1
\$\begingroup\$

Based on my experience and reading, I would generally put a stitching capacitor between power and ground (for example) if my signal was changing reference planes.

http://www.hottconsultants.com/techtips/pcb-stack-up-6.html

Do I need a via or stitching cap when I transition between physical reference planes of the same potential?

Let's say for the sake of argument that I have a situation were I absolutely need to run this signal from DGND to AGND briefly. Normally, I would put a stitching capacitor....but I also would tend to keep as little coupling as possible between AGND and DGND.

Assuming that this signal (SPI speeds) does need to briefly cross AGND, should I put a stitching cap there?

\$\endgroup\$
  • \$\begingroup\$ Can you make the SPI edges (Trise, Tfall) very slow, so the injected current is not a problem? Can you define "not a problem"? \$\endgroup\$ – analogsystemsrf Jul 21 '17 at 5:32
  • \$\begingroup\$ I think my actual implementation will be no issue with regards to noise. I guess I was trying to make a more general question regarding having stitching capacitors between AGND and DGND. Coupling between the two has always been a hard no for me, but I hadn't come across this issue before. As with all proper grounding questions, the true answer is probably "it depends". \$\endgroup\$ – scld Jul 21 '17 at 13:25
1
\$\begingroup\$

Assuming that this signal (SPI speeds) does need to briefly cross AGND, should I put a stitching cap there?

Sort answer, yes.

The real answer is: It depends. It depends on how the return current is going to get back to its source and how much current there is. If its a few mA's of switching current, it will cross the plane and return to the source through AGND. If AGND is starred off the board (you'll make a nice loop antenna), or is the ground is not a sufficiently low impedance then you'll get common mode noise of the current going down the resistance of AGND and returning back to the source. If the measurement your trying to make is below uA range, this could be a problem as a few mA's multiplied by a tens of mΩ will give a common mode voltage bump in that range.

The stitching cap is to equalize the return current while its being dumped onto the AGND side and help prevent the short term voltage bump that the AGND plane might see from switching. The stitching cap should be placed next to the digital traces as the return current's value is highest there (due to mutual inductance from the trace adjacent to the plane).

I prefer not to have AGND and DGND separate, if you read Otts book he mentions that in most cases split ground planes should not be used. Use a "bridge" somewhere on the board. (I prefer slots with the "bridge" on one side, but I have good EMC control)

enter image description here
Source HOT consulting Mixed signal Design

If the SPI switching current is low, then you probably wouldn't have to worry about common mode noise.

\$\endgroup\$
  • \$\begingroup\$ so in this particular case, I have a motor controller with a separate motor ground and digital/analog/signal ground as another. they are separate planes except all of the connectors are on one side, hence the slight reference plane change....as for the separate ground planes, they're coming in from the outside world as signal and power ground, so they're bridged together but in another housing. \$\endgroup\$ – scld Aug 2 '17 at 23:43
  • 1
    \$\begingroup\$ Post a picture of the PCB also post your requirments for ground plane bounce or ADC reqs \$\endgroup\$ – Voltage Spike Aug 3 '17 at 0:09

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.