The manufacturers don't even clearly specify these devices. They give you an equation by means of which you can attempt to approximate the pulse width a one-shot generates, up to a point, and then they refer you to a nomograph from which you can, in a sense, estimate what the pulse width is supposed to be.

When I need a delayed pulse, I prefer to build my own, which, by the way, I very seldom do, using a schmidt-triggered nand gate, with the same input to both inputs, albeit one delayed via series resistor followed by a cap to GND. This is essentially what a one-shot is, with the exception that one can calculate the time between the beginning and end of the delay. This type of circuit is just as sensitive to power supply noise as the one-shot, and just as likely to be influenced by board location, etc, as a one-shot, but, at least, I can reasonably quantify the timing. It's still not a digital circuit. It's just the sort of thing one uses as a short-cut when he's not willing to work out a precise, synchronous solution.

One-shots are a short-cut, and were popular during the '70's, before people learned how to design digital circuits. By the mid-'80's, they'd fallen into disfavor because of their unpredictable behaviors. They were still tolerated by some, as switch debouncers and the like, but in some industries, where digital design was taken seriously, it was a firing offense to put them in a circuit for delivery to a customer. Asynchonous methods, in general, were no longer welcomed by then.

RE