5
\$\begingroup\$

When making a two layer board with the bottom layer being a GND plane, should every GND pad for ICs and passives be directly connected to the GND plane using a via, or should I route all of the GND pins of components close to one another together and then use a single via to connect to the GND plane?

EDIT: Having a GND plane on the top would dramatically reduce the number of vias needed. However, this answer to a related question says:

You are right, there is really very little reason to use both the top and bottom of a two layer board for ground.

What I usually do for two layer boards is to put as much of the interconnects as possible on the top layer. This is where the pins of the parts are already anyway, so is the logical layer to use to connect them. Unfortunately you usually can't route everything on a single layer. Paying attention and thinking carefully about part placement will help with this, but in the general case it is not possible to route everything in one plane. I then use the bottom plane for short "jumpers" only when needed to make the routing work. The bottom plane is otherwise ground.

\$\endgroup\$
  • \$\begingroup\$ Do you not have a GND pour on the top layer? \$\endgroup\$ – Roger Rowland Jan 28 '15 at 18:51
  • \$\begingroup\$ @RogerRowland I'm currently trying to decide whether to use a GND plane on the top as well. I read another question on here where the answer said using a GND plane on top isn't a good idea because with so many signal traces it ends up being very broken. \$\endgroup\$ – Nate Jan 28 '15 at 18:54
4
\$\begingroup\$

It's more complex than this in a lot of occasions because sensitive analogue ground connections may benefit significantly from a local star-pointing regime in isolation to the ground plane with just one solid via-area to the bottom main ground plane.

This is done to avoid inevitable digital currents in the ground plain creating small unwanted digital noise signals between points on the localized analogue "star point system".

Also routing power directly to the most hungry components is also a good idea then, tee-off for the ground plane. Plenty of times I've seen examples of audio amps oscillating because the power tracks have gone to the chip via input connections despite a decent ground plane being used.

On the other hand, with switching regulators, different tactics are used and the main aim would be to group the components that produce a lot of ground current to one area of copper "land" - here I'm thinking of a buck converter where the input capacitor to the chip, the power ground on the chip, the flyback diode ground and the output capacitor ground are star pointed back to the chip's power ground.

This usually prevents upsetting the feedback point from rattling around and you get the expected result of a low noise and well-regulated output.

I'm sure there are other great examples.

\$\endgroup\$
  • \$\begingroup\$ Anyone thinking of doing this needs to do a lot more research first; a "local star-pointing regime" is going to create a virtual ground at a higher voltage than true ground if any of the nets it connects dissipate any current. \$\endgroup\$ – Nick Johnson Jan 28 '15 at 21:31
  • \$\begingroup\$ @NickJohnson - it's not unusual to implement such a method when the local ground regime passes very little current. The whole point is that hardly any significant currents pass in this area. \$\endgroup\$ – Andy aka Jan 28 '15 at 21:49
  • \$\begingroup\$ Thanks for your advice. I decided to use GND planes on both layers, then tie them together every so often across the board using vias. \$\endgroup\$ – Nate Jan 30 '15 at 17:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.