Gary, you are pulling my leg :) Alex, The fuzzy/messy BB noise you see and describe below is normal, it is Ethernet signal. You may or may not have superimposed other stuhh here as well, which I don't know. The level, however, if it is high relative to Class A radiated emission should not be normal. Good designs in multiport boards should typically have a decent margin if everything is done "correctly" (I know - it's not a 100% precise term). Center-taps at the transformer, especially the transceiver-side are critical here in this frequency range. They, together with capacitors to GND form series LC(R) circuit, which should have as low impedance as possible in order to suppress EMI (common-mode). That typically means the connection should have low inductance. Pulse is a good company and they might help you with your problem (at the transformer-level) if you contact their engineering in San Diego, assuming the problem is related to and can be rectified by the magnetics. Regarding the GND separation (at a risk of getting heated replies which I may reply to or choose to ignore) you should first distinguish between integrated connectors (with built-in magnetics) and discrete connectors, with magnetics in a separate package. With integrated connectors - NEVER split the GND. With discrete connectors and magentics, both approaches can work, and I would chose the approach depending on other factors, such as presence or lack of metal "pan" under the PCB, other traces that go to the front (e.g. LED or POE traces - would they violate the moat if I had one?), presence or lack od adjacent boards (e.g. plug-in modules or daughter boards over the I/O section). In no case should a moat allow coupling of signals (noise) from anyhting else that could potentially come across it (either through the on-board traces or by coupling mechanism from the adjacent boards, cables etc.). If you can't cahieve a clean moat, as I just descrbied, I! recomm end no moat - just a single GND for everything. In any case, pull back all junk that does not belong there and Vcc planes away from the front-end. It also helps in reducing noise coupling if you stitch the GND planes after the pull-back. BTW, I am assuming here multi-layer boards, not 2-layer PCBs, which people are also doing these days, and which is much more involved and many rules must be broken, and one has to hope it would work (it can work, at least on some designs). Having said that, a VERY important part of the picture is how you connect the PCB reference GND to the system (called "chassis", enclosure..). I strongly recommend a metal "pan" under the board, all PCB GND tied to the "pan" especially (but not only) on the front-end. That "forces" the reference voltage on the board to be close to the enclosure potential, so that the common-mode due to GND voltage on the PCB is low. This was a very short version. If you use Broadcom devices, you may find useful Application Note EMC-AN102. It is only available to companies who are BCM customers, so I am not sure if you can get it. I am not in a position to send it to you. Regards, Neven -------------- Original message -------------- From: "McInturff Gary" <[email protected]>
> Was kind of hoping Neven would speak up here. The digital and the I/O > grounds should be separate. You are going to mote between the main pcb > and the I/O connections. The common mode chokes span the mote. Neither > the power or the digital ground planes should extend under the mote. > You'll probably want to put a ferrite between + stuff and the isolation > transformer center taps, and you can experiment with the filtier caps > between the CT and digital ground for noise reduction. ON the output of > the transformers you had to watch the capacitance between the center tap > and the I/0 ground. You will violate the Ethernet specs pretty quickly > if you put much more than 1 nfd, 2kV but you can play with the R values > a bit. K. Make sure the Ethernet connector is tied to the I/O g! round - > don't break off the mounting tabs. You need 1500 Vdc between signal and > ground, pin-pin and between I/O ground and PCB digitial ground. Adjust > the mote spacing to get this right. In the past I've found it helpful > for this split to be as small as possible but still meeting the 1500 Vdc > dielectric gap. > > The "broad band" noise you describe suggests to me that you do indeed > have some sort of ground problem going on. > > Since this is EMI you'll do all the right things and still find > something hosing you up - plenty agrevating but good luck. > Gary > > -----Original Message----- > From: [email protected] [mailto:[email protected]] On Behalf Of John > Woodgate > Sent: Thursday, April 27, 2006 10:47 AM > To: [email protected] > Subject: Re: 100 Base-T chassis ground > > In message <20060427170820.34695.qmail@web34309! .mail.m ud.yahoo.com>, > dated Thu, 27 Apr 2006, Alex Horvath writes > > >In my experience it's sort of a no-brainer to have a separate chassis > >ground. My question is would you expect that the problems we are having > > >could be caused by the common ground implementation? > > It's very likely. The digital 'ground' is sharing its disturbing energy > with the output cable in common-mode. > > Just modify a sample to have a 'clean' ground on the output side of the > port and see if it helps. In EMC, a gram of test can outweigh a tonne of > > theory. > -- > OOO - Own Opinions Only. Try www.jmwa.demon.co.uk and www.isce.org.uk > 2006 is YMMVI- Your mileage may vary immensely. > > John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK > > - > ---------------------------------------------------! ------- ------ > This message is from the IEEE Product Safety Engineering Society > emc-pstc discussion list. Website: http://www.ieee-pses.org/ > > To post a message to the list, send your e-mail to [email protected] > > Instructions: http://listserv.ieee.org/request/user-guide.html > > List rules: http://www.ieee-pses.org/listrules.html > > For help, send mail to the list administrators: > > Scott Douglas [email protected] > Mike Cantwell [email protected] > > For policy questions, send mail to: > > Jim Bacher: [email protected] > David Heald: [email protected] > > All emc-pstc postings are archived and searchable on the web at: > > http://www.ieeecommunities.org/emc-pstc > > - > ---------------------------------------------------------------- > This message is from the IEEE Product Safety Engine! ering S ociety > emc-pstc discussion list. Website: http://www.ieee-pses.org/ > > To post a message to the list, send your e-mail to [email protected] > > Instructions: http://listserv.ieee.org/request/user-guide.html > > List rules: http://www.ieee-pses.org/listrules.html > > For help, send mail to the list administrators: > > Scott Douglas [email protected] > Mike Cantwell [email protected] > > For policy questions, send mail to: > > Jim Bacher: [email protected] > David Heald: [email protected] > > All emc-pstc postings are archived and searchable on the web at: > > http://www.ieeecommunities.org/emc-pstc - ---------------------------------------------------------------- This message is from the IEEE Product Safety Engineering Society emc-pstc discussion list. Website: http://www.ieee-pses.org/ To post a message to the list, send your e-mail to [email protected] Instructions: http://listserv.ieee.org/request/user-guide.html List rules: http://www.ieee-pses.org/listrules.html For help, send mail to the list administrators: Scott Douglas [email protected] Mike Cantwell [email protected] For policy questions, send mail to: Jim Bacher: [email protected] David Heald: [email protected] All emc-pstc postings are archived and searchable on the web at: http://www.ieeecommunities.org/emc-pstc
