@YDOVG.1 :
I have a PCB where only MCU is mounted, nothing else.
What power-supply decoupling do you have?
Have you wired up all Vdd and Vss lines? With a 0.1 uF decoupling capacitor within 5 mm of each pair of pins?
The "hat" moves closer to the edge when I decrease frequency.
By that, do you mean as a proportion of the period? In other words, is the "hat" a fixed time-duration from the signal edge?
In that case what you are seeing looks much like the effects of signal reflection from an incorrectly-terminated transmission-line.
I know you don't intentionally regard your SPI or other signal traces as transmission-lines. But at high-enough frequencies, everything is.
Now the "hat" occurs at about half a division of your 'scope screen at 20 ns / division. So that's roughly 10 ns. The speed of light is 3x10^8 m/s, so we're talking of a round-trip distance of 3m, so it looks like an unterminated wire length of 1.5 m.
What might be around 1.5 m long?
It could be the length of your 'scope probe lead (in which case maybe the hat isn't present when the 'scope probe isn't).
Or (if you don't have power-supply-decoupling capacitors within a few mm of the stm32) it could be the wires supplying power to your pcb from your good-quality bench power-supply.
Have you tried setting the GPIO pin speed slower (so the rise is less sharp); if so does the "hat" reduce in height?
Hope this helps you identify the source of this strange waveform,
Danish
