P2 - 12MHz or 20MHz Xtal or Osc ?

Comments

• TonyB_ Posts: 2,125

  2019-10-28 21:32

  In general, 12MHz can generate common video and other clocks with lower multipliers and dividers than 20MHz and based on that I'd use 12MHz.
• jmg Posts: 15,144

  2019-10-28 22:08 edited 2019-10-28 23:19

  Cluso99 wrote: »

  I am just finishing off my P2 PCB design.
  The P2D2 used a 12MHz OSCillator whereas the P2EVAL used a 20MHz XTAL.
  
  I am wondering whether to use the 12MHz XTAL which I use in my P2XBlade and I already have in stock, or use a 20MHz XTAL (yes, they work fine)?
  It would be nice for all boards to use the same frequency at least although we can divide down to a common 4/2/1/0.5 MHz.
  
  So what are all you designers going to use???

  
  Depends what matters most ?
  
  If size matters, higher MHz allows smaller packages, and higher MHz allows higher PFD operation
  
  The most flexible is to allow both a Xtal and an oscillator footprint, & with DAC-VC connection option for VCTCXOs, to allow pretty much any future use.
  
  If you want stability at low cost, 26MHz seems the most-available GPS TCXO frequency.
  Those are usually clipped-sine, & P2 can AC couple clipped sine sources fine. ±0.5ppm ~ ±2.0ppm are common.
  
  For very high precision (sub 1ppm) Clocks l like the look of these Murata Osc parts
  
  Sorting by price, the lowest cost small Xtals at LCSC are 25/26/40MHz
  This 26MHz xtal gives ±10ppm spec and is 4.8c/500+
  
  Looking at Murata Crystals, their best precision comes at the mid-high MHz of 24~32MHz
  XRCGB-F-H Series 24.000 to 32.000MHz ±10ppm max. ±10ppm max. ±2ppm max./year in (2.00mm x 1.60mm) package
  Digikey show 6,8,10,12pF versions, with 24MHz 6,8pF,12pF in stock and 26MHz with 6,8,9,10pF in stock.
  Looks like enough range there to pick an ideal-pairing for P2 Oscillator / PCB
• evanh Posts: 15,187

  2019-10-28 23:59 edited 2019-10-29 00:10

  Cluso,
  I'd be interested to know if a non-20 MHz crystal makes a difference to jitter progression as pre-XDIVP PLL (VCO) frequency is reduced on the revB silicon. My glob-top also has a bad spot peaking just above 20 MHz - https://forums.parallax.com/discussion/comment/1478442/#Comment_1478442 It has to be below something like 2 MHz to be that bad otherwise.
  
  Post-XDIVP can be as much as /30 on top of these numbers.
  
  
• JRetSapDoog Posts: 954

  2019-10-29 01:04 edited 2019-10-29 01:05

  On page 6 of the P2 Eval Board thread, Chip stated:

  I want to kind of steer things towards use of a 20MHz clock source, as it easily divides and multiplies to many needed frequencies. Our crystal oscillator and PLL were designed for 20MHz.

  But his wording suggests that he didn't intend for that to be written in quartz (stone).
• Peter Jakacki Posts: 10,193

  2019-10-29 01:20

  I'm using 25MHz on r2 and r3, either as a crystal direct or feeding the Si5351 which defaults to 25MHz at reset but I can optimise the frequency from there as well as feed other frequencies out to help with clocked devices..
• cgracey Posts: 14,133

  2019-10-29 01:20

  JRetSapDoog wrote: »

  On page 6 of the P2 Eval Board thread, Chip stated:

  I want to kind of steer things towards use of a 20MHz clock source, as it easily divides and multiplies to many needed frequencies. Our crystal oscillator and PLL were designed for 20MHz.

  But his wording suggests that he didn't intend for that to be written in quartz (stone).

  
  Anything from 4MHz up is probably fine.
• Cluso99 Posts: 18,069

  2019-10-29 08:12 edited 2019-10-29 08:13

  I was trying to get a standard value for everyone. Seems like that ship may have sailed
  So I may as well use the 12MHz I already use. It's a 3225 smt package.
• Peter Jakacki Posts: 10,193

  2019-10-29 11:16

  Those 25MHz 2520 rocks I'm using are 10ppm and 29 cents.