1
\$\begingroup\$

The AD7606 (and all variants thereof) is an excellent simultaneous-sampling ADC. Simultaneous sampling is necessary for my application. However, this ADC, unlike many others, does not have separate analog and digital power/ground pins. I'm seeing some crosstalk from the digital lines on the analog signals. I'm wondering what the best way to isolate the analog and digital signals is for this chip. Currently I just have one ground plane for the whole system, including the sensors that interface to the ADC.

The sensors are electret microphones amplified by AD8429 In-Amps. The microphones are always powered and so are always gathering, and they're not triggered in any way by the system.

Here's the datasheet: https://www.analog.com/media/en/technical-documentation/data-sheets/ad7606_7606-6_7606-4.pdf

And here's the pinout: AD7606 Pinout

The analog inputs are on the north side of the IC, and the digital lines are on the south and west sides. Should I split the ground planes under the IC and star ground them at pin 2? Pin 26? Pin 47? The datasheet reads as if pin 23, Vdrive, is the digital power supply, but there's no dedicated digital ground. Any guidance is helpful.

\$\endgroup\$
  • \$\begingroup\$ Just out of curiosity, have you already followed the layout guidelines on page 32 and are still having issues? \$\endgroup\$ – JYelton Nov 21 '19 at 18:09
  • \$\begingroup\$ @JYelton second that. There are very extensive layout guidelines in the datasheet. They are there for a reason. \$\endgroup\$ – Maple Nov 21 '19 at 18:58
  • \$\begingroup\$ @JYelton yes, I followed those guidelines. \$\endgroup\$ – Ben S. Nov 21 '19 at 19:01
1
\$\begingroup\$

Do NOT split the grounds underneath the IC. A single ground plane is fine. Instead, keep all analog components to the North and all digital components to the south west. That way, digital return currents are less likely to stray into the area where the analog return currents are circulating.


Is your issue cross talk with your ADC's own digital communications? Or cross talk from communications other devices sharing the same bus as the ADC?

If the problem is the latter, then the solution to this is to buffer the ADC from the rest of the digital bus with a transparent latch (sometimes even with two in cascade). The latch disconnects the ADC from the bus when the ADC is not communicating and thereby reduces crosstalk from the other devices on the bus.

enter image description here enter image description here https://www.analog.com/en/analog-dialogue/articles/staying-well-grounded.html

You should check out everything else in that link too.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ Thanks for the link, I'll check that out. There is only one ADC on the system, nothing else shares the bus. I'm sampling at ~16kHz. The microcontroller and ADC's digital lines are isolated from each other through a set of ADuM25x isolators. Im' still seeing the noise from the ADC's digital communications appear on the analog lines. I have a test point on the sensor input just before the ADC. When the ADC isn't sampling, it's quiet. When the ADC is sampling, the analog signal picks up some noise. \$\endgroup\$ – Ben S. Nov 21 '19 at 18:37
  • \$\begingroup\$ @BenS. There is no way to arrange it so that the digital communications do not have to be going on while the ADC is sampling? It's a SAR ADC, not sigma delta so it should not need to be sampling while the digital comms are going on. \$\endgroup\$ – DKNguyen Nov 21 '19 at 18:39
  • 1
    \$\begingroup\$ I see a contradiction in the description. "noise from the ADC's digital communications appear on the analog lines" and "When the ADC isn't sampling, it's quiet" just do not track together. Are you sure that what you are seeing is digital noise and not some tiny fluctuations on the input lines caused by sampling process itself? \$\endgroup\$ – Maple Nov 21 '19 at 18:55
  • \$\begingroup\$ @Maple Good eye. My brain had somehow interpreted that as "when the ADC isn't sampling the noise doesn't matter because there are no readings" \$\endgroup\$ – DKNguyen Nov 21 '19 at 18:57
  • \$\begingroup\$ @Maple what I mean is when I look at the sensor signal with a scope, there isn't any noise unless I start sampling with the ADC. The value at the sensor isn't changing, but the readings begin to fluctuate when I begin sampling. The sensor itself isn't triggered in any way, it's passive. \$\endgroup\$ – Ben S. Nov 21 '19 at 19:05

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.