Istvan, Thank you very much for your thoughtful and detailed reply! While the original question did not include converters specifically (assumed low impedance external supplies or similar), your points about converter circuits and their output capacitance is well taken. I'll mention another frequently overlooked point is INPUT bypass capacitance for converters, often it is under represented or poorly placed. But switching supply design is another topic. Your point about plane - plane coupling in a GND - PWR - PWR - GND stackup is also well taken. The power planes will effectively reference each other to some degree, bypass placement should consider this. At a minimum the overlap arrangement should be well studied, and best simulated. One key item in my post was possible operational failure modes due to sub-optimal PDS design. It's worth noting while a circuit may apparently function exactly as intended, it is could still essentially operationally defective due to EMI issues. Thanks for bringing up EMI. My original post was specifically not about return paths, and the GND - SIG - SIG - GND stackup sets specified were to minimize this consideration. Of course we know to always keep this issue at high priority, in terms of plane splits, via stitching, bypass locations, etc. Thanks for bringing it up. Also my thanks goes out to all the other responders to this thread. I continue to enjoy the collective wisdom here. Ivor -----Original Message----- From: Istvan Novak [mailto:istvan.novak@xxxxxxxxxxx] Sent: Tuesday, April 01, 2008 6:03 AM To: Bowden, Ivor Cc: si-list@xxxxxxxxxxxxx Subject: Re: [SI-LIST] PDS analysis? Ivor, There have been lots of comments and discussion following your posting. I would like to offer a few additional comments going back to your core question: "Assuming the split power planes utilize sufficient area to keep the point to point inductance and resistance to reasonable values, and 0.1uF ceramic bypass caps are evenly placed at device pins, and bulk capacitance is placed as needed, would there be reason to expect any problems, such as plane resonance, etc?" - as always, it depends, but there are circumstances when such a simple network works well. If you use only one value, 0.1uF, ceramics, this falls into the "Big-V" type, and your best bet is to use similarly a single value of bulk capacitor. By doing so you eliminated the potential antiresonances among bulk capacitors (because there is only one value) and among ceramic capacitors (because there is only one value). Staying on the board, you are left with three potential problem areas: a) the DC source and bulk capacitor interface, b) bulk capacitor and ceramic capacitor interface and c) ceramic capacitor to PCB interface. For instance, if your impedance target is around 0.1 ohms (may be good for typical rails with a few amperes current consumption), you can use bulk capacitor(s) with 0.1 ohms ESR together with for instance twenty to thirty 0.1uF ceramics. With typical ESR and ESL values, this gives a nice interface between the bulk and ceramic capacitors. Assuming a small plane puddle, this should be OK. - the DC-source to bulk interface is an often overlooked problem area. Too much bulk capacitance (too many uF and/or too low ESR) may drive the converter loop into instability, or may create startup problems. Too little bulk capacitance may result in an antiresonance peaking above your target impedance at low frequencies. There are good simulation aids for converter chips, but hardly anything for complete DC-DC converter modules. If you are willing to take the time to simulate the state-averaged converter performance with your planned PDN, ask your converter vendors; though they do not offer these simulators publicly, some will make it available for you if you ask. - the ceramic capacitor and plane interface has a few potential issues: a) antiresonance between the static plane capacitance and inductances of bypass capacitors, b) modal resonances of planes and c) too much inductance presented to active devices. These risk areas are inter-related. For instance, using thick laminates makes solve a) easier but solve c) harder and vice versa. Regarding b), your best bet is to make component placement such that your plane-shape size is minimized. One additional comment about the stackup: having paralleled power stack in the middle between ground planes is good for isolating power splits from high-speed traces (no issues with traces crossing splits), but the vertical coupling between the plane shapes above each other is very strong at high frequencies: you need to make sure that you dont allow vertical overlap between very noisy and very sensitive (supposedly low-noise) rails. The above are considerations of the primary PDN function, delivering clean power to the chips. As Chris always points out, the return-path function always needs to be looked even if on your board traces do not reference power planes.. Our EMI-prevention goals are usually covered in the regular PDN design process by suppressing plane resonances. Regards, Istvan Novak SUN Microsystems Bowden, Ivor wrote: > Hi SI Experts, > > Say you have a typical PCB with modern technology mix of CPU, DSP, DDR, GIGE, > PCIE, etc. Say it is a multi-layer stackup in the form of GND-SIG-SIG-GND > sets, with the power distribution centered in the stackup as solid ground > plane - split power plane - split power plane - solid ground plane, using 1oz > copper and 3.5 mil dielectric. Assuming the split power planes utilize > sufficient area to keep the point to point inductance and resistance to > reasonable values, and 0.1uF ceramic bypass caps are evenly placed at device > pins, and bulk capacitance is placed as needed, would there be reason to > expect any problems, such as plane resonance, etc? If so, what would be the > observable real world manifestations, in terms of circuit performance and > power pins scope waveforms? Would there be significant advantage to analyzing > this PDS, or should following this "industry standard practice" for PCB PDS > be sufficient to expect robust behavior? > > Thanks, > > Ivor Bowden > Senior Hardware Engineer > Curtiss-Wright Controls Embedded Computing > 10201 Wateridge Circle > Suite 300 > San Diego, CA 92121 > 858-452-0020 x 4405 > ibowden@xxxxxxxxxxxxxxxxx > _______________________________________________________________________ This e-mail and any files transmitted with it are proprietary and intended solely for the use of the individual or entity to whom they are addressed. If you have reason to believe that you have received this e-mail in error, please notify the sender and destroy this email and any attached files. Please note that any views or opinions presented in this e-mail are solely those of the author and do not necessarily represent those of the Curtiss-Wright Corporation or any of its subsidiaries. Documents attached hereto may contain technology subject to government export regulations. Recipient is solely responsible for ensuring that any re-export, transfer or disclosure of this information is in accordance with applicable government export regulations. The recipient should check this e-mail and any attachments for the presence of viruses. Curtiss-Wright Corporation and its subsidiaries accept no liability for any damage caused by any virus transmitted by this e-mail. ------------------------------------------------------------------ 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