Of *course* you can sync to better than a millisecond on the LAN. There's not a machine worldwide at my employer more than 600 micros off from each other, and the machines at my house are within 50.
You wanna start talking the sync-e+1588 test I'm doing? We're speaking in nanos then. My LTE Lite is the only USB pps I have presently - and it pulls my time well over 200 milliseconds off reference. That's a massive change from the 1 or less I am from the internet and the 50 micros from the other boxes. NS On Sun, Dec 14, 2014 at 10:30 AM, Chris Albertson <albertson.ch...@gmail.com > wrote: > > I would still like to experiment with it. As I wrote earlier I bought this > > for a frequency reference, not a clock, but I would not object to a bit > of > > fun messing around with it. > > > > > If the goal is just getting good enough time onto the Solaris machine then > use NTP and some pool servers on the Internet. You get about 10 > millisecond level accuracy and the cost is zero. If you have solaris > running you might even have this all setup and running. If not do this as > the first step and verify it works. > > If 10ms is unacceptable, next step is to connect the PPS signal. Doing > this will move you from milli to micro second level accuracy. It is easy > if the Solaris machine has a real serial port. If you have to go through > a USB dongle you loose about an order of magnitude accuracy but this is > still very good. > > There is zero point in buying a special computer to run NTP. Just use any > computer you own that is already running 24x7. Of course if you don't have > a computer that runs 24x7 then you would look for one that uses very little > power. > > Don't worry to much if the USB connection skews the time on the NTP server > by some tens of microseconds, your server can't transfer time to your other > computers on the LAN any better than millisecond level so a few tenths of > an millisecond hardly mater. > > My opinion of computer time is that for normal use being a few milliseconds > off is OK because the typical monitor is refreshed no faster than 100Hz so > you have lag cause by screen refresh times even if the internal clock is > dead-on perfect. Same for disk time stamps, these is lag in the IO system > too > > > -- > > Chris Albertson > Redondo Beach, California > _______________________________________________ > time-nuts mailing list -- time-nuts@febo.com > To unsubscribe, go to > https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts > and follow the instructions there. > _______________________________________________ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.