When I was working at FMC, designing controllers for naval missile
launchers and gun mounts, we made RS-232 drivers/receivers using
discretes, that really WERE short-proof. ;-)
Cheers!
Rich
How the chip handles short circuits depends on how the power supply
rails are created. For chips with on-board charge pumps or switchers,
the power supply rails are stiff enough to do the job (i.e. 232 and
drive serial mice), but not all that stiff should the output be
shorted. It is a cheesy protection scheme. For the chips I've
designed, the short circuit conditions were considered "indefinite".
Thus you could short it and it would survive, but a short for 24&7
until the cows come home would be a different story.
Using discretes doesn't necessarily save your ass. For instance, if
the power supply rails are stiff, the drivers could get some serious
current. It's been my experience that most discrete designs are not as
rugged as a well designed chip. This is because often chips have
protection circuits not shown in the datasheet. This is to take care
of screw ups in the design-in phase. That is, if you short some pins
that in real life won't get shorted but might get shorted by a clumsy
scope probe AND the chip fails, the product won't get designed in. The
customer will think the part is flaky. Such short circuit protection
is brute force, i.e. set high enough that it will never effect
performance of the part, and often high enough that the short can't be
left there infinitely.
There is a reason most datasheets only show simplified schematics or
just a block diagram. There is much not being shown.