Thank you all for the responses. I agree with Jeff that the only practical option so far is the cross-sectioning (it does not matter who does it). Also, it looks like Gary Brist's works is the only source of data that is usable to develop proper models that account for the distribution of geometry parameters. Though, such data are not readily available from the manufacturers. 3D CT scan is probably another option, but I have no idea how accessible such equipment and how much it would cost. There was a good paper on that subject by J. Hillebrand et al from Stuttgart University at DesignCon 2011. Just a few words to clarify my position. As the software vendors we are usually involved into the analysis to measurement validation only at the very last stage of the project - when everything is already designed, manufactured and measured (an example of such project is the "Lessons learned..." paper #2014_01 at http://www.simberian.com/AppNotes.php). Thus, all advices on working with the manufacturer at early stages are good, but not applicable. Usually, our customers put a few test structures on the board to do formal broadband dielectric and conductor roughness model identification with GMS-parameters. But even accurate material model identification up to 50 GHz already requires accurate cross-section geometry or high accuracy in manufacturing. It means that the board has to be cut into pieces or reliable geometry data available after the manufacturing to achieve reasonable correlation at 10-50 GHz frequency range (may be not so important at lower frequencies). Just changing geometry to fit the measured data (S-parameters or impedance) is very ambiguous technique - there are always multiple choices of what can be changed. For instance, if measured and modeled GMS-parameters are matched, but TDR impedances are off, changes in trace width and shape or spacing or dielectric thickness may be used to adjust geometry to match the impedance or reflection parameters. But if we change wrong parameter, it will work for only one structure and not for the others. Just tuning models for each structure is not an option. I think there is clear need for systematic approach to the analysis to measurement validation with the numbers from manufacturers. Best regards, Yuriy Yuriy Shlepnev, Ph.D. President, Simberian Inc. 3030 S Torrey Pines Dr. Las Vegas, NV 89146, USA Office +1-702-876-2882; Fax +1-702-482-7903 Cell +1-206-409-2368; Virtual +1-408-627-7706 Skype: shlepnev www.simberian.com Simbeor – Accurate, Fast, Easy and Affordable Electromagnetic Signal Integrity Software 2010 and 2011 DesignVision Award Winner -----Original Message----- From: Loyer, Jeff [mailto:jeff.loyer@xxxxxxxxx] Sent: Tuesday, September 9, 2014 7:39 AM To: shlepnev@xxxxxxxxxxxxx; si-list@xxxxxxxxxxxxx Subject: RE: [SI-LIST] Board design adjustments by manufacturers Hello Yuriy, The only data I would trust is from cross-sections of the actual trace I was trying to characterize, although I would argue that the identical trace from a different board (as long as it is from the same location on the panel) is going to have approximately the same geometries. And, the geometry of the trace might vary dramatically along the length of the trace. For instance, prepreg thickness will vary depending on local copper density. Variation is going to be very vendor-specific, so I don't think you'll ever be able to apply an assumed distribution to a particular trace. For instance, RTF etches better than standard copper (due to the nature of the grain structures), thus you would have to comprehend that variable in your assumption. For these reasons (and more), I've been very grateful for access to cross-sectioning ability; I don't know of an adequate substitute. Hint to entrepreneurs... If someone could come up with a compact, clean, cheap cross-sectioning tool, there might be significant interest. If you already have that available, let me know (and no, I don't mean a hacksaw and emery cloth :-)). If anyone has published data on the distribution of trace dimensions, it would probably be Gary Brist. A quick search revealed "Design Optimization of Single-Ended and Differential Impedance PCB Transmission Lines" which has some nice distribution data on slides 23, 24 , etc. He might have published more nuggets. Best wishes, Jeff Loyer -----Original Message----- From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] On Behalf Of Yuriy Shlepnev Sent: Monday, September 08, 2014 1:49 PM To: si-list@xxxxxxxxxxxxx Subject: [SI-LIST] Board design adjustments by manufacturers Hello Everyone, After going through a multiple analysis to measurement validation projects, I realized that almost all boards are not manufactured as designed (what a surprise :-)). As a consequence the analysis to measurement validation may require investigation of the manufactured board geometry, that includes cutting the board and taking measurements on the cross-sections (usually important only above 10 GHz). Now I am trying to figure out what board design adjustments are needed and where to get those numbers without cross-sectioning for the accurate post-layout electromagnetic analysis. In particular, is there any way to get the following numbers without cutting the board? - change thickness of layers (account for plating and variations in thickness of laminates); - change trace width and shape (account for etching and plating); - change spacing between traces in diff pairs (if adjusted for the "impedance control"); How board manufacturers actually adjust trace widths or spacing for a given design in the "impedance controlled" process if they have just Gerber or ODB++ data? Is there any software that does it? A relevant question - are there board manufacturers out there who provide probability distribution data for stackup layer thicknesses, strip widths, back-drilling depth etc.? Can anybody point me to a source of information on this subject. Best regards, Yuriy Yuriy Shlepnev, Ph.D. President, Simberian Inc. 3030 S Torrey Pines Dr. Las Vegas, NV 89146, USA Office +1-702-876-2882; Fax +1-702-482-7903 Cell +1-206-409-2368; Virtual +1-408-627-7706 Skype: shlepnev www.simberian.com Simbeor – Accurate, Fast, Easy and Affordable Electromagnetic Signal Integrity Software 2010 and 2011 DesignVision Award Winner ------------------------------------------------------------------ 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 forum is accessible at: http://tech.groups.yahoo.com/group/si-list List archives are viewable at: //www.freelists.org/archives/si-list 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 forum is accessible at: http://tech.groups.yahoo.com/group/si-list List archives are viewable at: //www.freelists.org/archives/si-list Old (prior to June 6, 2001) list archives are viewable at: http://www.qsl.net/wb6tpu