I have seen a bi-stable phase state, with phases flipping 180 degrees on alternate integrations, on baseline 5-23. It occurs in multiple datasets recently, at all frequency bands. I think it must be a correlator problem. Claire