Sean Breheny wrote:
> Let's say I have a FFs. Let's also say that the input signal changes only
> once. On the first clock edge, the input signal violates the set-up and
> hold time and the FF goes metastable. According to you, on the next clock
> edge, I am guaranteed that the FF will assume the correct output state.
> No more metastability after two clock cycles.

Yes.

> Yet, the basic underlying principles state that it is not possible
> to guarantee that you can make a discrete decision yea/nay about a
> continuously-valued input within any bounded time. You are saying
> that you can guarantee that this decision happens within three clock
> cycles.

Well, that idea (the actual quote is, "No choice between near-simultaneous
events can be made unambiguously within a preset deadline.") comes from the
same paper and the reference goes to an enormous website in which it would
take some time to track down the context in which it was made.

My understanding is that the "choice" we're talking about here is which
came first, the data edge or the (first) clock edge? The fact that the
metastability is gone after the second clock edge does not answer that
particular question; in fact, it erases all of the available information
about that question. You now have the answer to a different question:
Did the data edge occur before the second clock edge? (Yes!)

If there is no second clock edge, then the principle holds: The metastable
state -- and the inability to decide -- can persist indefinitely.

-- Dave
--=20
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.