Content originally posted in LPCWare by lpcwizz on Thu Apr 03 11:33:51 MST 2014
there are 2 things to your prev post
1) the source follower (old days transistor emitter follower) implementation of the bss138 is a nice way to generically address the step-up issue, however it requires 3 parts resulting in board space, component insertion cost and additional possibilities of failures and it's detection over a single part solution ... only a thought.
the high level drive of a pullup increases the waste of current (power) in the low state over a push pull out stage of a dedicated buffer solution ...
also as you indicated, in the source follower implementation it's not really buffering, since the sink current will be carried through the part to gnd.
also the low level spec of the part pin still apply.
2) general rule for digital I/O pins ... as long as only digital peripherals use a pin, the digital specs of the pin apply, regardless of the digital peripheral selected ... the exceptions are analog I/O peripherals using a gpio pin, which have in most cases different specs, caused by switching or re-routing the pin to the analog peripheral ...
it doesn't really matter in the case of an analog peripheral being connected directly to the pin or the digital pin being disconnected or just being switched in parallel to the pin, if the, for example adc, can only handle the 3.3 V and the digital circuitry. the rule of the lowest (weakest) spec overrules the others