I think what is happening here is the power supply voltages are climbing up until it triggers the over-voltage crowbar which shuts down the supply and it repeats. Are there any more questionable solder joints on the analog board? I don't have one of these to look at, but I would look at the feedback loop that regulates the supply. Usually there is a voltage divider formed by a couple of resistors hanging off one of the outputs, usually 5V. The output from that is fed back to the control side through an opto-isolator. Something is breaking that feedback loop and making the regulator think the output is low.