Background:
I have developed a control board for a large, intelligent air conditioning system. Last year, I had 4 control boards manufactured by MyroPCB, and everything came out fine. I just received 12 more of the boards, and am seeing this issue on 6 of the 12 (that is, 6 of the new boards work perfectly fine, just like the original 4).
The board uses a MSP430F5527 microcontroller.
The Problem:
on half of the boards, I load in the program (done with a FET430-UIF), and in less than a minute, the microcontroller stops operating normally. When I halt the debugger (I'm using Code Composer Studio), it shows an error like "No source available for 0x40000" (the address is never really the same, I've seen it at 0xFFFFE, 0x4, etc.), even though this should never be an area that the microcontroller should be in. My program is around 40kB, so it's still well within the lower flash section. The stack size is plenty big, there's more than enough ram, but for some reason the microcontroller is jumping to a random part of the code. The voltage to the chip looks perfect right at 3.3v, and doesn't ever appear to drop. Again, on the other 6 new boards (and 4 old), this has never happened - the current one in the field has been operating nonstop since September of 2014.
The Question:
My real question is if this is a known issue. The markings on the microcontroller are "OCCE4ZTC4 M430F5527 Rev E". Could this somehow have been a bad batch of microcontrollers? Could something like this happen if the manufacturer was soldering at too high of a temperature or something? My debug configuration is set to Full Verification and the program appears to upload just fine without any errors.