Posted on November 02, 2015 at 17:08I remember from the 1980s that my colleague responsible for the hardware had some trouble getting an oscillator for our MCU to work. But the last decade designing my own hardware, it seemed as if things had improved. All sorts of complicated mumbojumbo in the datasheets, but just plug a crystal on the board, two load capacitors (but not always populated), and things would work.
But now with STM32 it seems things are more complicated again. My standard crystal with two capacitors (now properly mounted) does not work.
I have found AN2867. It tells me to check Gmcrit as for the STM32F072 the Gm is specified. I tried following the calculation for MY crystal, and ended up with.... Exactly the same number as the example in AN2867: 0.23mA/V: The reason is that my CL1 = CL2 = 20pF (I think I mounted 18pF, is that ok?) and I accepted the AN2678 estimate of 7pF of trace capacitance. For my '072 with Gm of 10mA/V that gives a margin of more than 40. That should be enough, right?
I'm have a bunch of crystals where I don't have a proper datasheet of. And one of those is probably mounted. I'll mount this one tomorrow:
http://nl.farnell.com/abracon/abls-8-000mhz-b4-t/crystal-8mhz-18pf-hc-49us/dp/2467757
to see if I can get that to work.
But given that this one is supposed to work (it's the first and cheapest one that fits my board), would it be possible to chose wrong, and find a crystal that does not work?
Say: which ones of ''search for 8MHz crystals on farnell.com, select HC49, SMD''.
are unsuitable for use with an STM32F072?
