My freshman year of college I proposed a study to our hospitals IRB to strap small lasers to three week old infants in an effort to measure concentrations of a chemical in their blood. The most frustrating part was not the arcane insistence on ink and bolded study names, but the hardline insistence that it was impossible (illegal) to test the device before getting IRB approval - even on ourselves. Meaning that without any calibration or testing, our initial study would likely come back with poor results or be a dud, but we couldn't find out until we filled out all the paperwork.

Incidentally, I recall it being much less painful than this, at least to get to the approval step. I would expect that each IRB varies greatly from group to group.