On Monday, February 17, 2014 12:31:40 PM UTC-6, David Riley wrote:
you don't want an output cap that's too big on a 78xx part, because when you cut off the power supply, you could end up reverse-biasing the regulator
Actually I've pulled the datasheet and studied that very problem in the past and categorically the entire 7800 series specifically has an internal protection circuit for that exact problem which is good for up to 7 volts on the output. So shove less than 7 volts on the output, short the input, and nothing happens, by design, for any reg in the 7800 family. Some folks consider it a huge design bug in the 7800 series because they designed and shipped a family of regs with outputs up to 15 volts when they know the family protection ckt is only good to 7 volts out. So a 7805 is bulletproof, but a 7812 not so, by design, which sucks.
Anyway for S100 work its very difficult to reverse bias a 5 volt reg to more than 7 volts (can be done by interference, power spikes, lightning, drive an inductive load and flip it on and off (relays and motors, etc). So yeah you're mostly safe. You might be able to blow out a bare board, but some of the monsters with dozens of chips are drawing amps, and anything that fits on the PCB is going to drop in voltage very quickly. The RC constant for a hundred uF at the equivalent of a couple amps draw is a couple microseconds and the input can't drop that fast (or can it?)
I have blown up 7800-replacement FET based low dropout regs by turning them into expensive little oscillators. Got a nasty burn on the fingertip off one, before it vaporized itself. The classic 7800 bipolar family is better behaved than the new regs and much less picky about caps. Switchers are even more picky about caps.