8MHz Crystal used on Nucleo BOM does not seems to follow AN2867.

[JDe l.2]

Dear all,

I have tried to follow AN2867 for choosing an 8MHz crystal as HSE in the STM32L496.

Doing this:

1. Starting from the crystla included on Nucleo BOM, NX3225GD-8.000M-EXS00A-CG04874, I look for a 12pF crystal like this. Assuming Cstray=10pF, I have CL1=CL2=4pF (similar to 4.3pF used on Nucleo BOM).
2. If we have gm (transconductance) = 1.5mA/V (from STM32L496 datasheet), we get a gmcrit= 1'8mA/V, so transconductance gain is about 0'8, very far from proposed min value (=5), and very var from AN2867 example (107). I guess that proposed crystal has a very high ESR (500R), but I also think that STM32L496ZG transconductance = 1.5mA/V is too smaller.
3. More thant this, if I calculate needed RExt (as showed on figure 25), I get arround 4900R, so transconductance relationships is indeed worst.

I think my calculations are ok (I used an excell and compared data with AN2867 example).

Where I am wrong?

Kind regards,

Juan

[AY.2]

Same thing on G0 eval boards. Seems that the designers don't follow their own ANs.

[AY.2]

Moreover, it's not clear to me why they put Rext on OSC_IN line, not on OSC_OUT one, as AN says.