Re: PC Board with Accurate Clock?
From: W. Watson (wolf_tracks_at_invalid.inv)
Date: 01/13/05
- Next message: Greg Crinklaw: "Re: wasteful government spending??"
- Previous message: W. Watson: "Re: PC Board with Accurate Clock?"
- In reply to: Charlie Hubbard: "Re: PC Board with Accurate Clock?"
- Next in thread: W. Watson: "Re: PC Board with Accurate Clock?"
- Messages sorted by: [ date ] [ thread ]
Date: Thu, 13 Jan 2005 17:26:46 GMT
Charlie Hubbard wrote:
> On Thu, 13 Jan 2005 11:04:04 GMT, "W. Watson"
> <wolf_tracks@invalid.inv> wrote:
>
>
>
>>>I think there is some latency in the ntp (but works for me). There are
>>>a variety of SCADA devices that can use GPS and interface with PC's
>>>using millisecond accuracy. They are used for timing switching events
>>>in, say , critical environments, like electical distibution systems.
>>> Do a google search for "scada gps time" and maybe "modbus" and you'll
>>>see a bunch of hardware/software ideas.
>>>
>>>Uncle Bob
>>
>>I'll take a look. By latency do you mean there's some constant time shift
>>in the delivery of the time, but if latency could be removed the time would
>>be accurate to a low ms?
>
>
> NTP is sophisticated and complicated but can give good accuracy. To
> learn more than you ever wanted to know about NTP, see RFC-1305
>
> http://www.faqs.org/rfcs/rfc1305.html
>
> This is worth the read even if you don't plan to use NTP because it
> talks a lot about other sources of accurate time and the problems in
> keeping highly accurate time on computers in general. A couple of
> snippets from the document relevant to the ongoing discussion
> follow...
>
> The accuracies achievable by NTP depend strongly on the
> precision of the local-clock hardware and stringent
> control of device and process latencies. Provisions
> must be included to adjust the software logical- clock
> time and frequency in response to corrections produced
> by NTP...
>
> and
>
> When precisions not less than about 1 ms are required,
> the Fuzzball clock model can be adapted to the Unix
> model by software simulation, as described in Section 5
> of the NTP specification, and calling adjtime at each
> adjustment interval. When precisions substantially
> better than this are required, the hardware microsecond
> clock provided in some workstations can be used
> together with certain refinements of the Fuzzball and
> Unix clock models.
>
Good. I'll take a look. I definitely need something at a more general view
of accuracy than plowing through spec details, etc.
I think I'm starting to lean towards a radio frequency clock attached to
the Linux box. I think I mentione the How-to Linux doc somewhere above.
It's a bit out of date, but has sources for such boards that are $80-150.
It seems many have disappeared, but I think I've found a Garmin GPS 25
engine (board-like) that might do the trick. It and the powersupply for it
are $115. It may be possible to do this with a newer GPS unit than I have.
I understand that going through a serial port may cause a few problems.
Something more to be explored.
--
Wayne T. Watson (Watson Adventures, Prop., Nevada City, CA)
(121.015 Deg. W, 39.262 Deg. N) GMT-8 hr std. time)
Obz Site: 39° 15' 7" N, 121° 2' 32" W, 2700 feet
Web Page: <home.earthlink.net/~mtnviews>
- Next message: Greg Crinklaw: "Re: wasteful government spending??"
- Previous message: W. Watson: "Re: PC Board with Accurate Clock?"
- In reply to: Charlie Hubbard: "Re: PC Board with Accurate Clock?"
- Next in thread: W. Watson: "Re: PC Board with Accurate Clock?"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
