Santity check units. Is 10uW a plausible drive-level for mcu crystal?

[julian1]

I have a stm32 board, using a HC49 crystal for HSE. I want to confirm the drive level does not exceed the crystal manufacturer's power limit (0.5mW).
 
Following the approach in 4.4.2, p13.
http://www.emcu.it/DOC-Digitale-Generale/AN2867%20Oscillator%20design%20guide%20for%20ST%20microcontrollers%20rev5.2.pdf

With,

F=8MHz          xtal datasheet.
ESR=60ohm   xtal datasheet.
CL1=CL2=15pF
Cs=8.5pF      (implied from CL1=CL2=15pF)
Cprobe=9.5pF  "A typical 10Mohm probe is 9.5 pF"
Vpp=0.8V    measured with scope at drive pin of crystal.

I calculate,

> var square = x => x * x;
> 60 * square(3.14 * (8*1e+6) * ((15 + (8.5 / 2) + 9.5)* 1e-12)) * square(0.8 ) / 2
0.000010014198528
== 10uW.

The crystal is rated at 0.5mW, which is a big margin (50x). As long as I haven't made any mistakes with formula, values or units.

My gut says this detailed calculation is overcomplicated and unnecessary, and if common parts are used, and it starts oscillating then it's probably good enough.

On the other-hand most schematics show an optional current-limit resistor to reduce power.

Does 10uW sound reasonable for a mcu crystal drive-level with typical ESR and cap and Vpp values?

Is there a better (simpler) approach?- other than taking measurements with a current-probe which I don't have.

[Manul]

In practice, measuring crystal voltages will result in somewhat unreliable results. It is easier if oscillator has one crystal pin grounded, for eg. colpitts. But normally current probe should be used. You may try to DIY and calibrate by using signal generator at the frequency of interest.

10uW seems ok to me. Practically speaking, if you are worried about startup reliability, you may try adding extra ESR to the crystal (series resistor), to find out at what point it stops working. Then you may estimate how far from "not starting" the circuit is. It should not be "on the edge of working", because part variations would make your design unreliable.