Problem with Linux Machine's Request for Time from an XP Machine

Discussion in 'Linux Networking' started by W. Watson, Jan 9, 2005.

  1. Allen brings up a significant point, both in terms of how such
    equipment is implemented in commercial use (where the required
    reliability means redundancy is essential) and as a direction to
    look toward future developments of inexpensive time and
    frequency standards.

    Commercial implementations commonly use two Stratum 2 level
    clocks (typically rubidium based) synchronized by GPS to Stratum
    1 level accuracy. That arrangement can tolerate failure of the
    GPS or one of the Stratum 2 clocks and still provide sufficient
    accuracy for at least one month.

    Note that the definition of "stratum" levels is different than
    seen when discussing NTP. A Stratum 1 clock is defined by ANSI
    to have a minimum accuracy of +/- 1x10e11 over 20 year period.
    A Stratum 1 level clock is capable of standalone operation
    (plesiochronous synchronization) and does not require
    synchronization for accuracy, therefore no minimum stability is
    specified for a Stratum 1 clock.

    Stratum 2 is defined as having minimum accuracy of +/- 1.6x10e8
    and minimum stability is a drift rate of less than +/- 1x10e10
    per day, with the capability of being synchronized to a clock
    with the same accuracy. (Stratum 3 and 4 clocks are also
    defined, but have either very short term or no holdover
    accuracy, and require synchronization to maintain accuracy
    better than +/- 4.6x10e6.)

    Typically a cesium clock has +/- 1x10e11 accuracy, and a GPS
    synchronized clock can reach +/- 1x10e12. Historically, Stratum
    1 clocks were expensive until the recent advent of GPS
    synchronized units.

    Prior to GPS, typically Stratum 2 clocks were synchronized
    either from a timing source clocked by a Statrum 1 clock (T1
    carrier facilities) or from Loran C (similar to using GPS). GPS
    has the advantage of being available almost anywhere, and since
    it has become ubiquitous it has also become inexpensive.

    GPS has disadvantages too, which appear to be overcome by a
    recently developed miniature cesium based clock. GPS, as noted,
    requires satellite communications, which is simply not available
    inside buildings or behind mountains. GPS is also not small
    enough to fit into an individual cell phone. But DARPA and NIST
    have recently developed a cesium clock, with 1x10e11 accuracy,
    that is the size of a grain of rice and is expected to initially
    cost less than $200.

    For hobbiest the results may be two fold. First, the cost of
    these cesium clocks will probably soon be within range of the
    more dedicated (I'm hoping to see a PCI board with a cesium
    based clock available with a year or two for less than $300.)
    Second, as commercial companies replace older time and frequency
    standards with newer equipment there should be a flood of very
    interesting used equipment with rock bottom prices.

    Relatively soon we'll all have what Allen has!
     
    Floyd L. Davidson, Jan 13, 2005
    #61
    1. Advertisements

  2. W. Watson

    James Knott Guest

    In the communications company I used to work for, we had Loran derived
    clocks, made by Austron, to provide timing for the DS1 (the correct term
    for what is commonly called T1), DS3 and SONET systems.

    Incidentally, I recall an interesting discussion I once had with one of the
    techs. I pointed out that the office clock was wrong and he insisted it
    was correct, as it used the network clock as a time base. What he missed,
    was that while the time base was extremely accurate, the clock itself was
    simply set to the wrong time and was consistantly wrong by precisely the
    same amount. ;-)
     
    James Knott, Jan 14, 2005
    #62
    1. Advertisements

  3. W. Watson

    James Knott Guest

    Hmmm... That could be used in a wrist watch. Imagine, a stratum 1 clock on
    your wrist!
     
    James Knott, Jan 14, 2005
    #63
  4. W. Watson

    Mark South Guest

    It's hard to read the time accurately off something the size of a grain of
    rice ;-)
     
    Mark South, Jan 15, 2005
    #64
  5. Technically a T1 and a DS1 are two different things, though most
    people tend to use them interchangeably to mean either of them.
    DS1 is the interface. T1 is a carrier system (probably with a
    DS1 interface).

    I don't generally make much of a big deal about the difference,
    and in the above instance it would have been perhaps more
    correct to say DS1 than T1, though not all DS1s are suitable for
    timing, while all T1s are.
    That's the problem seen in the web page that was presented as a
    method for the OP to run a time server on XP. It listed
    precision at well within the specs necessary, but the *accuracy*
    was way too far out of bounds.

    In regard to Loran-C timing systems, I had an interesting
    experience once. We had one installed in the early 80's, and in
    the wonderful (i.e., fitting for a Dilbert cartoon) perception
    of management, exactly one technician was sent to maintenance
    school. So a dozen years go by... and one day it fails. This
    thing is located at the back of the equipment room. I just
    happen to be standing in the one position where I can look all
    the way down an aisle and see the _front_ door at the other end
    of the building. At just the point where the Pointy Hair Boss
    (who literally did look almost exactly like Dilbert's PHB) says
    "Who was it we sent to school on this thing???", I see that
    particular individual heading out the front door about as fast
    as he can escape. I just about died laughing, because it seemed
    also that I was the only one who actually remembered who it was
    that had been sent to school.

    (Now, in all honesty, I must admit... the tech did *not* even
    know that the Loran-C system had failed, and was not running
    away from a nasty job. If he had known, he probably would have
    been moving twice as fast. The fact is he had something like a
    doctor's appointment... But of course we didn't know that until
    hours later. Regardless, I didn't say a word until after he'd
    had enough time to escape. The point, in my mind, was that
    having only one person schooled on something that critical was
    poor planning.)
     
    Floyd L. Davidson, Jan 15, 2005
    #65
  6. It'll have to be put into an hearing aid. The trick is going to
    be designing the right interface... Perhaps voice recognition?

    You'll see people walking along, saying apparently to themselves
    "What time is it?" (And of course only they will hear the
    answer.)
     
    Floyd L. Davidson, Jan 15, 2005
    #66
  7. W. Watson

    James Knott Guest

    But they'd also have to put up with the continuous ticking of the
    clock. ;-)
     
    James Knott, Jan 15, 2005
    #67
  8. W. Watson

    Mark South Guest

    That would be fine, being a nutter walking along talking to yourself has
    become socially acceptable since handsfree set for mobile phones became
    available.

    If the Linux system in which it is embedded works on VR then you will hear
    people walking along talking to themselves saying stuff like "liss space
    dash ell pipe grep myconfig dot conf".
     
    Mark South, Jan 15, 2005
    #68
  9. W. Watson

    James Knott Guest

    If it looks like a grain of rice, someone might come up with the slogan
    "Takes a licking and keeps on ticking." ;-)
     
    James Knott, Jan 15, 2005
    #69
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.