Let’s say that I want something to happen “100 milliseconds after” something else has occurred.   What I would do is simply to arrange for a timer to go off at one-half that interval.   (Thus controlling latency.)   When an event occurs, it sets a flag.   The timer handler checks to see if the flag has been set, and if so, clears the flag and does something interesting.   If that “something interesting” needs to wait until several successive quantums have passed, I will either use a counter, or, more likely, I’ll capture a snapshot of the timer value to use as a “deadline” value.   (“If the alarm goes off at 6:30, I have to be awake and on-the-road by (the next deadline is...) 7:00.   Is it 6:55 yet?   No?   Snooozzzzzeee...”   ;-)  )   Maybe, after doing something, the timer routine calculates the next deadline.

Being, properly, “lazy,” I won’t get any fancier than this.   I’ll just let that timer run forever, even when there is nothing to do.