??? 04/18/06 19:22 Read: times |
#114475 - Not necessarily ... Responding to: ???'s previous message |
Erik, Richards,
11 July, 2005 Maxim/Dallas issued a Customer Alert Notice regarding Flash corruption. Info regarding this is copied below. The issue is not characteristic of the DS'420-B5 or the DS'430/DS'450-A5 which are errata free chip steps. regards, p Subject: DS89C4x0 Flash Memory Corruption Problem Description: Maxim / Dallas Semiconductor's DS89C430, DS89C440, DS89C450, and certain versions of its DS89C420 Ultra-High-Speed Microcontrollers have exhibited symptoms of flash memory corruption during a power-on cycle. When this issue occurs, portions of the code memory and/or internal device configuration may be lost. Upon investigation, this may manifest itself as: 1) Erased/corrupted program memory, noticeable during program operation or when a verification function is performed using the internal ROM loader. 2) An incorrect device number, displayed in the internal ROM loader sign-on banner. For example, a DS89C430 reports itself to be a DS89C450 in the ROM sign-on banner. 3) Erasure of previously programmed lock bits. 25% of parts evaluated exhibited corruption independent of supply ramp rate and at temperatures ranging from 53°C to 77°C, but there is no set of conditions under which this issue is guaranteed not to occur. In addition, no screening process has been identified. Root Cause & Corrective Action: During a power up sequence, the signals driving the flash erase interface were left un-initialized. If these signals power up in a certain state, they cause the flash to perform a random erase cycle, ultimately corrupting flash. The design fix implemented on the devices is to force the flash interface signals to a known state during power up, removing any potential from this type of corruption. The corrected version of the DS89C420 will be marked revision B5 and the DS89C430, DS89C440 and DS89C450 will be marked revision A5 |