Greetings everybody.
I write here to ask if anyone can help me with an issue that happened with a TMS3206727 that we use in one of ours products (in a custom board).
This product is on the market and it was installed in one of our clients in 2010. The single core DSP firmware installed on the product was generated using the Code Composer 3.3.79.5, the Bios 5.31.02 and the compiler 6.0.8 (Int. Devel. 5.98.0.364). In 2012 we developed a new product that makes uses of an OMAP L138. So at that time an effort was done to see if it was possible to use the same set of the TI tools with both the chips. After the success of both the products tests procedures we started using the Code Composer 3.3.83.19, the BIOS 5.41.00.06 and the compiler 6.1.9 (Int. Devel. 5.98.0.419) with both chips. In June of this year, after testing the firmware, the units that were installed in 2010 suffered a DSP firmware update, generated by the latest TI tools that I mentioned earlier. During the update procedure already on the client facilities the firmware of the dsp started to hang (followed by a reset forced by the watchdog) after some dozens of minutes of correct operation. We tried to replicate the problem in lab and it did not happened. So, after some negotiations with the client we obtain permission to analyze the installed system and we reached the conclusion that if the firmware is generated using the oldest set of tools everything works without problems. If we generate the same code (same project tag) with the newest set of tools the problems of stability start to manifest.
During the debug procedure (that was implemented only in lab) we detected also that when we use the Jtag xds510 usb with the newest set of tools with the TMS3206727 the connection seems to fail after some inactivity time from the part of the user. This behavior does not happen with oldest set of tools using the same Jtag connection.
The emulador is from Spectrum Digital Inc.. And the driver (sdusb2em.sys) has the version number 2.0.0.880. This driver is used in both versions that i described of the development tools.
All the versions of the product firmware have been generated with the newest set of tools for the last 2 years without problems. The problem only showed it self on a particular instalation and we still cannot reproduce it in our lab (in the lab the system behaves correctly). The conclusion that i gave you earlier was obtained in a week of tests that were done in a substation (client installation) that is already in service.
Because we are trying to replicate this problem in lab for the last two months without any success I decided to post the problem here in the hope that the community had the knowledge of any issue that existed with the use of this set of versions of the development tools over the 6727 that could explain this behavior.
In our project we use the dsp Bios with multithread support and the following hardware is crutial for this project:
-DMAX,MCASP,SPI and UHPI;
The project is also compiled with no optimizations.
I already posted this problem in the Code Composer thread of this forum in :
http://e2e.ti.com/support/development_tools/code_composer_studio/f/81/p/370831/1304987.aspx#1304987
And Ki-Soo Lee kindly replied and gave me the advice to post it also here.
Thank you in advance for your help.
With my best regards.
Rafael Rodrigues.
