crazy-geeks / stm32-argb-dma Goto Github PK

В процах серии f4xx (проверял конкретно на f407vet, но скорее всего и на других этой серии так же), тактирование таймеров от от шин APB, отличается о тех что указаны в библиотеке. В коде использую таймер 1 с DMA2 CH2.

Для себя сделал следующую коррекцию файла ARGB.c :
/// Timer's RCC Bus
// #if TIM_NUM == 1 || (TIM_NUM >= 8 && TIM_NUM <= 11)
// #define APB1
// #else
// #define APB2
// #endif

/// Timer's RCC Bus (для stm32f407VET) //отредактировал ShadowPRM
#if (TIM_NUM >= 2 && TIM_NUM <= 7) || (TIM_NUM >= 12 && TIM_NUM <= 14)
#define APB1
#else
#define APB2
#endif

Первоисточник: RM0090 Reference manual STM32F405/415, STM32F407/417, STM32F427/437 and STM32F429/439 advanced Arm®-based 32-bit MCUs (с.180).
Дополнительная информация: https://microsin.net/programming/arm/stm32-timers-clocked-from-internal-clock-frequency-apb1-i-apb2.html

Можно прошерстить разные серии процов и написать опцию в ARGB.c (выбор серии проца).

If odd count - data continues to transfer

The datasheet is for SK6812 RGB but the implementation is for SK6812 RGBW.

Hello Crazy-Geeks,

Thank you for this lib. I try to add it to my Nucleo G0B1RE project but I have several problems.

First of all I had to change APBfq init because of "implicit declaration of function" :

Unfortunatlly, I the execution get stuck in the second ARGB_show() call. Whane checking DMA_HANDLE.State it is HAL_DMA_STATE_BUSY.

Please find my configs :

I use a picoscope for debugging which is directly connected to the TIM4_CHANNEL1 pin and the it stays low.

Do you think the problem comes from the APBfq changes ? Could you please help me find out where does the problem comes from ?

Thank you for your time.

antoi-q

Hi there,

Thanks for all the work in putting this together, very clear code & instructions. I am trying to use this lib on an STM32F411 (Nucleo STM32F411RE) and the output signal is incorrect:

This is the signal supposed to be sent for to clear a strip of one pixel:

ARGB_Init();  // Initialization

  ARGB_Clear(); // Clear strip
  while (ARGB_Show() != ARGB_OK); // Update - Option 1

  //ARGB_SetBrightness(100);  // Set global brightness to 40%

  //ARGB_SetRGB(2, 0, 255, 0); // Set LED №3 with 255 Green
  //while (!ARGB_Show());  // Update - Option 2

  //ARGB_SetHSV(0, 0, 255, 255); // Set LED №1 with Red
  //while (!ARGB_Ready()); // Update - Option 3
  //ARGB_Show();

  //ARGB_FillWhite(230); // Fill all white component with 230
  //while (ARGB_Ready() == ARGB_BUSY); // Update - Option 4
  //ARGB_Show();

  //ARGB_FillRGB(200, 0, 0); // Fill all the strip with Red
  //while (!ARGB_Show());

I have checked all the CubeMX configuration which is as per your instructions and really don't understand what's happening. Any idea?

For reference, here is my clock configuration:

Timer configuration:

RCC configuration:

In this method:

Hi!

I'm trying to run this library on STM32F030K6T6, but I have a trouble with it.
All configs are correctly set, but the frequency is 4 times slower than that supposed to.

Screens:

Of course I added the TIM_NUM 17 to ARGB.c and hard-coded frequency, which should not effect on performance:
void ARGB_Init(void) { u32_t APBfq = 48000000; // Clock freq #ifdef WS2811S

ARGB.h:
`#define WS2812 ///< Family: {WS2811S, WS2811F, WS2812, SK6812}
// WS2811S — RGB, 400kHz;
// WS2811F — RGB, 800kHz;
// WS2812 — GRB, 800kHz;
// SK6812 — RGBW, 800kHz

#define NUM_PIXELS 100 ///< Pixel quantity

#define USE_GAMMA_CORRECTION 1 ///< Gamma-correction should fix red&green, try for yourself

#define TIM_NUM 17 ///< Timer number
#define TIM_CH TIM_CHANNEL_1 ///< Timer's PWM channel
#define DMA_HANDLE hdma_tim17_ch1_up ///< DMA Channel
#define DMA_SIZE_WORD ///< DMA Memory Data Width: {.._BYTE, .._HWORD, .._WORD}`

Fix info about buffer width in CubeMX
Compile Doxygen page

where to get

#include "stm32_dma.h"
#include "pwm.h"
#include "math.h"

??

this is not a part of HAL package

Now strip works only when APB frequency is 72 MHz. I tried to set it to 8 MHz and change counter to 9 (8x10^6/8x10^5 = 10), but strip didn't work at all. In several weeks I'll buy logic analizator to watch what happens at GPIO pin.