I'm presently having an EMI problem with dot clock emissions from a PC video card. Adding additional ground layers isn't an option since I'm already at the maximum. The return plane is connected to the chassis on this card and the video cables are very good. The emissions are coming right out of the box and not from the monitor or the I/O panel, or cables, etc. All of the signal layers are routed between a power and ground plane. The power plane is, however broken up into a couple of island to accomodate different voltages. All of the unused areas on the signal layers have been filled in with ground pads. The power to the RAMDAC is well isolated and physically separated from the digital signals. The digital and analog grounds are not isolated. The particular RAMDAC being used doesn't facilitate this. I'm only have a problem at one frequency and am presently measuring about 1/2 dB under the limit. I would like to be about 4 to 6 dB under.
The PC that I'm using is very good for controlling emissions. The problem is with my card and not the PC. The next step that I am considering is to add guard traces along the signals carrying harmonics of the problem frequency. I'm also thinking of covering a couple of chips with cans. Another possibility is to put capacitor mounting pads at the load ends of the signal lines and try to clean up the signals a little bit. I don't have much faith in this idea though. If I did this it would be on series terminated lines with series resistors. I was thinking of using resistor mounting pads that would also accomodote ferrites and then swapping back and forth between the two. I was also think of isolating the power (with a ferrite and lots of caps) on the chip that is sythesizing the frequency and on the ones that are doing a lot of switching at the dot-clock frequency. I would appreciate any tips or ideas anyone might have. In addition, I would consider opening a PO and paying consulting fees to a PC video card guru. Max mkel...@es.com
