[time-nuts] GPS lock of the FE5680. Current experiment and question
hmurray at megapathdsl.net
Sat Feb 11 02:19:53 UTC 2012
> Flip-flops hardened against metastability are available.
Do you have a part number in mind?
The only data sheet that I remember for a part that looked interesting was
actually 2 FFs inside one package. For something like that, you save the
prop time of the output driver and the input receiver relative to a classic
pair of FFs. That should be a significant improvement relative to the
classic pair of FFs. Other than that, the parts I remember were nothing
special. Maybe the designers actually understood metastability and did some
testing and wrote a data sheet with promises. I don't remember the details.
There is no cure for metastability. You can only make the probability lower.
There are two ways to make things better. One is to get faster logic and run
it at the old clock rate. The other is to wait longer.
If you wait long enough, you can make the probability low enough, as in a
MTBF of greater than the age of the universe.
Modern high speed logic has a (much) faster settling time which makes the
window smaller, but then we run the whole system at a faster clock rate so
the probability of events per second stays roughly the same.
These are my opinions, not necessarily my employer's. I hate spam.
More information about the time-nuts