Hi,I am trying to programm a lm3s5b91 with CCS 5.1.1 and I get the following error message:
CORTEX_M3_0: Flash Programmer: Timed out while writing to Flash memoryCORTEX_M3_0: Trouble Writing Memory Block at 0x0 on Page 0 of Length 0x14a0
I already managed to program that device before. Anyone had the same problem? Could it be caused by the 100 cycles limit on flash writes as described in errata 4.4 ? I used the device for development, and have flashed it around maybe 30 times.
Have you a 2nd board - with identical MCU? Simple "A-B" programming comparison test should reveal if yours is system or specific MCU issue...
I can still programm the chip on the EVK that I use as aJTAG adapter, but not the one on my robot, and I do not have a second board.
Hi Thomas,
Under the "Other Utilities" tab in the LMFlash Programmer there is an option to unlock the debug port. I would suggest you try that and re-attempt the flash programming. The programming count seems a bit low for the errata, but if the unlock does not help, then you should contact your local TI representative (FAE/sales) and RMA the board.
To implement "Unlock" - must not friend Thomas know the "class" of his LM3S5B91 MCU? (somehow this delightful class distinction does not (even now) make it into MCU datasheets...)
Ah, yes good catch CB1.
Thomas, you can find your class information by looking in the processor's datasheet within the DID0 register description, CLASS bitfield. In this case, you should find that to be Tempest.
Greetings Stellaris Mitch-
Plz understand - this reporter/others - know that Stellaris Mitch doesn't make the rules/regulations. Poster Thomas is unable to communicate with his MCU - thus his ability to access any MCU Register is tad impaired!
Have complained loud/long about "MCU Class determination" being "buried" - inconvenient & time/effort eating to find. (there are other ARM vendors - long existing road-blocks cannot be good for sales.) Stellaris team seems nearly "dead in water" to effectively deal with this issue - despite multiple client & poster protests.
Seems that simple cursor "hover" over MCU's part no. should yield class info - quieting unwanted, time-wasting, vector-away efforts...
(we strongly suspect that "class" was never wanted - forced upon you - but management of the issue remains ineffectual...)