Adc dma support

[xC0000005]

This adds support for the ADC in DMA mode (I tested by soldering & connecting thermisters/pull up resistors). Longer term, it would make sense for the ADC read functions to be macros that pull the data from the right index in the DMA buffer since the ADC runs in continuous mode and the buffer is always up to date, but this works perfectly well for now.

[victorpv]

Great, let me merge this one right away since it doesn't conflict with anything. I will try to test this later too.

[victorpv]

What is this line for?
USBSerial SerialUSB;

The SerialUSB should normally be automatically declared as "Serial" if you select the option to include SerialUSB in the board settings. I haven't had to add that line to get communication using USB serial, did you face any problem?

[xC0000005]

I thought I deleted it before I committed (I have "Do not commit" markers around other problems unique to my build. I'm sorry.

[victorpv]

Ok I just tried to resolve all the conflicts and merge your changes and the changes upstream from the official Marling. Now I need to try to merge my changes. After that you should probably resync since all together there will be a lot of changes. I'll post once I have uploaded that.

Keep your own configuration.h file, I try to avoid that in the uploads since it's specific for each user, but there is a folder with configuration.h example files for the different boards, I added 1 folder for the stmf1ret, if you want to add a folder to the Malyan board and put the config there, we should be able to submit that as a PR to Marlin and get it merged.
Thinkyhead corrected a lot of formatting in the code, I had to manually edit a lot of files to keep his changes...

[victorpv]

@xC0000005 I am having trouble getting a reading for the TEMP_BED pin. First I had it set to PA0, so I thought perhaps there is a problem with that enum being 0, so I changed to have PA1 as BED, and PA0 as the extruder, and the problem followed to PA1.
So to clarify, PA0 and PA1 work when assigned to the extruder, but both fail when assigned to the BED.
I also have updated the repo with all the latest changes. Before you resync to it, I advise to make a copy of your local folder, I have lost changes more than once because the github client overwrote them when changing to another branch.

[victorpv]

This seems to work fine for me, if you can test.

enum TEMP_PINS
{
  #if HAS_TEMP_0
    TEMP_0,
  #endif
  #if HAS_TEMP_1
    TEMP_1,
  #endif
  #if HAS_TEMP_2
    TEMP_2,
  #endif
  #if HAS_TEMP_3
    TEMP_3,
  #endif
  #if HAS_TEMP_4
    TEMP_4,
  #endif
  #if HAS_TEMP_BED
    TEMP_BED,
  #endif
  #if ENABLED(FILAMENT_WIDTH_SENSOR)
    FILWIDTH
  #endif
};


void HAL_adc_start_conversion (uint8_t adc_pin) {
  TEMP_PINS pin_index;
#if HAS_TEMP_0
  if (adc_pin == TEMP_0_PIN){
      pin_index = TEMP_0;
  }
#endif
#if HAS_TEMP_1
  if (adc_pin == TEMP_1_PIN) {
    pin_index = TEMP_1;
  }
#endif
#if HAS_TEMP_2
  if (adc_pin == TEMP_2_PIN) {
    pin_index = TEMP_2;
  }
#endif
#if HAS_TEMP_3
  if (adc_pin == TEMP_3_PIN) {
    pin_index = TEMP_3;
  }
#endif
#if HAS_TEMP_4
  if (adc_pin == TEMP_4_PIN) {
    pin_index = TEMP_4;
  }
#endif
#if HAS_TEMP_BED
  if (adc_pin == TEMP_BED_PIN) {
    pin_index = TEMP_BED;
  }
#endif
#if ENABLED(FILAMENT_WIDTH_SENSOR)
  if (adc_pin == FILWIDTH_PIN) {
    pin_index = FILWIDTH;
  }
#endif

  HAL_adc_result = (HAL_adc_results[(int)pin_index] >> 2)& 0x3ff; // shift to get 10 bits only.
}

[xC0000005]

I had the enum exactly like that and for some reason, thought this was unnecessary. When I'm home I'll compile, hook up the thermistor and test. Sorry for the problem.

[victorpv]

No problem, let me know if it works for you. I think we still may have to account for a case in which temp_0 is not defined, in case someone defines temp_bed and temp_1 for example. I still need to make more changes around the timers so the channel are picked automatically depending which are not used for pwm and exist in that mcu version, haven't had time for that yet. I forgot to check, what speed is the adc sampling at the moment? Get Outlook for Android<https://aka.ms/ghei36>

…

________________________________ From: xC0000005 <[email protected]> Sent: Wednesday, October 4, 2017 8:21:45 PM To: victorpv/Marlin Cc: victorpv; State change Subject: Re: [victorpv/Marlin] Adc dma support (#2) I had the enum exactly like that and for some reason, thought this was unnecessary. When I'm home I'll compile, hook up the thermistor and test. Sorry for the problem. — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub<#2 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AKphQbG3xUnYHFmyTxo0Dvqita1nxpusks5spC8pgaJpZM4PsKBj>.

[xC0000005]

It's working. Given the way enums work, if someone doesn't have a value defined, they should drop out of the list. The result is that if TEMP_0 was not defined, the zero value of the enumeration (and the first entry in the array) would be that referenced by TEMP_1. Any combination should work in this manner.

[xC0000005]

Ah - I know what's missing. ADC_PIN_COUNT should be the last enum, always present. Then we size adc_pins based on that. I swear I had that working and decided to "simplify." I should have left it in the working state.

[victorpv]

By the way I alse tested i2c eeprom yesterday and works fine.