| ??? 11/19/07 16:48 Modified: 11/19/07 16:52 Read: times |
#147195 - Inter-plane coupling Responding to: ???'s previous message |
David said:
On your suggestion of using another plane for a high current or noisy ground, rather than splitting a single groundplane - wouldn't that be counterproductive? If there are two parallel planes, I'd have thought that at high frequencies, the large capacitive coupling would make them perform as one. This would ruin the point of separating them, surely?
At least splitting a single plane minimizes inter-plane coupling. You are entirely right, two adjacent planes show a heavy capacitive cross coupling and when using separate layers to isolate currents and potentials from each other you should also put a shielding ground plane between them, when HF (high frequency) signals are involved. If you have a HF carrying copper trace running in a separate layer and put two adjacent ground planes to it, then a transmission line is formed keeping all the magnetic field inside the gaps between these three planes, like you can see it with every coax cable. The room outside of the adjacent planes is free of electric and magnetic fields. You will only see something, if the impedance of planes is high enough to introduce some ohmic voltage drop, which can happen, if you route very high currents. But in usual applications the surrounding by adjacent ground planes forms an effective shield. Even the supply planes can provide this shielding, but then lots of supply decoupling caps distributed over the whole plane must exist. I am in touch with EMI testers from time to time, and one day they showed me a board, that couldn't pass the CE tests. It was an inverter with a solid ground plane. A very high and noisy current was also using the solid ground plane and generated so much common mode noise on the entire ground plane, that the cables were always radiating too much. Only, when they diverted a certain area from the ground plane for this noisy current, which means a kind of splitting of ground plane the radiated EMI was ok. Here, it wasn't so much the HF content, but more the height in Ampères, which caused the EMI. Splitting the ground plane, on the other hand, doesn't really work for highest frequencies. The gap becomes rather low impedant due to the stray capacitance between the two splitted planes, because the gap is not so much wide usually. I prefer a 5mm (at least 4mm) gap at the closest point and make it as wide as I can elsewhere. For audio frequencies and slightly beyond, the splitting of ground plane can be very very helpful. In digital designs you must do some careful anaylsis though (testing, what works best and radiates lowest). Kai |
