0
\$\begingroup\$

I stumbled over this simple solution to drive a crystal oscillator, but I need to connect a 25.17MHz crystal instead of the 10MHz in this schematic. How do I calculate R1 and C for a given frequency f(max)? If C is taken from the crystal's datasheet, how do I calculate R1 then?

schematic

simulate this circuit – Schematic created using CircuitLab

I'm going to use this as a VGA pixel clock driving a couple of counters. If anyone has an even simpler solution, it's mostly welcome.

\$\endgroup\$
3
  • \$\begingroup\$ Check this document out: crystek.com/documents/appnotes/pierce-gateintroduction.pdf \$\endgroup\$
    – Leoman12
    Commented May 3, 2020 at 3:42
  • \$\begingroup\$ Note that both NOT gates are not Schmitt triggers. \$\endgroup\$
    – Andy aka
    Commented May 3, 2020 at 8:22
  • \$\begingroup\$ As a rough guide, the resistance of R1 needs to be about equal to the reactance of C at the crystal frequency. \$\endgroup\$
    – user173271
    Commented Jun 10, 2021 at 17:00

1 Answer 1

Reset to default
1
\$\begingroup\$

An easy solution is to buy a ECS-100A-251.7 25.175 MHz oscillator, if that's close enough for you.

R1 has to do with limiting the drive power to the crystal. C is calculated from the load capacitance on the crystal datasheet and the input+ stray capacitance- twice the load capacitance minus the input+stray capacitance.

You could also buy a pre-programmed programmable oscillator such as the SIT2018BE-S1-18E-25.166000G which is a Digikey value-added item and available in 25.166 MHz. Also available in 25.175MHz as SIT8008BI-13-18E-25.175000G.

I would not suggest using this circuit- oscillators made with Schmitt triggers are notoriously unpleasant. It should only be attempted with an unbuffered inverter in the oscillator position.

\$\endgroup\$
4
  • 1
    \$\begingroup\$ 25.175 MHz would be exactly the correct frequency for VGA pixel clock. \$\endgroup\$
    – Justme
    Commented May 3, 2020 at 9:54
  • 1
    \$\begingroup\$ I once took advantage of a 'feature' of a particular series of logic (74LCX04) to make a very reliable oscillator. The part had significant ground bounce due to lead inductance so it always started up correctly. \$\endgroup\$
    – Peter Smith
    Commented May 3, 2020 at 15:02
  • \$\begingroup\$ @PeterSmith That part has a really complex internal circuit. I guess the EMI suppression circuitry also suppresses VHF/UHF oscillation that can occur sporadically with buffered inverters. The classic solution is to use the 74HCU04, of course, but if the transconductance isn't high enough (typical graph only is supplied) it might not start (low temperature and low supply voltage would be worst-case for a given unit). \$\endgroup\$
    – Spehro 'speff' Pefhany
    Commented May 3, 2020 at 15:11
  • 1
    \$\begingroup\$ @SpehroPefhany I seem to recall that the classic unbuffered inverter was the CD4049UB but it was of course somewhat limited in frequency range. I needed 20.48MHz in my particular application (in 1995). \$\endgroup\$
    – Peter Smith
    Commented May 3, 2020 at 15:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.