1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Grabbing the time, possible?

Discussion in 'Tomato Firmware' started by rommels, Nov 5, 2007.

  1. rommels

    rommels LI Guru Member

    Is it possible to grab the router time and use it to set the windows time?

    I normally use TimeSync, but I figure since the router is grabbing it every 4 hours why not grab it from there instead if possible.

    So is it? ;p
     
  2. acidmelt

    acidmelt LI Guru Member

    The stock tomato firmware does not have an NTP server embedded therefore you will not be able to grab the time directly from the router (not over NTP at least), if youd like you could modify the firmware to include the OpenNTP daemon or any other NTP server which will allow all of the computers on your network to synchronize with the router.
     
  3. rommels

    rommels LI Guru Member

    Can a openntpd be like uploaded onto the router like custom css can and then be installed and ran using the console?

    Edit: Found another thread talking about this. Guess things haven't progressed on this front.
     
  4. mraneri

    mraneri LI Guru Member

    My router is a REALLY bad timekeeper. It drifts about 20 seconds per day. I have it syncing with NTP every 6 hours, so it gets up to 5 or 6 seconds off. Would be better off setting up a windows box to sync with the internet once a day, and setting the router to sync off of the internet box once per day..

    Either way, wouldn't want to have the router keeping time for anything, really.

    Why these guys can't bias up a crystal oscillator so it oscillates at the right frequency, or why they can't use a 50PPM crystal is beyond me...
     
  5. RonWessels

    RonWessels Network Guru Member

    Why do they need to? Given that syncing with NTP keeps it to 6 seconds maximum error, it is perfectly acceptable for what it needs to keep time for: access restrictions. Nobody is going to complain if access restriction rules kick in 6 seconds late. And I'll bet most users of these routers (not necessarily most users in this forum) don't even use any feature that requires the router to know what time it is.

    Being an accurate time server was _never_ a requirement for the hardware of these routers, and to spend the time biasing the oscillator or to use a higher accuracy crystal is simply "wasted" expense.
     
  6. Maggard

    Maggard LI Guru Member

    Not only would the (admittedly trivial) hardware required for accurate timekeeping be an almost universally unrewarded additional expense, it would also be solving a nearly non-existent problem.

    NTP takes so little traffic, and the greater reliability of an off-site pool of servers so superior, that including accurate timekeeping in a consumer router would almost be more hassle then benefit.

    For a large site where critical processes must be accurately synchronized & logged then a ‘real clock’ is a decent investment. But for an additional couple bucks on a ~US$50 consumer device it’d be a competitive disadvantage.
     
  7. mraneri

    mraneri LI Guru Member

    Couple of bucks? Biasing a crystal properly costs 0 cents (basically, they need to read the app-note for the crystal they're using and use the right resistor and capacitor values instead of the ones they're using). Also, a 50PPM crystal is probably 0.2 cents more expensive in volume than a 100PPM crystal, and you would have trouble finding a crystal at a mainstream frequency that's less accurate than 100PPM...

    In other words, it wouldn't cost anything.

    Yes, you're all right, otherwise.. A router doesn't need to have ultra precision accuracy. The point of my post in the first place was that you don't want to use it as a source to set your system time.
     
  8. Maggard

    Maggard LI Guru Member

    It's 0-2 cents IF they use the same design AND the right resister AND the appropriate capacitor, all of which manufacturers are free not to use right now. Also that's one more set of requirements that parts can fail and thus that additional percentage of undesirable units.

    Therefore, as folks in manufacturing will happily tell you, the additional 0-2 cents on parts plus the now stricter components means they're not so free to purchase on the spot market. Also validating this subsystem would require another step in Q/A. Thus the final additional manufacturing cost goes up maybe 50 cents which translates into a buck or so more to the consumer.

    So all told that trivial change can end up costing unreasonably much for, as we've mostly agree, a nominally useful feature.
     

Share This Page