Quantcast
Viewing all articles
Browse latest Browse all 4325

Other RP2040 boards • Re: custom RP2040 PCB - crystal does not oscillate

Thank you for this very thoughtful response!
If the only thing that has changed is the core (spacing between layers 2/3) then it doesn't matter at all.
That could still possibly be your problem here, as you already have the crystal a bit further away from the RP2040 than typical. This extra capacitance is in effect parallel with your C20/C21, so you could try reducing the value of those parts to give something equivalent to what you started with - maybe 10pF or 12pF would be better. For experimental purposes I might try removing them altogether and see what happens.
I will try to remove the capacitors tomorrow, and report back.
(note BTW that RP2040 has an internal linear regulator not an SMPS (RP2350 has SMPS), but the 1V1 will still be extremely noisy as it has all the CPU logic switching at high frequencies).
Ah, thank you.
Proposed Improvements - PCB 3
I've reworked the design to minimize distance between the RP2040 and the crystal, let's call it PCB 3.
Image may be NSFW.
Clik here to view.
Screenshot 2024-09-01 at 10.25.58 PM.png
Using an online microstrip capacitance calculator, estimates for parasitic capacitance of the XIN/XOUT traces, plus 0.2 pF for the 1k 0402 resistor:
- PCB 1: 2.4 pF
- PCB 2: 2.0 pF
- PCB 3: 1.2 pF

Regardless of the the accuracy of the estimate, the shortened distance should make a big difference in extra capacitance.

I also notice that "Hardware Design with RP2040" mentions both 2-layer and 4-layer stackup, and the corresponding "Minimal Design Example" use two layers. Again using an online microstrip capacitance calculator, it's appears that my 0.2104mm dielectric DOUBLES capacitance relative to a 1mm dielectric. That could explain how the very same crystal and capacitors are given so much extra room in the "Minimal Design Example".

I've also reviewed two of Robert Feranec's videos about crystals, and noticed that none of his examples use ground fences. In case my fence was creating coplanar capacitance, I've removed it, but left 0.9mm distance from the MCU and a nearby decoupling capacitor.
- https://www.youtube.com/watch?v=vfRLHWiydLQ
- https://www.youtube.com/watch?v=dHgyZ7Pn9fw

Though unlikely material, it is very easy to move the +1.1V trace away from the crystal, and to revert to 1.6mm PCB thickness.

Statistics: Posted by jacobmarble — Mon Sep 02, 2024 5:48 am



Viewing all articles
Browse latest Browse all 4325

Trending Articles