[ntp:questions] Re: 1pps signal,20us width

Fei Wu fxw115 at yahoo.com
Fri Nov 5 16:22:30 UTC 2004


Hi Paul:

> The PPS signal is usually connected to the DCD pin
> of a computer's
> serial port, where it triggers interrupts. As long
> as each pulse is
> long enough to trigger, it doesn't matter how long
> it is. The
> pulse length is not related to the port's baud rate,
> so changing
> that won't help you.
Thanks for your explanation. It is very clear to me
now.

> 
> You won't really improve performance by distributing
> the master clock
> signal to several stratum-1 servers -- you still
> have a single point
> of failure. Consider using one stratum-1 server and
> a couple of
> stratum-2 servers. That's easier to wire up and just
> as reliable as
> your suggested configuration.

I agree that I still have a single point of failure.
What I try to improve is the synchronization. I am
actually comparing the results by using 1pps
discipline and net synchronization now. The results
show that: with 1pps time source(CDMA
clock)discipline, the root distance and root
dispersion of synchronization are one order better
than net synchronization by a gps server(which is 1pps
disciplined).

Fei 


		
__________________________________ 
Do you Yahoo!? 
Check out the new Yahoo! Front Page. 
www.yahoo.com 
 




More information about the questions mailing list