I've been evaluating AD8302 temperature stability and temperature drift and room-temperature trim. Look carefully at the phase versus temperature curves before you lock in your design if you require excellent temperature stability. The nonlinear temperature coefficient values over the full 0C to +70C range make it very difficult to make the phase response stable to within 0.1 degrees using linear calibration even over a +/- 5C range. The ground pin temperature is the most-accurate temperature measurement point available once the device warms up to its quarter-watt operating temperature.
I required insulation (as in precision frequency sources) over the AD8302 to prevent convection cooling which creates a device temperature measurement error and to create a uniform thermal gradient in the PCB under the chip.
temperature_drift