Ken, If you log onto the Procket web site, www.procket.com, you will see a terabit router as one of their products. Every PCB in that box has signals as fast as 2.4 GB/S. In every case some or many of these signals are routed on the outer layers, or more precisely layer 2 and n-1, with the two outer layers being "cap" or no signal layers. This box passed CISPRA on its first test and CISPRB on its second test after some gaskets were improved. In no case did any PCB interfere with any other PCB. I've designed high performance products using this technique for more than 20 years with good success in every product. I'm not sure where the notion that traces on outer layers causes problems comes from, but is had never been a problem with any product I have designed. Don't worry about it. Just make sure the ground planes are not split and that you do a good job designing with your power decoupling. Lee > [Original Message] > From: Ken Hayden <khayden@xxxxxxxxxxxxxxxxxx> > To: <MikonCons@xxxxxxx> > Cc: <si-list@xxxxxxxxxxxxx> > Date: 1/30/2004 5:00:46 AM > Subject: [SI-LIST] Re: Traces don't cause EMI - really? > > I read this thread last fall with great interest, and I think I learned a lot. But I > have an immediate application about which I'm still confused, and could use some help. > > This application is for a PCB that will be assembled into a shielded card cage > assembly. This particular PCB will not have any cables leaving the shielded enclosure, > but any number of other PCBs in this 36-card assembly could have cables of various > geometries and filtering characteristics. (This is a telecom/datacom application, and > the PCBs are line cards with numerous flavors of DSL, Ethernet, POTS, and digital > telecom interfaces.) > > We have traditionally laid out PCBs for this application with outside planes, and > absolutely everthing with any high-frequency content was run on inside signal layers. > Using this kind of stackup, and taking many of the usual precautions, we have had good > success in building cards that pass FCC class A, and usually class B. > > In the current (very quick-turn) project, we will be using an embedded microprocessor we > haven't used before. In the interest of building fully functional boards with good > signal and power integrity as quickly as possible, we are considering directly lifting > the processor, DDR SDRAM, gigabit GMII, PCI, and HyperTransport artwork section from the > microprocessor vendor's evaluation kit artwork, and incorporating the artwork into the > rest of our design. > > It turns out that this evaluation kit is in PCI plug-in board format (clearly intended > to run in a desktop PC), and has the following stackup: > > TOP > GND > SIG1 > GND > PWR > PWR > GND > SIG2 > GND > BOT > > This is of course a wonderful stackup for power distribution, but it requires running > close to half of the signals on the outside layers, since there are only two internal > signal layers. The evaluation kit runs many of the DDR SDRAM, PCI, and GMII traces on > the top and bottom layers. Only clocks and HyperTransport traces seem to be strictly > limited to SIG1 and SIG2. > > My question is this: Is this kind of layout likely to radiate significantly into other > boards in the enclosure (relative to a board with more layers, having added planes > outside all signal layers), thereby risking excessive CM currents from the cables > leading from these other boards? As I have mentioned, we have always believed that it > would, and have avoided such a stackup. But I wonder if we've been going overboard and > wasting money on extra layers. > > Ken Hayden > Consulting Engineer > Integral Access > > > > ------------------------------------------------------------------ > 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.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 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