I'd look at what the resistors feed to- on my IRIS (as far as I can see) a resistor
was smoked because multiple filter cap failures ^ ripple frequency with resultant lowering
of reactive impedance, increasing current flow in this filter stage (between the outputs
of the 5v rectifiers and ground, (.1?f cap -> 330 ohm -> gnd) and blowing the
resistor. (at least that's what it looked like- the cap wasn't shorted and seemed
to work properly under test)