Lee, spreadsheets that include the plane=20 capacitance as a lumped branch do simulate=20 accurately for cases where the bypass network=20 does not push the PRF up into the half lambda=20 region of the unloaded board. I've seen a number=20 of these tools and built them as well. What=20 spreadsheet models I have seen do not do: a. They don't create accurate mounted capacitor=20 models. Garbage in =3D> Garbage out. b. They only account for what those glorious ICs=20 do if the IC parasitics are included. But this=20 is also true for any other tools. c. They do not in general account for the=20 distributed behavior of the planes including=20 loaded modal resonances. Uncorrected, 1D branch=20 models tell interesting lies in the phase plane. Used with reason they do make good first cut=20 estimates. For anything serious, I use=20 distributed PEEC models that with accurate=20 capacitor and IC models correlate extremely well=20 to lab measurements. There are a number of=20 papers available some that date back 11 years on=20 that methodology. More recently Todd Hubing=20 wrote a paper on reducing the 2D representation to an equivalent 1D form. Regards, Steve. At 12:50 PM 5/23/2006, Lee Ritchey wrote: >I agree that the spreadsheet method has same hazards. Namely failing to >show resonances between the plane capacitor and the parasitic inductance of >the discrete capacitors. It's good for a first order look at things. A >more precise method is to use a SPICE model. That will account for these >resonances, but it won't account plane resonances. The Ansoft and Sigrity >tools will. Perhaps there are other tools that will as well. > >It is possible to achieve a reasonable impedance vs. frequency with about 5 >values of capacitor and avoid any serious resonances. I've got many dozens >of measured results that show this. There is good correlation between >predicted and actual with the simulators named above and reasonable >correlation with the spreadsheet method. > > > > [Original Message] > > From: <Alan.Hiltonnickel@xxxxxxx> > > To: <JeanPierrePoulin@xxxxxxxxxxx> > > Cc: <si-list@xxxxxxxxxxxxx> > > Date: 5/23/2006 12:27:08 PM > > Subject: [SI-LIST] Re: Capacitors arrays=85worth it or not? > > > > Hi Jean-Pierre, > > I've done spreadsheet analysis before, and it's a useful first > > approximation. The problem is that it is difficult to take resonances > > into account. > > > > I have not yet read Lee's book, and I'm not sure where he stands on > > capacitor selection technique (although I think it's safe to say he's in > > favor of it). There seem to be two schools of thought, both of which I > > have found useful. You can check the SI-List archives for more details. > > I'm sure I won't do justice to either method here. > > > > Under the "Big V" approach (championed most notably and ably by Steve > > Weir) you use the largest value capacitors you can find in any > > particular size and add them until you reach your target plane impedance > > over the frequencies of interest. Although I have used this > > successfully, you have to watch out for resonant peaks in the impedance > > profile caused by the inductance of a larger cap interacting with the > > capacitance of the next smaller cap. > > > > The distributed approach requires choosing multiple values in each size > > to create a flat (or nearly so) impedance profile. This approach was > > developed by Ray Anderson, Larry Smith, and Tanmoy Roy. The problem I > > find with this approach is that the end result can be an impractically > > large number of capacitors. > > > > Ultimately you need to simulate and measure. There are several good > > simulation tools available. Naming a few off my head, Cadence has a > > decent Power Integrity tool, as do SiSoft, Mentor, Ansoft and Sigrity > > (This is not a comprehensive list. Google: "Power Integrity"). For any > > of these tools to work well, you'll need good models. Once you've built > > the board, use a VNA to evaluate your power impedance. The results of > > that should inform your decisions on capacitor changes or design= changes. > > > > I won't go into placement and connection methods here, I'm sure you're > > aware of how crucial it is to keep connection inductance from dominating > > capacitor package impedance. Steve's comments elsewhere in this thread > > have me wanting to revisit the array and X2Y caps to see if I'm missing > > something. > > > > Alan > > > > jeanpierrepoulin wrote On 05/21/06 10:05,: > > > > >Thanks Alan for the recommendations on a formal analysis... > > > > > >It did seem - given the sub 1ns rise time - that this was required > > >even if the two parts are right close to one another... so I've been > > >trying to improve on my previous approach (i.e. follow ref design > > >and 'sprinkle a few more caps just in case' methodology) to a more > > >scientific approach... > > > > > >Reading on the subject to improve the understanding and reach some > > >approximation to science, I have thus far adopted Lee Ritchey's > > >approach, added some theory from Dr. Howard Johnson and verified the > > >whole thing with an old Micron app note... > > > > > >However, the 'excel spreadsheet' that resulted is surely not optimal > > >and could induce us in some errors... > > > > > >Are there any tools / methodology you have seen in your travels that > > >can assist with capacitor selection for high-speed designs such as > > >ours? > > > > > >Many thanks! > > > > > > Jean-Pierre > > > > > >--- In si-list@xxxxxxxxxxxxxxx, Alan Hilton-Nickel > > ><Alan.Hiltonnickel@...> wrote: > > > > > > > > >>Jean-Pierre, > > >> > > >>You have some analysis to do, both electrical and mechanical. > > >> > > >>Keep in mind that as the form factor of the capacitor gets > > >> > > >> > > >smaller, the > > > > > > > > >>interconnect inductance becomes more dominant. This means that the > > >>lowest-inductance caps should generally be closest to the point of > > >>consumption (the chip). This often means underneath the chip. The > > >> > > >> > > >arrays > > > > > > > > >>and the odd form-factor caps may simply not be placeable without > > >>eliminating power or ground vias, which at some point defeats the > > >>purpose of using low-inductance caps since it increases the > > >> > > >> > > >inductance > > > > > > > > >>of the power-plane connection. > > >> > > >>I've often seen the arrays and X2Y/LICCs used most effectively on > > >> > > >> > > >the > > > > > > > > >>package, not the board. The higher cost of the parts is a key > > >> > > >> > > >factor as > > > > > > > > >>well. > > >> > > >>Good luck with this. > > >> > > >>Alan > > >> > > >> > > > > > > > > > > > > > > > > > >------------------------------------------------------------------ > > >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 FAQ wiki page is located at: > > > http://si-list.org/wiki/wiki.pl?Si-List_FAQ > > > > > >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 > > > > > > > > > > > > > > > > -- > > Alan Hilton-Nickel > > Signal Integrity Engineer > > Sun Microsystems Inc. > > Netra Systems and Networking > > Newark, CA > > > > > > > > ------------------------------------------------------------------ > > 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 FAQ wiki page is located at: > > http://si-list.org/wiki/wiki.pl?Si-List_FAQ > > > > 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 FAQ wiki page is located at: > http://si-list.org/wiki/wiki.pl?Si-List_FAQ > >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 FAQ wiki page is located at: http://si-list.org/wiki/wiki.pl?Si-List_FAQ 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