|
Message
From: John Sheahan<jrsheahan@o...>
Date: Fri Sep 17 12:09:54 CEST 2004
Subject: [oc] setup time,hold time and clock skew
wjx197733@e... wrote:> Though consult so many data,I can not understand the conception
>about setup time,hold time and clock skew.Do the setup time have
>relation with hold time?if have,what is the relation?Do setup time and
>hold time have relation with the clock skew?if have,what is the relation?
>what relation do the three conception have with the data out?
>_______________________________________________
>http://www.opencores.org/mailman/listinfo/cores
>
>
Setup time is the time before the clock edge that data must be stable.
The latest time it may hange if you want to predict the answer.
Hold time is the time after the clock edge that data must be maintained
stable.
meaning you keep the data constant from the setup time limit until
the hold time limit.
In reality - the flipflop has a limited window where it actually samples
the data. It is
between the setup limit and the hold limit. That window will be bigger,
as there are
other uncertainties.
If you transition in the window, you may get a metastable. Where the
flipflop makes a
long time making up its mind which way to flop.
Clock skew is a measure of when the clock actually gets to different
flipflops in the same design. Obviosly clock skew may result in hold
time violations
if its bad enough.
the flipflop will have a clock-> q spec. as long as you don't hit a
metastable
(by violating setup and hold - its not always clear which one you got wrong)
clock->q will not be affected by the prior.
john
|
 |
Browse
