STR912 ARM with SSD1906 TFT controller
in [Embedded]
|
Prev: ATSTK525
Next: I2C Firmware Code for TI MSP430F2013
From: Spen on 15 Jan 2007 16:32 kikuzo wrote: > Hi Spen, thanks for the tips..!! > > I see that you used a Generic #1 Bus. I had to use #2 because the > board with the SSD1906 was set that way. > > I also tried to use the WAIT signal to stop momentarily the STR912 > connecting it to an external interrupt (P5.6, line 14), but I have not > been lucky with that. The interrupt routine never gets called, so I > lowered the bus speed to 40 MHz. With that and the aid of Wait States, > I let the WAIT signal to come up nicely. > Why not use the STR9 EMI to insert wait states, i have the emi running at 48MHz, cpu at 96MHz with no problems at all - using the TX09 at 16bit/pixel. Spen
From: Spen on 15 Jan 2007 16:41 > I visited Ramtex web page, but they don't have a demo library. > That's a shame. I'm starting to write my own library, but graphic > functions are very heavy to deal with. You have to buy the lib from Ramtex - not bad pricing though. On the Open Source front - just managed to get microwindows running, its taken a bit of porting, but seems to be working well. The touchscreen is next. Spen
From: Anton Erasmus on 16 Jan 2007 02:38 On Mon, 15 Jan 2007 21:41:25 +0000, Spen <spen(a)nospam.com> wrote: >> I visited Ramtex web page, but they don't have a demo library. >> That's a shame. I'm starting to write my own library, but graphic >> functions are very heavy to deal with. > >You have to buy the lib from Ramtex - not bad pricing though. Unfortunately for some situations one have to modify the source substantially. It is fine for targets they have tested on, but they do stupid things like: 1. Have a #define PC when selecting non 8-bit mode. This same define then selects little endian as well. If one use a 16-bit or 32-bit target that is big endian everything breaks. 2. Their font arrays must be packed. This is an assumption on their part which is not documented anywhere. If one selects the packed-array option on gcc for an LPC ARM, it breaks the code that uses the defined structures for accessing hardware registers provided by many for their LPC Development boards in conjunction with arm-gcc. This is of course a very clear example why using structures to access hardware registers is a bad idea. 3. The documentation for getting it going on a different target MCU is very incomplete. The various #defines have side effects like in 1 above which is not documented. 4. They have an LCD emulator program for Windows. This is nice in that one can test menus etc. on the PC. Unfortunately it looks like most of the code was developed by PC orientated programmers using this emulator. If Ramtex fixes the #defines with undocumented side effects, and provides better documentation for using on a different target, it would be quite nice. As it is I would stay clear unless one is using it on a target MCU that they themselves have tested it on. >On the Open Source front - just managed to get microwindows running, its >taken a bit of porting, but seems to be working well. The touchscreen is >next. What sort of footprint do you get for microwindows ? How does it compare with the Ramtex footprint ? Regards Anton Erasmus
From: Spen on 16 Jan 2007 13:59 > If Ramtex fixes the #defines with undocumented side effects, and > provides better documentation for using on a different target, it > would be quite nice. As it is I would stay clear unless one is using > it on a target MCU that they themselves have tested it on. > I have had to port to the STR9, but so far so good. > What sort of footprint do you get for microwindows ? How does it > compare with the Ramtex footprint ? Not a fair test at the moment as microwindows is doing a lot more, but: about 25K for Ramtex and 60K for microwindows. Regards Spen
From: Spen on 16 Jan 2007 13:59
> If Ramtex fixes the #defines with undocumented side effects, and > provides better documentation for using on a different target, it > would be quite nice. As it is I would stay clear unless one is using > it on a target MCU that they themselves have tested it on. > I have had to port to the STR9, but so far so good. > What sort of footprint do you get for microwindows ? How does it > compare with the Ramtex footprint ? Not a fair test at the moment as microwindows is doing a lot more, but: about 25K for Ramtex and 60K for microwindows. Regards Spen |