I have an issue with TLV707285 which is used to charge an ultracapacitor in the following schematics:
+3.3V -> TLV707 -> Ultracap (anode) -> Diode ->| -> [Microcontroller supply] <- |<- (diode) +3.3V
The microcontroller gets supply either from 3.3V directly via diode or from the ultracapacitor via diode when no 3.3V is present.
The microcontroller also drives directly the EN pin of the TLV707285 to control charging of the ultracapacitor.
The problem is that _sometimes_, when the ultracapacitor has discharged to certain level (the TLV707 has output voltage present, but no input voltage) and then the input 3.3V is applied again, the TLV707 goes to a strange state where it seems to not regulate it's output voltage (~3.3V is present at its output) and a large current (over 200mA) flows through TLV707285. When this happens I could track that only the TLV707285 heats up on the PCB which hints that the excess current is consumed only by the TLV707285 itself. The output of TLV707285 is isolated from the rest of the schematic by diodes and the ultracapacitor, even when presented with higher than the rated 2.83V, will not cause such high current drain.
As mentioned, this situation happens only sometimes and it is very difficult to replicate. This situation typically happens when the ultracapacitor voltage dips under, or near, the microcontroller lowest operating voltage which may cause the TLV707285 EN signal to transition into tristate, toggle or stay in particular state. At the same time micocontroller seems to operate correctly. Regardless, the TLV707285 will not come out of this state by toggling the EN pin low/high, etc - it only comes out of this by powering off and on again the input supply.
Is it possible, theoretically, for TLV707285 to enter into such faulty state and how to prevent this from happening?