Bug #3857

Use SAMD XOSC / PLL / GCLK to allow lower reference frequencies

Added by laforge 3 months ago. Updated 8 days ago.

Target version:
Start date:
Due date:
% Done:



Dieter has suggested a feature to allow using a reference clock at lower frequencies, such as 1MHz.

This is too low to use as an input directly into the Si5351C. However, the XOSC of the SAMD allows for input frequencies down to 400kHz. The crystal oscillator can also be disabled, allowing logig-level clock input. Internal fractional PLL can be used to generate a clock up to 96 MHz from it, and that clocks can then be output via a GCLK block on any of the GCLK_IO pins.

The resulting output could be routed to the secondary input of the Si5351C. This way, we could support e.g. multiplying from 1 MHz to 10 MHz in the SAMD, and then drive the Si5351C as we would with a direct 10MHz reference.

I of course don't know how good the phase noise of the SAMD internal PLL is...

Related issues

Precedes osmo-clock-gen - Feature #3858: Make more GPIOs avaliable for future useResolved03/25/201903/25/2019


#1 Updated by laforge 3 months ago

  • Target version set to hw-v2

#3 Updated by laforge 3 months ago

  • Status changed from New to In Progress
  • % Done changed from 0 to 10

Most likely candidate is a ATSAMD21E18A-MUT which is in QFN-32 package (current: QFN-24) and has 256k flash and 32kBytes of RAM. It's readily available from e.g.Digikey in large volume.

#4 Updated by laforge about 2 months ago

  • Assignee changed from laforge to mschramm

#5 Updated by mschramm about 1 month ago

  • Precedes Feature #3858: Make more GPIOs avaliable for future use added

#6 Updated by mschramm 28 days ago

  • Status changed from In Progress to Resolved
  • % Done changed from 10 to 100

XA input of Si5351C is now connected to Cortex' PA10, which also can be GCLK_IO4. They are connected via a series resistor and a voltage divider of 47k/30k , as the XA input has V_max = 1,3V. Those two resistors are already in the BOM, and they give 1,286V at 3V3 input. The series resistor ahead of 47k might get little bigger than 0R.

#7 Updated by laforge 13 days ago

  • Status changed from Resolved to In Progress
  • % Done changed from 100 to 70

It looks like only the XA output is connected to the Si5351C, but the XOSC input part on the SAMD has not been implemented.

#8 Updated by mschramm 13 days ago

PA14 is XIN, and is already routed to the GPIO expansion header (see #3858). As discussed with laforge, we simply don't have more PCB space for adding a RF-grade connector. maximum frequency on that pad is 32 MHz, so I think this is acceptable.

All external connections on this header and on UEXT will get additional TVS diodes added.

#9 Updated by mschramm 8 days ago

  • Status changed from In Progress to Resolved
  • % Done changed from 70 to 100

(all TVS diodes placed)

Also available in: Atom PDF

Add picture from clipboard (Maximum size: 48.8 MB)