1

The crystal I want to use is the ABM8W-12.0000MHZ-4-B1U-T3 (datasheet). I know that to calculate the capacitors for a crystal is:

CL = (C1 * C2) / (C1 + C2) + Cstray

However, the load capacitance of the crystal itself is only 4pf. Seeing as I usually assume cstray to be around 4-6pF, would I even need any capacitors for the crysal circuit? Or should I go with a, say 1pF cap?

Of course I can simply leave the footprints for a capacitor on the board in case things don't work as expected, but I would like to know the "theory" behind it.

Or, should I instead be more accurately calculating the capacitance of the trace itself? I tried finding out how to do this, but haven't been able to find a way to do this without manufacturing the board first.

Cyborgium
  • 159
  • 1
  • 7
  • [See also this answer](https://electronics.stackexchange.com/questions/527486/crystal-load-capacitance-for-low-power-applications/527505#527505) for more details on the load capacitance. – Andy aka Aug 26 '21 at 14:48
  • 1
    Small capacitors external to the crystal increases risk of **no oscillation**. Crystal area had been large years ago - your crystal is small, and runs at lower power, higher impedance...favouring smaller external Pierce capacitors. Your worries about how much stray capacitance contributes is a bigger design problem than it used to be. Choose: cheap or reliable,stable. – glen_geek Aug 26 '21 at 18:13
  • @glen_geek What kind of values would you suggest? – Cyborgium Aug 27 '21 at 10:38
  • 1
    Too bad this was closed - Your emphasis on low-C xtals is valid. You are squeezed into a narrow range of added C. Abracon suggest \$C_L=4pf\$ total. To that is added stray C depending on your layout, and gain element that you've guessed. And you haven't stated requirements for frequency accuracy. Relying on stray C may make frequency jittery from adjacent signals - another consideration. Your plan of adding caps to unpopulated PCB is OK if re-work is allowed after fabrication - that might be costly. – glen_geek Aug 27 '21 at 13:56
  • @glen_geek This board is a proto so we will be assembling the pcb ourselves for now, re-work after fabrication will certainly be required. As for the frequency accuracy, are you referring to its PPM value? I picked this crystal specifically for the low PPM value as we have had issues with ethernet on higher PPM. So in that case, 10PPM. If it would help, I could provide the trace details? – Cyborgium Aug 27 '21 at 14:56
  • 1
    Haven't much confidence calculating/estimating PCB stray C. Have run MCU oscillators with stray C only (on breadboard) noting that oscillations sometimes didn't start 'till extra C was added. If you must hit 12 MHz with small error, I'd also add the extra series-R that is sometimes recommended (a kilo-ohm or two)....otherwise you may over-drive the crystal beyond its recommended 10uW - yours is a small crystal that can't take much power. Over-driving can change frequency - consider a bigger xtal. Good luck to you. – glen_geek Aug 27 '21 at 19:11
  • Hm, I hadn't thought of that. I'll look into crystals and their overdrive. Thanks! – Cyborgium Aug 27 '21 at 19:25

0 Answers0