I think you're asking to basically shoot yourself in the foot...

Quote Originally Posted by kevj View Post
The data sheet specs 25 mA as the max current sourced by a pin. I'd like to source 75ma to 100ma, but for relatively short duration - say 20 uS max - in short groups of pulses (maybe 20 to 30 pulses together) followed by several seconds to several minutes of no pulses.
What if the PIC locks up for some silly reason?

That's probably technically out of spec, but most LED's that spec 100 mA may spec up to 2A or more for short durations.
True...and I know I've shorted PIC output pins directly to ground before and pulled whatever current I could get out of them. The pins worked afterward...but for how long? I went to a 2 month class last year, had a load of slides that showed things like ESD damage, overvoltage/overcurrent damage, talked all about why things would work afterwards and fail some time after that.

Ideally I'd use a transistor but the prototype boards have already been run and it's wired direct and space is very limited for this application. Just wondering if it's at all safe or what I can expect as problems if we do run it and drop the resistance of the output low enough to generate the 75-100 Ma.
Got enough room to run a jumper wire and tie a few pins together and rewrite the firmware a bit? Not an optimal solution, but it might get you by...