Max, I don't think your comments are off the subject at all. From your e-mail, you appear to describe routing the power pin of the oscillator through the ferrite rather than isolating one plane from another.
It makes me think about my comments on islands from the last e-mail and qualify them by saying that if an island is used, the decoupling on the isolated side had better be excellent. The signal return path issue still exists though. You describe the basic concept used for board decoupling and isolation from the "off board" system DC supply. That is, build series impedance looking back from the load through the main system power supply. Simultaneously, supply local charge to the load through a low imedance, low area loop. This is what one is trying to do when one builds an island. Again, the local decoupling had better be adequate at the frequencies of interest or one will end up with a large voltage drop across the isolation device (ferrite in this case) and with a noise problem. It is probably smart that, as I think you imply, an island is NOT used, since it is difficult to get the decoupling right at high frequencies. One reason may be that the island is isolated from the distributed board capacitance (other stray parasitics notwithstanding) created by the large main board area and adjacent stackup of the power and logic return planes. Thus, decoupling of high order harmonics (200 - 300+ MHz) may actually be degraded to the extent that the island is not allowed to "see" this distributed capacitance and that the island's distributed capacitance may be inadequate. The distributed capacitance becomes the charge source of choice when the discrete circuit is operating in the inductive reactance part of its impedance curve. I don't like to think of a component of being "forced" to take charge from a source because that implies that there is not a quality available (inadequate HF decoupling); the component should find what it likes close by and take it freely. Semantics?...;-). Regards, [email protected] PS: If everyone else is tired of hearing about this subject, in the future, I will respond only to the sender. Please let me know. ---------- From: Max To: Tony Fredriksson Cc: mikev; mkelson; emc-pstc Subject: Re: RFI problems with PC's ... List-Post: [email protected] Date: Tuesday, April 16, 1996 3:15PM [snip] %> %>It is common to see high frequency circuits on PCBs separated with %>ferrite beads and having isolated, DC-coupled 0V planes. Often, the %>oscillator is designed with its own Vcc/0V island. % [snip] This is a little bit off topic from making islands, but based on my memory of some material I've read and my own opinion, I think the following might (i.e., I'm guessing) be a good idea. With all oscillators, you could run the power (not the ground) through a ferrite. Then you would place a bypass cap to ground, between the ferrite and VCC. The ground plane would not be filtered or separated in any way--it remains in one solid piece. The net effect should be a cleaner power and ground system. What this ferrite/cap configuration would do is to force the oscillator to draw all transient current from the capacitor. Or, in otherwords, the rest of the caps on the board would be unable to help provide fast-transient current because of the ferrite. This would keep the current loop (power AND GROUND) small and prevent it from infecting the rest of the board. The path for the transient current surges would be from the capacitor to the IC's power pin, out the IC's ground pin and back to the negative side of the capacitor (a relatively small loop). Come to think of it, this EMC method would probably work with any type of IC, with the only limitation being the storage capacity and ESR and ESL of the capacitor and the impedance of the leads and via's, etc. If memory serves and I interpreted the article correctly, I think this method of keeping VCC and ground clean was described in the EDN article that I mentioned previously. Max [email protected]
