Good points. Perhaps the units to be tested need to be given a simulated shipping ordeal prior to testing. Dropping, shaking, heating, and cooling. Dave Cuthbert
From: Gary McInturff [mailto:gary.mcintu...@worldwidepackets.com] Sent: Wednesday, March 26, 2003 8:35 AM To: lfresea...@aol.com; randall.flind...@emulex.com; emc-p...@majordomo.ieee.org Subject: RE: PC EMI Derek, While I have never seen the extent of failure (amplitude) you're talking about I have run into a tower or two that didn't comply. It was some few dB out hither and yon. (The monitor was the next biggest concern but I had few troubles with those, although Gateway went through a bad period for awhile some years ago). I was able to take care of it in these few cases by cleaning all of the mating surfaces, resetting all the cards in the bus, and tightening the enclosure fasteners. Given everything else you've been through I suspect your tried that already and I suppose there is no reason you should have to do that. But one of the things that isn't addressed is how well these things travel when shipped or when people add internal cards etc. Joints loosen etc. Obviously, that's not the complete answer by any stretch, but it got me what I needed - a system that met class B. Then I left the unit at the test lab, and spent a little time "refurbishing it" every so often. Design wise, I've always found the I/O card slots to be the most problematic. They really aren't designed very well from an attachment perspective. I've usually had a couple of small indents added to the card face plate, in the center of the face plate to provide a little positive contact. Typically they hook into a slot on one end and have no real contact force from that end to the end which screws into the chassis. If the face plate, or the chassis sheetmetal has a bow into a relatively large seem is left open. I was dealing with gigabit network interface cards (GNICs) at that point and they have some high frequency stuff very close to this interface. Basically even a good chassis can go bad if it isn't maintained over time or after shipping. I'm not suggesting this is the overall solution to the problem that has your snuggies in an uproar, :) but it may help on the onsey-twosey case for your clients in the lab. Gary The note below is probably better and worth 2 cents so this must then be the 1 cent version. From: lfresea...@aol.com [mailto:lfresea...@aol.com] Sent: Tuesday, March 25, 2003 5:59 PM To: randall.flind...@emulex.com; emc-p...@majordomo.ieee.org Subject: Re: PC EMI In a message dated 3/25/2003 5:10:12 PM Central Standard Time, randall.flind...@emulex.com writes: I am sure you also addressed this but since I have 2 cents to add....... A front-end overload condition on the receiver/analyzer, or an overload at the pre-amplifier, can cause errors in measurement that may seem transparent. A local transmitter could be overloading your measurement system and increasing your emissions readings.... Again, take it at what it is worth! (2 cents) Hi Randy, Where I live, lucky to have electricity :-) When I set my software up, I played with attenuators to make sure that with my signal path, I have at least 20 dB over the Class A limit before I get into any of the signal measurement chain elements either going into compression or saturation. When ever I get an outage like this, I always add 10 dB in the signal path and remeasure... Just to make sure the two curves track. Cheers, Derek.