PVDD/PGND in ADAU1701 --- digital or analog?
I'm planning to have two distinct areas of the board for digital ground plane and analog ground plane, joined at a single point.
Should PGND be connected to the analog ground plane, or digital ground plane? The datasheet states, for PVDD: "3.3 V Power Supply for the PLL and the Auxiliary ADC Analog Section". That sounds extremely scary :-\ PLL sounds not only digital, but as noisy as it gets for a digital signal, since it is all edges. Does the chip really mix that with circuits in the analog section???
Another "scary" detail in the datasheet is that it states that "The AGND, DGND, and PGND pins can be tied directly together in a common ground plane" --- I would *much* rather isolate digital and analog ground planes, connecting them at a designated point only.
Any comments? Have people out there successfully designed boards with a single ground plane? (by successfully, I mean: do they work well, with good noise and EMI behaviour?)
Thanks,
Carlos
--
Audio Signal Processors
SigmaDSP Audio Processors
adau1701
Hello Carlos,
Yes it may appear scary to mix analog and digital grounds, yet this is exactly what's recommended by AD engineers in this document:
https://ez.analog.com/dsp/sigmadsp/w/documents/5199/sigmadsp-pcb-layout-best-practices
Where I work we have built a twelve channel mixer using six ADAU1701s and a PIC uC. It's on a four layer board with only one ground plane. Initially the prototype's performance was disappointing, yet it turned out that our crude measurements as well as an unstable 3.3V supply were adding to our measured noise. Presently not only are we happy with this product but we sold out our last production run and now our only problems with producing more are mundane ones like unavailable parts. Our experience with the original issues is recorded here if you're curious:
https://ez.analog.com/dsp/sigmadsp/f/q-a/65702/too-much-noise-in-a-sigmadsp-based-mixer/
So yes, one ground plane is all you need. In fact we have also made a successful -1701 based product with no ground plane at all, just generous copper pours on a two-sided board. This worked for us thanks to one of our engineers who is really good at PCB layout.
As for EMI -- I did some quick tests on the 2-layer board mentioned above, the results are here:
https://ez.analog.com/dsp/sigmadsp/f/q-a/66609/emi-from-sigmadsp-processors
Best regards,
Bob
Please log in to reply.