[time-nuts] PC Time Servers

[Hal Murray]

rchrdm at earthlink.net said:
> Faros depends on an accurate time source that has consistent delays.  I guess
> “delay” here means the time it takes for the time signal to get to the PC and
> Faros. 

I don't know anything about Faros.  How accurately does it need to know the 
time?  (and/or what happens if the time is off)

In the context of NTP (Network Time Protocol), delay refers to the time it 
takes for the request packet from the client to get to the server and for the 
response packet from the server to get back to the client.

If the routing is symmetric, the client can correct for the transit time by 
adding half the round trip "delay" to the time from the server.  But that 
depends on symmetric routing and queuing.

In general, larger delay means more opportunity for asymmetric routing.

If you know the time at both ends, you can measure the network transit delays 
in each direction.  As an example, from San Francisco to NIST in Boulder CO, 
it's 15 ms there and 40 ms back.  (Yes, that's a horrible example.)


> I have a simple GPS receiver (from Adafruit) connected to the PC that sends
> data to Lady Heather.  Is there a way I can use the data from the GPS in
> Faros? 

Model number or URL?

Does it have PPS?  (Pulse Per Second)

There are 2 ways to get time from a GPS unit into a PC.  One is to time-stamp 
the serial data messages.  The other is to time-stamp the PPS signal.  Linux 
and *BSD have kernel support to grab the timestamp from a PPS signal connected 
to one of the modem control signals on a serial port.

Most low cost GPS units use USB without PPS.  On a time-nut scale, USB adds a 
ms of noise but the timing on the serial data is usually much worse than that. 
 For good timing, connect a PPS signal to a real serial port.

For more fun, see hanging bridges.



-- 
These are my opinions.  I hate spam.