Trimming 16MHz crystal in production for BLE??.........

[Smokey]

I'm making a (hopefully) standalone production test fixture for a product with bluetooth.  The DUT has a 16MHz crystal, and from what I understand that needs to be +/-50ppm max for BLE to work.  I can output the clock on a GPIO for measurement and the DUT micro has trim capabilities built in, but I need to accurately measure the clock and get that data to the fixture processor to set the trim.

Because I plan on making many of these fixtures, I would rather not have a boat anchor frequency counter hooked up to each one.  Cost is also always an issue for many fixtures.  Something self contained that doesn't itself require calibration would be optimal.

What are the best options for measuring a 16MHz clock accurately enough (+/-5ppm?) to perform this trim in a formfactor that is ideally board mounted, self contained, and inexpensive?

[exmadscientist]

I always just buy a better crystal. It quickly pays for itself in test complexity.

[Smokey]

Normally I would agree that a better crystal is easier.  In this case, they recommends both +/-10ppm spec and trimming, which is done entirely in software (no external caps).  Better than 10ppm gets a little exotic.

You got it with the crystal frequency effecting the RF output.  That's the concern.  I don't have a spectrum analyzer in the production test setup, but that would probably work trimming it that way.  But at that point a counter would be easier/cheaper I think.