Re: getting time from WWV
From: Floyd L. Davidson (floyd_at_barrow.com)
Date: 06/27/05
- Next message: A. Ben Hmeda: "Re: Need mp3 to mpy file converter for RCA Lyra RD1080 mp3 player"
- Previous message: Leonhard Vogt: "Suse9.3: Siemens PCCard11 works, but I need help"
- In reply to: phil-news-nospam_at_ipal.net: "getting time from WWV"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Mon, 27 Jun 2005 05:11:48 -0800
phil-news-nospam@ipal.net wrote:
>I have a potential project to build a small cluster of computers to do
>a certain complex task which requires accurate timekeeping. But these
>computers are also not going to be connected to the internet, so using
>some NTP servers on the internet to get time isn't an option. What are
A discussion of various time sync methods for a local ntp server:
http://www.wraith.sf.ca.us/ntp/
>good choices of _hardware_ to obtain to get accurate timekeeping based
>on receiving WWV transmissions at 60 kHz, and have stable oscillators
>to hold that time within 0.1 ppm when WWV can't be received ... AND can
You won't find anything that will sync to wwvb to get even near
that accuracy. Likewise you won't be able to sync anything in a
PC to better than about 1 ppm (primarily due to bus jitter).
http://www.isoc.org/isoc/conferences/inet/01/CD_proceedings/T42/inet2001.html
I don't know if it is showing up in any commercial products yet,
but NIST has developed an inexpensive atomic clock. Within a
short time we should see $200 atomic clocks on a PCI card, and
soon enough the price will be below $100. (Actually, in not too
long there will probably be an atomic clock in every cell phone.)
http://tf.nist.gov/ofm/smallclock/index.htm
>work with Linux (presumably anything with an ethernet and NTP support
>should be just fine, but something with another means might need its own
>driver).
In the mean time, either a rubidium based Stratum 1 clock or a
GPS synchronized Stratum 2 clock (both of which are expensive)
in a standalone NTP server is just about the only way you will
get 0.1 ppm or better.
Also, be aware that NTP sources and documentation have a very
different definition of what a "Stratum" level means than those
used by industry. Hence you'll see references that are
conflicting, and others that are ambiguous.
Stratum 1 means 20 year minimum accuracy of +/- 1 x 10e11. A
Stratum 2 clock has a 20 year minimum accuracy of +/- 1.6 x 10e8
*and* a minimum (free running) stability of +/- 1 x 10e10 per
day and be capable of synchronizing to an accuracy of
+/- 1.6 x 10e8.
Essentially a Stratum 1 clock is meant to be a standalone free
running clock, while a Stratum 2 is intended to be synchronized
but have a very long (more than 1 month) holdover accuracy if
synchronization is lost. (There are lower Stratum levels too,
but they are not intended to provide holdover accuracy for more
than hours, or even be synchronized to better than 5 ppm.)
The NTP view of Stratum levels is merely an indication of how
many synchronization levels exist between any given clock and
a genuine atomic clock. Hence a Stratum 2 clock may or may
not be accurate, and may or may not remain stable when not
synchronized.
-- Floyd L. Davidson <http://web.newsguy.com/floyd_davidson> Ukpeagvik (Barrow, Alaska) floyd@barrow.com
- Next message: A. Ben Hmeda: "Re: Need mp3 to mpy file converter for RCA Lyra RD1080 mp3 player"
- Previous message: Leonhard Vogt: "Suse9.3: Siemens PCCard11 works, but I need help"
- In reply to: phil-news-nospam_at_ipal.net: "getting time from WWV"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
