 Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me. |
Forum Index : Microcontroller and PC projects : MM2: High res SSD1963 displays revisited
| Author | Message | ||||
| matherp Guru  Joined: 11/12/2012 Location: United KingdomPosts: 8603 |
See the Youtube video for an idea of the performance achieved. With the new facilities Geoff has included in 4.7b8 it is possible to make significant improvements to the code to drive displays other than the ones directly supported. I must also thank Peter Carnegie (G8JCF) who is creating, testing, and documenting the mechanisms to call internal Micromite firmware functions from CFunctions. Without his work this development wouldn't have been possible. In this case, I have updated the code for the 44-pin MX170 to drive the 4.3", 5", and 7" SSD1963 displays. I did more research on the SSD1963 and found a mode in which the colours can be loaded as two 9-bit parallel data elements. This supports RGB666 which gives better colour resolution than the RGB565 used by other displays. The advantage of the 9-bit mode is that the 44-pin MX170 has 10 data pins on port C which do not overlap things like I2C or SPI. These connect to the 9-bit data bus and to the WR pin which is used to load the data on the bus. pin set up allocation is: 'Fixed pins setpin 25,dout 'C0 - connect to DB0 setpin 26,dout 'C1 - connect to DB1 setpin 27,dout 'C2 - connect to DB2 setpin 36,dout 'C3 - connect to DB3 setpin 37,dout 'c4 - connect to DB4 setpin 38,dout 'C5 - connect to DB5 setpin 2,dout 'C6 - connect to DB6 setpin 3,dout 'C7 - connect to DB7 setpin 4,dout 'C8 - connect to DB8 setpin 5,DOUT 'C9 - connect to WR ' Connect RD to VCC ' Connect CS to GND 'Flexible pins Const RS=12 'define the pin to use for RS const Master_reset=32 'define the pin to use for REST setpin RS,DOUT SETPIN Master_reset,dout Note the display only uses 12 pins on the processor. Pins DB9-DB15 on the display can be left unconnected. The code now uses the same in-built fonts as the directly supported displays and uses the same 24-bit colour codes (e.g. rgb(red)) All of the facilities are included in the demo code shown in the youtube video 2015-06-19_133829_ssd1963-44.zip If you have used my display code before you will see that the new version is considerably smaller 
The code has been tested on the 4.3" (480 x 272) and 5" (800 x 480) displays. I haven't got a 7" display. You will see in the code there are a number of "IF" statements that set up the display size. These can of course be simplified for a single size display. I'm very pleased with this version - it allows the use of the larger displays on the MX170 and gives very good performance. The C source for the Cfunctions is attached: 2015-06-19_135537_SSD1963.zip The code only runs on 4.7b8 - do not use b11!!!! |
||||
| matherp Guru Joined: 11/12/2012 Location: United KingdomPosts: 8603 |
Here is a version that has completely flexible pin assignment allowing use on the 28-pin MX170. Of course it is quite a bit slower but still perfectly usable for many applications. You specify the pin allocation in the top of the program: ' ' Define the pins to use for all connections to the display ' CONST RS=12 ' define the pin to use for RS CONST WR=5 ' etc CONST DB0=25 CONST DB1=26 CONST DB2=27 CONST DB3=36 CONST DB4=37 CONST DB5=38 CONST DB6=2 CONST DB7=3 CONST DB8=4 CONST Master_reset=32 'define the pin to use for REST ' Connect RD to VCC ' Connect CS to GND ' The program will then take care of everything else to do with setting up and using the pins. 2015-06-19_173957_ssd1963flex.zip |
||||
| nibbler Newbie  Joined: 18/04/2013 Location: CanadaPosts: 11 |
matherp - This is very interesting. As of today I have been waiting 44 days for my 2.4" screen to arrive here in Canada and have just about given up on it. Now I'm thinking of buying a larger one but might wait to see what further developments you or others make especially with the 7" screen. Alan |
||||
Zonker Guru Joined: 18/08/2012 Location: United StatesPosts: 761 |
Wow... Sweet Pete..! I have the 7" version... I think I have a 44 pinner sitting on an unfinished LCD-uMite Board somewhere... Will start gathering parts... (i think)  |
||||
bigmik Guru Joined: 20/06/2011 Location: AustraliaPosts: 2870 |
Perter(s), This is wonderful... There is no stopping you blokes... Is it possible to adapt to the 28pin '170? I realise that the there would be only 9 pins left over but as a `BackPack' display option they may still be enough for many applications and if we can keep the I2C lines free then the board can be an intelligent display and receive commands from a main Mite/cpu that handles all/most IO.. Love these TFT displays.. 
Regards, Mick Mick's uMite Stuff can be found >>> HERE (Kindly hosted by Dontronics) <<< |
||||
| matherp Guru Joined: 11/12/2012 Location: United KingdomPosts: 8603 |
The second version (SSD1963flex) is for the 28-pin or 44-pin. It allows you to specify each pin individually so you can leave both the i2c and SPI pins free. Here is a slight update with some speed improvements and better comments. 2015-06-20_073815_ssd1963flex.zip |
||||
| Chris Roper Senior Member  Joined: 19/05/2015 Location: South AfricaPosts: 280 |
or use the MCP23017 16-Bit I/O Expander on I2C and use the same MicroMite for both functions, display and the extra 16 I/O lines. Much cheaper and the MCP23017 works well, it even handles interrupts on the individual pins. Cheers Chris http://caroper.blogspot.com/ |
||||
| matherp Guru Joined: 11/12/2012 Location: United KingdomPosts: 8603 |
I2C is much too slow. A complete refresh (CLR) of a 800*480 screen takes 800*480*18 bits + toggling the WR line after each 9-bits (or 16 if RGB565) Just transferring the bits to I2C at 400KHz would take 17 seconds not including any protocol exchange. A pair of 74HC595 on SPI at 10MHz is usable - I've done this before but it is still much slower than direct drive parallel For reference a complete screen refresh of a 480x272 display on the 48MHz 44-pin using 9-bit parallel takes 27msec |
||||
| Chris Roper Senior Member Joined: 19/05/2015 Location: South AfricaPosts: 280 |
I meant use the MicroMite to drive the Display directly and the Port Expander to replace the I/O lost. I agree driving displays via a port expander is fine for a 2x16 LCD text display but no good for graphics. If you did need SPI though the MCP23S17 can run at 10MHz. Personally I have never needed anything faster than the MCP23017, in my projects, for general I/O use. http://caroper.blogspot.com/ |
||||
| bigfix Senior Member Joined: 20/02/2014 Location: AustriaPosts: 124 |
Is it also possible to use the XPT2046 Type Touchscreen controller ? (which some of the SSD1963 displays have included i.e. ebay item 281353506869) I guess touch is a completely independent function from this new mode of display control anyhow |
||||
| matherp Guru Joined: 11/12/2012 Location: United KingdomPosts: 8603 |
Of course
Exactly so. I've appended the test routines using the touch code from this thread modified to use the Cfunction SSD1963 drawing routines to the standard graphics demo. The first version is for the 44-pin MX170 running 4.7b8 and uses dedicated pins for the SSD1963 databus to get maximum performance. This also now includes a demo of the scrolling capability. 2015-06-23_122334_ssd1963-44-touch.zip The second version will run on either the 28-pin or 44-pin Micromite (still needs 4.7b8) and allows completely flexible pin use (except of course for the SPI pins used to access the touch controller) at the expense of some performance. The flexible version does not support scrolling so this part of the demo is missing 2015-06-23_122435_ssd1963-flex-touch.zip You will see that the touch calibration routine outputs 6 numbers XSAMPLE0 YSAMPLE0 XLEFT0 XRIGHT0 XTOP0 XBOTTOM0 rather than run the calibration routine each time, these can be set as constants for any given display and the calibration routine then isn't needed. |
||||
| bigfix Senior Member Joined: 20/02/2014 Location: AustriaPosts: 124 |
Thank you so much for this full solution package ! I bought two of the original TFT Maximites last year at Segor in Berlin But with all the recent additions from Geoff & you they got somewhat "stale" now I do not expect any more updates to the TFT Maximite - so its time to switch... |
||||
| matherp Guru Joined: 11/12/2012 Location: United KingdomPosts: 8603 |
Please find attached an updated version. I have now combined the flexible pin allocation and the optimised version. In addition the scrolling function is now available in all screen orientations. This allows you to scroll in any direction by setting the orientation before calling the scroll routine and then setting it back if/as required. The software will automatically use the optimisations if the pin allocation is suitable. The scrolling function is only available with the optimised pin allocation and RD connected to a pin on the Micromite. This snippet of the program header hopefully explains this. The 28-pin part cannot be configured for optimised operation or scrolling. However the update rate on the 28-pin is still fast enough for most applications even with a 800x480 display. 2015-06-24_122431_ssd1963-flex-touch-scroll.zip The C code is attached if anyone wants to look and or modify it 2015-06-24_123206_SSD1963.zip |
||||
 Print this page |
