MC9S12XE Emulated EEPROM

[528887301]

MC9S12XE Emulated EEPROM，if the sector erase count more than five times 65534, How will sector header change ？What are the effects of reading and writing the  D-flash?    

[gordondoughman]

If the sector erase count reaches the maximum, it means the DFlash sector(s) associated with the EEE have been erased at least 65,534 time which is beyond the rated specification of 50,000 minimum. So, while typical is stated as 10x this number, reads of the data (in EEE RAM) stored in the Emulated EE and writing new data to the Emulated EE would not be guaranteed.

Regards,

Gordon

[danielmartynek]

Hi,

The ECOUNT value for the sector does not roll over, it will stop incrementing when it reaches the value 65534. It will not affect reading, writing to the D-flash.  

Regards,

Daniel

[528887301]

Hi, Daniel :

        Thank you for your reply!

         Reference manual  on Table A-20 mentioned D-Falsh erasing times, the minimum value is 50k(Under the conditions mentioned in the data sheet, The same below). Typical values are 500k. When the typical value is exceeded, the possible failure mode is?

         A product of customer using D-flash EEE, used to save important data. Appeared in two cases ：from the D-flash to read back the data （The read data is 0XFF） are different from written data, and Sector header is 65534 . Products have been used for some time, and  has been no problem before. Why?

[danielmartynek]

Hi,

Regarding the NVM Reliability Characteristics, minimum is 50k (-40°C < junction temperature < 150°C), which guarantees that the data are correct. Typical 500k, but there is a note: average junction temperature does not exceed 85°C in a typical temperature profile over the lifetime of a consumer, industrial or automotive application.

Moreover, it depends on the partitioning of D-Flash, as you can see in Figure A-2 in the RM.

Regards,

Daniel

[gordondoughman]

Daniel,

Wen,

While the total number of writes to the EEE RAM buffer is dependent on the DFlash to RAM buffer ratio, the minimum guaranteed number of erase cycles for any given sector is still 50,000. The trouble with using the EEE beyond a point where the erased sector count reaches its maximum, is you really have no idea how many times past the minimum or even typical specification the sectors have been erased. Unless of course you can calculate it based on the software doing regular periodic writes.

The Flash in the 180nm devices is very robust and in most cases can be used way beyond the specified minimum, as indicated in the typical specification. However, at some point, the sectors will begin to fail as indicated by the dead sector count.

Before I retired from NXP, I was involved with an automotive customer that had a bug in their software that wrote new data to the EEE RAM buffer every 10 mS while the car was running. It in fact took years and an estimated millions of erase cycles before problems began to show up, however, the EEE still began to fail.

The minimum erase cycle specifications are set based on voltage, temperature and process variations. While fab processes are generally very well controlled, they can at times drift to "corner " or worse cases where the devices will still meet the minimum specifications, but may not be quite as robust and I'm not saying this is the case here.

So, bottom line is, you can use a device to the typical specifications, but you shouldn't.

Regards,

Gordon