BlackST wrote:
A corrupted eeprom that will be fixed in 2 minutes or even less if someone else does it for you. Can you ?
Everything you do seems to require only 2 minutes. Yet when I asked you to perform a simple measurememt with a multimeter, I waited for 2 weeks while you consulted with colleagues, and in the end I had to show you how to do it. So you'll have to excuse me if I don't believe any of your self indulgent stronzate. I'm not impressed with chest beating, nor by people who blow their own trumpets. I don't care who you are, or who you think you are -- if you make a claim, then be prepared to back it up if someone questions or challenges it.
As for how I would test the idea that a corrupted NVRAM, with a bad checksum, would prevent the motor from spinning up, I suggest you make two copies of the NVRAM, one with PUIS enabled, and the other with it disabled. Compare the two images. I expect that there would a difference in one particular bit corresponding to the PUIS flag, plus a second difference either at the start or at the end, corresponding to the checksum. I suggest you then take the PUIS-enabled image (which prevents the drive from spinning up), and then flip the PUIS flag, while leaving the checksum untouched. This will corrupt the NVRAM without corrupting the data. If the MCU executes the code without regard to the checksum, then it should spin up the drive. If it examines the checksum, then the drive won't spin.
As for the suggestion that it is possible to corrupt the data without invalidating the checksum, I have read that in some drives the MCU will rewrite the NVRAM if the board is swapped to a different HDA.