We experience a serious issue with few pieces of the MVF61NS151CMK50/2N02G microprocessor. Currently I have one of them on my table, soldered on a board. The problem is that it boots highly unreliably. I guess the origin is in the BootROM, as u-boot does not even start to boot (no console output; though the problem is not in the console).
On power-cycle, the MPU sometimes boots, and sometimes it does not. It might be 50/50. It behaves the same when booting from MMC and NAND (set by GPIOs). On boot failure, it does not enter the Serial Boot Mode (at least no USB device connects to the PC).
We discovered one very "interesting", but also crucial thing. Triggering the RESETB pin (T4) does not change anything! I.e. if the MPU starts booting after power-on, triggering the reset does reset the MPU and it boots successfully again (and again and again). If it does not boot after power-on, triggering the reset does not reset the MPU, it never starts booting. Hence it is not even possible to workaround this issue with watchdog!
We watched the voltage levels and signals development (XTAL, RESETB), but did not find any significant difference between the working and broken state. We welcome any advice on what to measure to solve the issue. Anyway, the RESETB observation seems to point on a silicon issue, doesn't it?
Is this a known issue? Is there a way to fix, or do we need to replace the MPU?