STM32H750 Discovery kit Clock Generation on GPIO Pin @ 10Mhz speed using DMA (Memory to peripheral copy).Clock is not coming properly and it is ringing

[Andrew Neil]

@Community member​ "Are you driving a print head?"

@Community member​ "Do you have a Crystal Cube? :)"

Did you mean crystal ball ?

Anyhow, printhead may be inspired by:

https://community.st.com/s/question/0D53W00001dw96OSAQ/is-there-any-board-which-can-support-more-than-30-mhz-toggle-frequency-on-gpio-and-parallel-processing

@Rajesh Kannan​  - you should review that thread as it contains some (potentially) relevant suggestions - whether or not you're actually working towards a printhead...

[Tesla DeLorean]

As I think we've previously established I have a tesseract and I am an elephant..

[waclawek.jan]

> Did you mean crystal ball ?

No. It's all Cubes when it comes to STM32, said ST.

The flats throw nasty reflections and the edges get into way of sight, but hey they are sooo easy to stack onto each other!

Jan

[Tesla DeLorean]

The TIM's could generate a couple of outputs, for broader patterns I'd look at other things.

The TIM could modulate streams into LED daisy chains.

Things like FIFO memories can be use for capture or output problems where time domain accuracy is important/critical and burst bus access at optimal rates can sustain. ie some disparity in rates where you need some elasticity, or change widths, etc.

Yes the H7 GPIO is particularly quirky

[AScha.3]

just i was not aware, H7 could be slow on port access, i try : have DevEBoard here, for my audio player fun project...

simple 1. test: set port by HAL call (slow...maybe. i know. )

now H743 at 250MHz clk , caches on, -Ofast , port speed set: low ->

so 50ns with HAL. next direct write ..

[AScha.3]

direct write about same : 50ns lo, 50ns hi.

then

optimzed clock tree, to get (maximum) 200MHz on AHB4 , where ports are.

and direct write >> GPIOA->BSRR = 0x00000002;

 so 30 ns seem optimum with H743. :|

just for fun: test write port lo-hi-loop on stm32F303 (M4 core) - Ofast :

about 15ns , at 72MHz clk. =)

(1 clk 13,8ns at 72 MHz)

and while-loop is 100ns total, because core much slower than H7

same on F303 without optimizer , -O0

 about 340 ns !!

and funny, with -O2 faster than -Ofast :

about 83ns , 1 tic faster than fast. ;)

so optimizer -O2 speed up 400% in this small loop. :astonished_face: