On Mon, Dec 29, 2014 at 7:02 AM, David J Taylor <
[email protected]> wrote:

> Using a mixture of PPS and local servers is exactly what I was aiming for,
> but:
>
> - NTPns doesn't support my PPS source
>

I don't understand this.  But getting a compatible GPS seems reasonable to
move forward.


> - for testing, using local servers alone, even the seconds don't get
> "named" correctly, even though NTPns shows they are being detected and one
> is even marked as "SELECTED".
>

This would be the part where I said reach out to PHK regarding NTPns.

Getting back to the "try" -- I assume you've read <
http://www.febo.com/pages/soekris/>.  Notice that while the gpio results
are a win over the serial port they're nothing to write home about although
that may be because he used ntp rather than NTPns.

[columns are host, refid, delay, offset, jitter]
His Soekris numbers (presumably 10Mb ethernet):

-tock.febo.com .GPS. 1.012 -0.117 0.499
+tick.febo.com .PPSC. 0.990 -0.052 0.729

My numbers (typical desktop with serial PPS talking to a "dedicated" ntp
server also using serial PPS):

*ntpa   .GPPS.  0.179  0.003  0.003

Or the Raspberry Pi and its jittery network interface:

+rPi1  .GPPS.  0.436 -0.006 0.036

So for a "try" you could build a low-latency, high-res time-stamper (per
PHK) based on a 4501 but you can't get that precision out of the box so
it's of limited practical use. It sounds like you want something practical
but I'm not sure.
_______________________________________________
time-nuts mailing list -- [email protected]
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Reply via email to