MM9Z1_638 Produces Erroneous Data

[chrisescalante]

I am working with the demo software provided by Freescale for KIT9Z1J638EVM, which has a battery sensor that measures current, voltage, temperature.  Every so many clock cycles there is erroneous values displayed.  When I observe the memory location where the measurements are being stored, I can see cycles where there are 5 to 10 memory locations being overwritten with a "EE" or "FF", and in some instances it is writing over memory location used to store the battery sensor data.  When I add a watchpoint or breakpoint, the problem stops which is puzzling and as soon as I remove all breakpoints the EE's and FF's re-appear.  This makes it difficult to debug.  How do I fix this so the sensor data is accurate?

[reyes]

Hi Chris Escalante,

I checked with the applications engineers and they mentioned that they are already in contact with you about this case.

Basically they mentioned that they reviewed your SW, and confirmed that it is not compilable, so it cannot be the one you use for testing.

They also mentioned that the most likely issue is caused because you have code fragments from FreeMaster and PE in the project. As he explained “the PE stuff uses an SCI routine “On-Top” of the already existing SCI routines”.

In case you have further questions regarding this case, I would recommend you to keep in contact with the applications engineers directly.

Have a great day,
Jose Reyes

-----------------------------------------------------------------------------------------------------------------------
Note: If this post answers your question, please click the Correct Answer button. Thank you!
-----------------------------------------------------------------------------------------------------------------------

[chrisescalante]

Hi Jose,

Yes, I am working with the application engineer, but I also tested the last software he emailed me that does not contain the added PE and SCI code as well as the demo software available for download on the Freescale website and they both produce the same erroneous data at random clock cycles.  I provided screen shots of this to prove the issue.  Perhaps it is associated with the modifications we made to KIT9Z1J638EVM to get it to interface with 24 volt batteries, but I still believe it is a software bug.

Regards,

Chris