There’s always a gap between how a design performs on paper and how it performs (or even works at all) in the “real” world, a phenomenon which is most often apparent in the timing glitches that plague complex timing chains. Closing that gap to commercially-acceptable levels is often a forensic task worthy of V.I. Warshawski, Sam Spade, or other legendary pulp fiction detectives. But if you arm yourself with the right tools, interrogate the right suspects, and learn to think like a detective, you’ll be able to crack the toughest cases. Most of the tools of the trade and necessary clues you’ll need are close at hand: it’s just a matter of applying some investigative skills to the problem at hand.

 

Consider, for example, exhibit A: the humble data sheet. Just because a number exists on a data sheet under a tersely-explained mnemonic doesn’t mean the information is as useful as you assume. It doesn't take a lot of imagination to suppose that a company might lie on the data sheet to get into a favorable position on your Bill Of Materials. Well, they don’t exactly lie: but they are not above omission of critical information or use of vague and often undefined terminology. Things like rise time, fall time, jitter and duty cycle figures are often stated as “typical”, “rms” or “average” values. Does the vendor’s idea of “typical” match what is typical to you? Sometimes key information is just plain omitted. Can you orthogonally add the jitter, phase error and output to output skew up on a device and calculate the output clocks placement windows? Absolutely not! Can you get this data from a data sheet? Only one vendor has ever addressed the issue with the term TTB (Total Timing Budget)!