0
\$\begingroup\$

I am reviewing a 6 layer board that has analog signals on the left and right side of the board.

I have always read that analog circuitry should be grouped together for easy plane segmentation. However in the real world that is not always possible, such as this case...

planes

The image labels, in text, where the analog circuitry, power circuitry, and star grounding point are. The star grounding point is where the the Vin/GND terminal block is.

The layer shown is the internal ground plane, with the darker green area being the analog plane and the whiter/lighter green area being the digital ground. All of the analog input traces are covered by the AGND plane area.

Is this is valid grounding scheme if the bottom layer is a complete digital ground pour?

Technically I know the grounding for analog and digital should be on the adjacent layer to the signals, but this is a basic relay output and analog input (8 bit) board and not high speed.

\$\endgroup\$
  • \$\begingroup\$ Your design may not be high speed but try telling that to EMI hitting the board from an external source. \$\endgroup\$ – Andy aka Dec 13 '18 at 18:19
  • 1
    \$\begingroup\$ Sometimes GND plane separation is a good, sometimes it is a very bad thing because it can introduce additional GND loops. Instead, check the following rules in your design: 1) Think of every! possible current loop and try to minimize all the loops with the highest signal rise time or sensitivity. 2) Two current loops should have little common trace as possible. This is done with star grounding. \$\endgroup\$ – Stefan Wyss Dec 13 '18 at 18:46
  • 1
    \$\begingroup\$ My main concern with this approach is with the signals crossing the boundary between planes. Because of the disruption of the return path, these lines would be particularly susceptible. \$\endgroup\$ – Cristobol Polychronopolis Dec 13 '18 at 20:03
  • \$\begingroup\$ @CristobolPolychronopolis yea I am aware of that. The analog signals do not have any breaks in the plane for return path. And I am hoping that the digital signals that cross the boundaries can use the bottom digital ground pour for their return path. \$\endgroup\$ – zme Dec 13 '18 at 20:20
  • 1
    \$\begingroup\$ @zme The current in those traces crossing between the planes will actually split and use both the bottom digital pour AND will try to use your split plane, the edge cannot see the future! In the first instance I would forgo the whole split plane and star point thing totally, it is only ever a win with low frequency precision small signal stuff and usually turns out to be unnecessary. uV and uA level stuff I would worry about it, but for anything else, meh! \$\endgroup\$ – Dan Mills Dec 15 '18 at 18:09

Your Answer

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

Browse other questions tagged or ask your own question.