The industry-standard practice, since the early days of Intel, is to advertise the p*ss out of a product, writing all sorts of promotional articles to appear in industry magazines, and then, after the response appears to be at least luke-warm, hire the engineering team to design it.

Often this leads to disappointing results, but, if the applications aren't firmly designed before the first silicon is sampled, the mfg's are afraid they won't be able to sell enough of the things to get their bonuses.

Most engineers are, quite justifiably, leery of designing from a datasheet marked "preliminary" or "objective specification" at least nowadays, because one can't rely on things changing ... important things like Vcc, or Fmax. It's really disappointing when a part that was advertised and sampled as a 50 MHz part turns out to operate at 33 MHz max. It's also difficult to explain to the boss, why a circuit board that was designed for a 5-volt part now won't work becuase it also needs a 2.5-volt regulator, ferrites, bypass cap's and 30 bits of level translators. It's also hard to explain that the circuit may not work because of the associated propagation delays introduced by those level shifters.

If the "TBD" appears on the no-longer "preliminary" datasheet, I toss the entire folder in the can. Someone else makes just as good a part that doesn't require that I "hope for the best."

If a manufacturer doesn't want to spend the money to determine precisely what the operating parameters of his product are, then I don't want to risk my money, or anyone else's on his "guesstimated" spec's. f he's not even willing to publish an estimate, publishing "TBD" instead, he belongs in that can along with his bogus product description. That's true of ANY parameter, including the Icc variation over temperature, and other seemingly innoccuous parameters.

RE