Crystal Capacitance Tuning

[rezarethink]

I've been able to use a 26Mhz and 40Mhz xtal in testing, but for production if 40Mhz is more supported, that's what I would like to design in. I'm considering this XTAL: http://datasheet.octopart.com/FA-128-40 ... 700903.pdf

For other designs, I typically double the capacitance rated on the datasheet for the xtal in the design. However, I imagine that there are higher tolerances required for this design given that it'll be doing RF so I want to make sure I'm picking the right values.

In one of your datasheets (which I can't seen to find, except for a screenshot of the relevant section here in chinese viewtopic.php?t=1924) there were some specified ranges of capacitances and they were asymmetric for the 40Mhz option.

How should I go about picking the right capacitance values? Do I need to measure the the output freq of the xtal and adjust the caps till I'm smack in the middle of 40Mhz? And I would imagine the oscope probe will influence the oscillating frequency. Is there some way to output the 40mhz on an external pin so I can measure it there while making adjustments?

For the XTAL that I posted the datasheet above, I'm planning on using two 20pF caps -- how reasonable is that?

Thnx,
Reza

[iot-bits.com]

What I would do for a 12pF crystal is exactly what you mentioned. But if you are using a thin, especially 4 layer PCB, you should account for the trace capacitance as well, it can be of the order of 4-6pF in some cases.
Apart from that, you can put out 40MHz clock, or a divided version, on the CLK_OUT pins by programming and count that output. Using 0402 or 0201 capacitors with crystals will be preferable.
Directly probing a crystal will definitely affect the output. Direct probing is only to see if the clock is present, won't trust that any further in an RF application.