ADAU1701 DSP Crystal stop oscillating

We have been in production with an ADAU1701 design for 3 years but we have recently been seeing failures that we have identified as the DSP crystal oscillator stopping.
This has probably been happening for a while as we have had a few products with unexplained failures that sound like this fault in hindsight.

The product will happily run for multiple hours or days and then the crystal stops oscillating. a power cycle normally fixes the problem.

During the development we had issues with the crystal oscillator and submitted a post for help on the layout.

Dave Thib kindly helped and reviewed a revised layout which is used in production pcb so we hope this is good. The original post is here,

https://ez.analog.com/dsp/sigmadsp/f/q-a/64396/adau1701-1401-recommended-crystal-layout

Question, can you provide any suggestion as to what may be causing the crystal oscillator to extremely intermittently stop.

Are there any suggested tests we could do to identify the fault.

The crystal used is an Abracon ABLS-12.288MHZ-B4-T.

(The schematic says a ADAU1401 but we use the ADAU1701 is production)

The final PCB layout is below, It is a 4 layer design with a ground plane directly under the top layer

We have tried the following,

  • Thermally stressed the unit both hot and cold with no effect
  • Changed the drive resistor to 200R and 0R0 which appeared to do nothing
  • Performed readback every 100ms on the DSP to catch when it stops responding to SPI commands. This indicates the clock has stopped

We are struggling to think what else to test as the clock is such a self contained block in the DSP with little testing possible.