The design changes being evaluated for the next batch of Milkymist One devices, M1rc4, include a fix that should put an end to the NOR corruption issue for good, without even needing the work-around we have in place now.
Last week, I got a prototype with that change and I let it suffer over the weekend. In total, I let it power cycle 14021 times and checked for problems (byte-wise comparison of the standby partition, CRC of all partitions, and the setting of the lock bits) every ~5000 cycles. I didn't observe any problems. If the NOR corruption issue was still around, with an average time to corruption of 700 cycles (this is at the upper bound of what tests with M1rc3 produced), it would have occurred in this test run with a probability of 99.9999998%. So it's quite likely that the fix did indeed eliminate the problem. The detailed log is here: http://projects.qi-hardware.com/index.php/p/wernermisc/source/tree/master/m1rc3/norruption/4/LOG Next in line for testing: the USB power switch with current limiter. - Werner _______________________________________________ http://lists.milkymist.org/listinfo.cgi/devel-milkymist.org IRC: #milkymist@Freenode
