| Electronics > Projects, Designs, and Technical Stuff |
| Crystal oscillator near board edge - EMI concern? |
| (1/1) |
| blueduck577:
I'm currently laying out a board which includes ethernet and the requisite PHY. In the current arrangement, it would make my layout nice and tidy to orient the PHY such that that the 25 MHz crystal oscillator will be close to the board edge. Close is 0.1-0.2" from the edge. Would placing the oscillator here be a bad idea for EMI? The board is 4 layers with two internal planes. This board will eventually go into a commercial product and thus will be subject to EMC testing. The board will mounted on a drawer with a metal drawer rail right next to the side in question. The enclosure is all aluminum. |
| T3sl4co1l:
Funny, I did that just the other week (the boards came in; no, I don't have any pictures handy :( ). They haven't been to the test lab yet, though. Do you have a tentative layout to discuss? Consider the facing surface of the clock trace. How close it it to the edge? All parts of the net: trace, component pins, vias, the whole path. Is it surrounded by ground? (Ground fill on the top side is best, but inner layer ground on a 4+ layer board is a close second and almost always adequate!) How sharp is the oscillator, are harmonics a hazard relative to the trace length (as an antenna)? A metal enclosure can cure a lot of ills, but aren't magic. You may need to grind off paint or anodization from mating surfaces, perhaps add EMI fingers as well. Consider that every seam, gap or opening is an aperture, which can transmit some range of frequencies corresponding to its size. If you happen to induce currents which flow directly across those openings, they will radiate. Tim |
| rhb:
I can't say a priori about a design, but a scope probe without a ground lead or grabber is an amazingly sensitive E field EMI probe. I made a little "condom" out of heatshrink to preclude shorts and it is amazing how sensitive it is on the 1 mV/div scale. With about a 1/4" extending past the shield it doesn't detect anything until you are very close to the source. A small loop of copper wire makes a very effective H field probe. I have been quite amazed at the EMI from the LCD screen on my Keysight 33622A at 1-2 inches. |
| Navigation |
| Message Index |