We have a system with a Buck regulator with a 3.3V output driving several paralleled ADP151's each feeding separate sections of circuitry to act as a low noise supply. Each ADP151 is fed through a ferrite bead and bypass capacitor. We are having problems with the ADP151's failing in a mode in which the output is still in regulation but the input draws much more current than is supplied to the output. When the failure occurs any one of the parallel regulators may fail, but the rest continue to work normally (we can use a thermal camera to find the part that is getting hot). My initial thought was that the buck regulator is producing a transient voltage sometimes that exceeds the absolute maximum input voltage and that it is enough to destroy some parts but not others. Unfortunately the failure occurs infrequently enough that it is difficult to reproduce. I monitored the output voltage of the buck regulator and never saw any bad waveforms across multiple boards - of course none of these boards produced ADP151 failures when I was monitoring them. We built a pilot run of this design and saw this happen once but did not properly identify the failed part at this time. This is happening much more frequently in our subsequent (but still low-volume) production run.
Thinking that it is not an input over-voltage problem I started to consider what I have seen in other LDO data sheets warning against output voltages higher than the input voltage. The ADP151 data sheet makes no mention of this issue, but it does use a PMOS pass transistor, presumably with a body transistor that can get turned on. Do the symptoms I describe (IC still regulating but drawing excessive input current) make sense with a failure caused by this kind of condition? Is the ADP151 known to be susceptible to problems with reverse voltage due to output caps being charged up while the input voltage is removed?
I would appreciate any help as this is a major problem in our production at this time.