Well all the four MCP1603 ADJI that I have are dead, I don't know why, all of it release its magic smoke, maybe it does not support change of bottom resistor on the feedback voltage divisor on the fly or maybe I was wrong with something, so suspecting of my voltage level logic converter circuit (based on this) I recycled a 3.3v LDO voltage regulator, desoldered the last failed MCP1603 from the board and performed tests and the five bi direccional logic level converters worked as expected!!!.

I raised a ticket on microchip about this, I want to know what happened here, if was my fault then why the LDO is not getting hot while I use it? time will tell.

The next image show the AccessB without the MCP1603 connected to the LDO