Problem with "official" LCD software example on STM32F4 Discovery Expansion Board

So I too was having issues with the code provided by Embest so I wrote some of my own. You can pull a copy from Github here:

https://github.com/ChuckM/bb-lcd This puts up a simple display on the LCD with a running timer (counting up seconds and milliseconds)

A screenshot of what it looks like when it is running is here: https://github.com/ChuckM/bb-lcd/blob/master/screen-shot.png

I've noticed a couple of things, first that the viewing angle on this display is quite poor, the second is that since the LCD isn't actually memory

mapped it would be really really hard to get it to update without flickering. If you run the demo it animates moving the RED/GREEN/BLUE/MULTI

blocks around every 10 seconds.

Also a shout out to the AdaFruit folks for their GFX library, it let me get something up and running quickly where it was more than just draw a

line or draw a box. Originally the demo drew a clock face and animated the hands but that was really hard to watch with the flickering.

--Chuck

if you want fast updates, you must use write to internal ram function. Idea is simple, you send special commands to define area , and start send raw data. It's like paint bmp function. i can get up to ~400fps with this modification on SSD1963 480*272 lcd.

void LCD_WriteBMP(uint16_t Xpos, uint16_t Ypos, uint16_t Height, uint16_t Width, uint8_t *bitmap)

{

  u32 index;

  u32 size = Height * Width;

  u16 *bitmap_ptr = (u16 *)bitmap;

  bitmap_ptr+=Height * Width-1;

  LCD_WriteCom(0x0036);

  LCD_WriteRAM(0x0040);

  LCD_StartPixelLineAddress(Xpos,Ypos,Xpos+Height-1,Ypos+Width);

  for(index=0;index<(size);index++)

  {

    LCD_RAM = *bitmap_ptr--;

  }

  LCD_WriteCom(0x0036);

  LCD_WriteRAM(0x0000);

}

Problem is, you need to have frame buffer in sram if you want to refresh all screen.But you can optimize all paint functions by adding write directly to internal ram function.

Sorry about being late to reply, in this particular case the Embest board uses the SSD2119 (not the 1963) however there is a similar sort of mode where you can set up the address counter internally and then repeatedly write to the 'Data' address and have it autoincrement. However, in this particular case the board is implemented poorly, it does not take full advantage of the STM32F4 FSMC controller, and doesn't wire up either the WAIT line or explicit RD/WR lines, (in part because the chip used doesn't have all FSMC pins due to the low pin count). Bottom line is that writing the frame cannot recognize a 'slow' write (as you get durring a DRAM refresh in the controller chip) so the software sets up delays in the DATA hold to insure that all the writes "work" (but even then you can over run the chip if you aren't careful.) The display works more reliably if you wire it up to the SPI port. That has its own limitations of course but at least you can then wire up the 'sync' pin to know when it is displaying and trying to do beam following (well display following as there is no "beam" :-).

Sorry about being late to reply, in this particular case the Embest board uses the SSD2119 (not the 1963) however there is a similar sort of mode where you can set up the address counter internally and then repeatedly write to the 'Data' address and have it autoincrement. However, in this particular case the board is implemented poorly, it does not take full advantage of the STM32F4 FSMC controller, and doesn't wire up either the WAIT line or explicit RD/WR lines, (in part because the chip used doesn't have all FSMC pins due to the low pin count). Bottom line is that writing the frame cannot recognize a 'slow' write (as you get durring a DRAM refresh in the controller chip) so the software sets up delays in the DATA hold to insure that all the writes "work" (but even then you can over run the chip if you aren't careful.) The display works more reliably if you wire it up to the SPI port. That has its own limitations of course but at least you can then wire up the 'sync' pin to know when it is displaying and trying to do beam following (well display following as there is no "beam" :-).