[ntp:questions] SNTP server + ntpd 4.2.4 client

Unruh unruh-spam at physics.ubc.ca
Wed Mar 19 17:00:56 UTC 2008


Paul.Croome at softwareag.com writes:

>The web page http://www.heoldesign.com/index.php?id=58 twice
>refers to NTP ("High Precision PG NTP Micro Server"; "The
>HEOL-T101 NTP server...") and twice refers to SNTP
>("compliant with SNTP protocol"; "Timing Ethernet protocol:
>SNTP V4").

>Giving the manufacturer the benefit of the doubt, I would
>infer that it is a genuine NTP server, and that the packets
>sent over the wire comply with RFC 4330 (SNTP v4), which is
>the most recent RFC to be published on the subject of NTP
>and its relatives. As the RFC states, "The NTP and SNTP
>packet formats are the same".


Yes, I was completely wrong in casting doubt on the existence of SNTP
servers. I finally read the rfc again, having remembered only about the
clients, and this is exactly the kind of situation for which SNTP server
was designed for-- a single reference clock with no net backup.

I still worry about their claim that their timestamp has a 10us accuracy
while their PPS has a 40ns accuracy. For a dedicated box, 10us seems like
an very very long time. It should be able to timestamp those PPS to far
better than 10us and deliver an ntp time stamp to much better than that.
There is nothing in the box to cause it to lose time. 

Again, I wonder what the price on this thing is. Just to compare it with a
$60 18LVC which should be able to do just as well-- with a bit of soldering
to do. The advantages of the Heol is supposed to be that it works in poor
reception conditions, and after it has found where it is, it will work with
timing from only one sattelite ( instead of the 4 that the 18LVC requires).
It may also be designed to flywheel a bit better if no sattelites are
available. Depends on whether those things are worth the money. 

May I make an estimate as to the cost? >1000 euros. 








>Paul




More information about the questions mailing list