• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 98 subscribers
  • Views 477 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

AM335x Eval board, OSC load capacitance

Randy Maruschock
Intellectual 320 points


I'm a little thrown off by the load capacitors on the OSC0 and OSC1 crystals on the AM335x eval board.  From the app notes of various crystal MFGs the caps on the 32.768KHz crystal is what I would expect it to be.  The crystal has a CL of 12.5, and CL is calculated as CL = (C1xC2)/(C1+C2) + Cstray, so assuming Cstray is negligible, 25pF crystals for a CL of 12.5pF makes sense.  However, the 24MHz crystal has a rated CL of 18pF.  With 18pF caps used on this crystal, the math would indicate the load capacitance generated would be closer to 9pF.   Nothing is popping out on the datasheets to me to imply CL should be calculated differently between the two.  Am I missing something?

Regards,

Randy

  • 0
    TI__Intellectual 2150 points

    Yes, this should be closer to 24 pF for the load capacitors on the 24MHz crystal.  However, on the EVM we kept it at 18pF based on initial tests run on the first prototype boards.

    For the EVM, we kept the Cstray in the calculation as nominal 5 to 6 pF.  This let to load capacitor values of 24 to 26pF each. However, on initial EVM board builds, there was a supply issue with the capacitor values that we wanted.  After trying the 18pF capacitors on the initial boards, there were no problems found.

    Nonetheless, on future EVM updates, we may move back to the 24pF capacitor.

  • 0 in reply to ddomke
    Intellectual 320 points

    Thank you for the clarification.

  • 0 in reply to Randy Maruschock
    Prodigy 20 points
    I am also confused about the load capacitance.
     
    I want to use a 25 MHz crystal and the oscillator must oscillate with 25 MHz +/-50 ppm over the full temperature range. I hope to be able to drive my Ethernet PHY from this clock base via CLKOUT1.
     
    In the AM335x datasheet we have the following requirements on the osc0 crystal circuit:
     
    Stability/tolerance    +/-50 ppm
    C1 and C2                    12pF... 24pF
    Cshunt                          < 5 pF
    ESR                                < 46.6 Ohm
     
    From the datasheet I conclude that Cpkg = 0.01 pF (!) which is negligible. Assume Cpcb = 0
     
    The datasheet specifies:
    Cshunt = C0 + Cpkg + Cpcb
    CL = C/2 + Cshunt
     
    With Cpkg = 0, Cpcb = 0, C0 = 5, C1 = C2 = 24 I get CL = 17
    With Cpkg = 0, Cpcb = 0, C0 = 5, C1 = C2 = 12 I get CL = 11
     
    If I for a moment ignore crystal aging, I would need to select a crystal where the sum of the tolerance and the stability is 50 ppm or less. In other words I need a crystal with this specification:
     
    f = 25 MHz
    Sum (tolerance, stability) < 50 ppm
    Temperature range  -40C … +85C
    ESR < 46.6 Ohm
    C0  < 5 pF
    11 pF < CL < 17 pF (depending on C0)
     
    I have difficulties to find such a crystal. Am I doing something wrong here?
     
    I am also asking myself if it is correct to include C0 in CL. A crystal manufacturer only needs to know the external CL in order to tune the crystal correctly. Like it is specified here.