| ??? 11/13/07 02:32 Modified: 11/13/07 02:43 Read: times |
#146915 - About naked firefighters... Responding to: ???'s previous message |
Farshad said:
I've provided solid-ground return paths as much as possible in a two layer board. That's a contradiction in terms! You cannot provide a solid ground plane "as much as possible". A solid ground plane, by definition, is a complete and entire plane without any interruptions, gaps or holes. That's what the word "solid" implies. What you provide is wide ground traces, but not a solid ground plane. That's something totally different. Farshad said:
I was cautious to let all ICs have a solid and low-impedance return to ground and fill the empty area between traces with ground layer. This cannot cure the absence of a solid ground plane. More, if the ground fills aren't connected to the solid ground plane by many many vias, you get even more trouble. Farshad said:
The amount of EMI emitted by the board itself is trivial. Have you measured the emitted EMI? Even only one 74HCMOS gate's output signal not routed properly over a solid ground plane can make you fail the CE radiation test! Farshad said:
That's correct. Maybe not in magnitude of 100 to 1,... Yes, maybe even higher! Farshad said:
It's not fair to call any 2 layer board as improper design. It's entirely fair to state this, because you will hardly pass the CE radiation test when using a two layer board. Farshad, this is even difficult for a multilayer board design, when cables are involved, which leave the board. I exactly know about what I'm talking, because this is my daily job. Farshad, your problems are related to an improper design (which is improper because of the trouble you observed). Like Russell stated, this has not only to do with the PCB, but with the whole wiring scheme and other issues. But I can tell you, that using a two layer board without a solid ground plane will hardly make you solve the problems. The customer might move the board some centimeters from its actual position, or route one cable somewhere else, and problems will arise again. You need a solid ground plane as a radio frequency plane for the cable shields and input/output filters, as an extremely low impedant reference plane to reduce the common mode noise, and as extremely low impedant path for all ground return currents. Remember, you no longer play electrostatics with digital boards but you enter the field of electrodynamics, where the signal traces and the solid ground plane form transmission lines for all digital signal and supply currents. Through your wiring wave packets are running, needing the solid ground plane for proper termination. A board with solid ground plane is an universe more powerful, than any two layer board design. I would instantly lose my job, if I would begin to use two layer boards for our embedded products. It would have been one thing if you had stated, that your setup isn't immune to EMI, and how to correct this. But, to mention that the micro seems to be more sensitive to EMI than the earlier one isn't the proper way to argue at all. YOU decide how much EMI the micro has to withstand, by suited measures, like grounding, shielding and filtering (which need the solid ground plane to work properly!). You can suppress EMI by nearly any amount you want. Have you ever seen a naked firefighter battling against the fire?? Kai |



