Subject: [SI-LIST] Re: Decoupling capacitors
> 
> 
> 
> 
> 
> 
> > Specctraquest has a power integrity option which allows the simulation of
> > parallel resonaces of capacitor decoupling, which also accounts for
> > propagation delay between the capacitors.  It uses a multinode mesh of
> > transmission lines to model the decoupling capacitor system.  I haven't
> had
> > a chance to use it, but it sounds like a good strategy for modelling
> > decoupling capacitors as a complete system, taking into account capacitor
> > locations as well as capacitor values.  Has anybody in this group had a
> > chance to use this?
> > 
> > Wayne Cooke
> > Signal Integrity Engineer, Innovance Networks
> > 19 Fairmont Ave., Ottawa, Ont.  K1Y 1X4
> > email: wcooke@xxxxxxxxxxxxx
> 
> 
> Disclaimer: I am one of the original developers of the power integrity
> analysis tool that was licensed to Cadence and is now integrated into 
> Specctraquest, so my views may be biased....
> 
> 
> As mentioned by Wayne, the SpecctraQuest Power Integrity tool (SQPI)
> is based on a plane model comprised of an orthoganal mesh of transmission
> lines where various decoupling capacitors can be placed at any of the 
> mesh nodes of the plane model for performing a frequency domain analysis.
> 
> There are actually two modes of analysis in the tool: single-node and
> multi-node. 
> 
> The single-node analysis places all the decoupling caps in
> parallel, connected to just a single pair of nodes, (the hot side and
> ground),
> with no plane circuit in the model. This mode of analysis can be utilized
> as an initial step in the selection of a set of decaps that provides a
> broadband low Z response. It is quick but not particularly accurate as
> it doesn't account for the spatial placement on the planes.
> 
> The multi-node analysis adds the plane model into the mix. In this
> scenario, the decaps as well as a noise source(s) and a VRM are spatially
> placed on the plane. Then a frequency domain simulation is run and the
> resultant impedance profile is generated. Note that the inductance
> associated
> with the mounting structures (pads, escapes, and planes) can be specified
> in addition to the 'intrinsic' inductance of the capacitor which is included
> in the decap model.
> 
> When you place decaps on the plane and perform the frequency domain
> simulation,
> there are three types of resonances which you will observe and must be
> managed to produce a properly decoupled power distribution system. These are
> the a) plane resonances (actually cavity resonances), b) series RLC
> resonances
> where each decap resonates, and c) anti-resonances caused by the inductance
> of the decap resonating with the interplane capacitance of the planes.
> 
> For more details on the methodology, go to
> http://groups.yahoo.com/group/si-list/files/
> and follow the links to a number of published conference papers dealing with
> the subject of power distribution system design and simulation authored by
> our group here at Sun.
> 
> -Ray Anderson
> Sun Microsystems
> ------------------------------------------------------------------
> 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:
> http://www.freelists.org/webpage/si-list
> 
> For help:
> si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
> 
> List archives are viewable at:     
>               http://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:
> http://www.freelists.org/webpage/si-list
> 
> For help:
> si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
> 
> List archives are viewable at:     
>               http://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:
http://www.freelists.org/webpage/si-list

For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field

List archives are viewable at:     
                http://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: