Crystal PCB design for custom Teensy

[tyzjames]

Hi everyone,

I'm currently designing a Teensy on a board, and have a question regarding the PCB design for the crystal. From that I have read:
1. Ground plane under crystal
2. No digital signals below crystal
3. No parallel lines with crystal signals

I am not certain if I should be grounding the Inner plane (purple) with the bottom ground plane (red).

Is this design sufficient?

 

[HWGuy]

I think the connections to the crystal are not connected to the correct crystal pads. Probably mixed up top and bottom view. Check again with the datasheet.

PCB layout: Add a GND guard ring on the top layer and connect the 2 free pads from the crystal to GND. The Inner plane is not needed. But if you add it, connect it to GND.



More: 2.1.3.3.2 MCG oscillator http://www.nxp.com/assets/documents/data/en/quick-start-guide/KQRUG.pdf
Here as example the KL26Z Freedom Board https://i.imgur.com/2m8lmeN.png

 

[tyzjames]

Thanks for spotting that!! I mis-read the diagram in the datasheet.

Thanks for the links as well. Pasting a part from the nxp manual for future reference

The crystal or resonator should be located close to the MCU. No signals of any kind
should be routed on the layer directly below the crystal. A ground plane on the layer
directly below the crystal is recommended. A guard ring should be placed around the
crystal and its load components to protect it from crosstalk from adjacent signals on the
mounting layer. This guard ring can originate from the VSS pin adjacent to the crystal
pins. Note that the guard ring (and load capacitors) is connected to the ground plane in
Figure 2-4 and Figure 2-5.

 

[madmattd]

HWGuy to the rescue on rotated crystals again. He caught my same goof (after I spun the boards, as I didn't post beforehand), which fixed a startup issue I had (no functioning crystal will do that...).

Another image of the guard ring + isolated ground plane for the crystal. Note: with the wrong crystal pin orientation. The guard ring also really should continue across the top which was later corrected, though the board operates as-is (when I rotate the crystal when reflowing). I don't have an image handy of the correct setup, but this is close enough to give the idea.