Tom, I see Scott has already answered this. In order to avoid turning every design into a college thesis, the best that I can recommend is that you rigidly adhere to a fixed mount / via design for your capacitors. The next problem to overcome is variation in the distance from the capacitor mounting surface to the first RF plane in the PCB. As you will find in our papers depending on just how accurate you need to be, you can zero this out provided that it isn't more than about 15 mils into the board. The remaining problems will be those of variation from mfg to mfg, and cover layer which tends to grow for a particular capacitor value over time. What happens is that as the MLCC manufacturers get better and better at making thinner and thinner plates, they use fewer plates for a given capacitor value. They build the plates out from the middle, so the cover layer gets thicker over time increasing mounted inductance. Today's 2uF capacitor that fully packs a capacitor to within a couple of mils of the outer surface will become next year's capacitor with a cover layer nearly twice as thick. If you are doing everything right: power cavity close to the surface, good capacitor mount and via design, etc then the cover layer variations can have a visible impact. If you are sloppy with things like the power cavity far from the capacitor surface, then via inductance will swamp out any cover layer impact. You can get around the cover layer variation problem by remaining conservative, which will also help your pocketbook. If you are going to use a big Vish kind of design with only one or a few capacitor values, instead of using the largest capacitor in a given body size as your primary value, begin two capacitor sizes down. MLCC capacitor manufacturers charge for Farads when the capacitors are relatively fully packed. This has a strong influence on price for only the largest and second largest values in a particular case size. Below that, you are paying mostly for the shipping ( no joke! ). You can still use the biggest values, just use them as a small proportion of your network, and model them as though they have a thick cover layer. Because next year they will. This will save you money, and BTW it will keep your ESR from getting unnecessarily low which can help with resonance management. See various papers by Larry Smith on that. If you are designing a flat array ala Larry's FDTIM, then you will naturally only use a couple capacitors that fill out their cases anyway. Once you adopt these procedures, you can reasonably design your network using fixed models and pretty much any of the tools out there from: Ansoft, Sigrity, Cadence, etc and get reasonably good correlation from simulation and what your design really does. Further, you won't have to worry about constantly tracking source control and manufacturer variations in their capacitors. The really big issue in PDN design is getting accurate information from your IC suppliers: What current spectrum does your PDN need to support, and what parasitics are buried inside the IC package. If you don't have that information, then you'll end up with a PDN design that either costs much more than it should, or that doesn't perform as needed, or both. Steve. X2Y measured several capacitor package configurations with very accurate fixtures that we developed for that purpose. You can find the results on the X2Y website in the decoupling section. One of the papers has a summary chart that shows mounted inductance versus cavity depth. tom_cip_11551 wrote: > Hi, > > I am trying to determine an optimum decoupling strategy. I have found > lots of manufacturers data conerning capacitor models, from Murata and > others. But I keep thinking that these models can't possibly take into > consideration the layout footprint, because of process, material and > other variations, like the layer height above copper. > > So, when attempting to model a capacitor in spice or other, what is a > good strategy in trying to take into consideration the footprint itself? > > Yes, I do realize that in some cases, a 3D model can be created, but > that can be very time consuming and I am really only talking about a > first order spice approximation. > > > Thank you to all who respond. > > Tom > > > ------------------------------------------------------------------ > 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 > > > > -- Steve Weir Teraspeed Consulting Group LLC 121 North River Drive Narragansett, RI 02882 California office (866) 675-4630 Business (707) 780-1951 Fax Main office (401) 284-1827 Business (401) 284-1840 Fax Oregon office (503) 430-1065 Business (503) 430-1285 Fax http://www.teraspeed.com This e-mail contains proprietary and confidential intellectual property of Teraspeed Consulting Group LLC ------------------------------------------------------------------------------------------------------ Teraspeed(R) is the registered service mark of Teraspeed Consulting Group LLC ------------------------------------------------------------------ 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