Ground in RF System: Separating Analog and Digital

[fidel]

Hi all. I'm working on an RF system including a 2.4G transceiver and a GPS receiver. I'll use a ground plane. What I intend to is to separate analog ground and digital ground and connect them together at the entry via a 0ohm resistor. My friend argues that a whole ground layer that connect to both analog and digital signals would work better than splited ground. It seems to me that digital noise will enter analog circuit more easily as the analog and digital pins are not separated into different parts on these chips. Could you give me some advice on this if you have such experience? Thanks.

 

[MATLABdude]

Hi all. I'm working on an RF system including a 2.4G transceiver and a GPS receiver. I'll use a ground plane. What I intend to is to separate analog ground and digital ground and connect them together at the entry via a 0ohm resistor. My friend argues that a whole ground layer that connect to both analog and digital signals would work better than splited ground. It seems to me that digital noise will enter analog circuit more easily as the analog and digital pins are not separated into different parts on these chips. Could you give me some advice on this if you have such experience? Thanks.

As you suspect, it's done to keep noise (from currents returning to ground) isolated in their individual systems (analog, digital, RF, etc.) For instance, the current returning through analog ground will not push up (V=I*R) the potential of the digital ground. The small (and relatively higher impedance) connections between the various grounds in your system keep the nominal ground potential roughly the same across your various separate grounds.

Here's a more complete and involved answer that came up when I Googled your response (Why separate grounds?) to see what was out there (agrees, at least, with my understanding of things):
http://www.epanorama.net/wwwboard/messages/1942.html