With 0.4 sec integration time it is easy to see that the problem is what Rick has referred to as 'shredding'. The amplitude takes one of several states and the phases are not too erratic. This is pretty clearly a DTS problem. There have been reports of 26D misbehaving for about a week.