What's Your Favorite Grounding Scheme?

[Aleph Null]

For those that design your own PCBs, what's your preferred grounding scheme and why?

 

[pettert]

I just want to know what's right, and now I think I'm going back to have planes on both bottom and top layer after a couple of boards with only bottom

 

[Aleph Null]

I just want to know what's right, and now I think I'm going back to have planes on both bottom and top layer after a couple of boards with only bottom

My understanding is that at low frequencies (audio) it doesn't make any real difference, at least most of the time. I use ground plains on both sides and try keep vertical traces on top and horizontal traces on bottom. That's down to aesthetics and workflow, not really performance.

 

[pettert]

My understanding is that at low frequencies (audio) it doesn't make any real difference, at least most of the time. I use ground plains on both sides and try keep vertical traces on top and horizontal traces on bottom. That's down to aesthetics and workflow, not really performance.

Alright, I read something about benefits of having traces separated by ground that sounded convincing enough for me to consider having a ground plane on top

 

[jessenator]

Maybe it's the placebo effect, but I've "noticed" a difference between board revisions in how higher noise parts (timers, charge pumps, etc.) interact if their grounds are isolated, quickest path to the main ground (DC jack).

I'm not an EE, and so if I'm working with a 2-layer board, I keep all the supply traces on one layer, and typically the bottom, with a ground plane. Maybe an additional top-layer ground plane is good—I've done it a lot—but maybe it's unnecessary.

If I have to go to 4-layer, I take cues from actual EEs who know what they're doing (even though I'm probably still messing up). I have a dedicated ground plane with nothing else in it, and use it as a separation, and try to keep all other supply traces off of part layers.

Again, I'm not an EE, just a poser.

 

[Aleph Null]

Maybe it's the placebo effect, but I've "noticed" a difference between board revisions in how higher noise parts (timers, charge pumps, etc.) interact if their grounds are isolated, quickest path to the main ground (DC jack).

I was careful to qualify my statement with "at low frequencies". The higher the frequency, the more well defined the return path (ground) needs to be. I haven't worked with timers are charge pumps much, but it makes sense that they'd benefit from a short, unobstructed path back to the ground lug on the power jack. I know when you get to magahurtz and gigahurtz digital stuff you have to start paying attention to the gap between the trace and the ground plane and/or the thickness of the board because of the capacitance they create.