The control loop senses T, and when T drops below internal set value, the
amp stages, typically bias, are turned on. Take my post with a grain of salt in
that the precise behavior of the control loop best commented on by an
actual IC designer.
A circuit is arranged to provide a thermal shutdown signal in response to an overtemperature condition. The thermal shutdown circuit is typically implemented in the same die as the circuitry that is t
LDOs will shutdown when Tj goes above 125 C, and restart only when Tj falls below approx. 100 C regardless of the application of power. Normally a short circuit at the LDO output will cause the device to overheat, and power down is required to allow the device to cool down. Repeated thermal shutdown events will stress the part and reduce reliability.
You can have real time or you can have latched overtemp
protection (see similarly in overcurrent / short circuit
protection). Depends on what you want the fault response
recovery to be. Systems that operate away from "intervention"
will probably want a real time protection that resets when
conditions are back in range. Consumer systems that
have significant tort exposure, probably are best off with a
reset that requires power cycle, to put the responsibility
squarely on Somebody Else. Easier to pass a UL flammability
test if the "heater" quits after the first trip, than "motorboating"
until the next domino does its thing.