[ntp:questions] How to keep Linux server in Chicago and Mumbai in sync to within 5 microseconds

Chris Albertson albertson.chris at gmail.com
Wed Jan 12 02:11:32 UTC 2011


> Two GPS receivers ( Garmin GPS 18X for about $70 each and a bit of time
> soldering on an RS232 plug.
> )
> Certainly cheaper and much more accurate. And if you can see the cell
> tower to get the CDMA you should be able to see the GPS sattelites.

Yes the GPS can be even cheaper than $70.  But the CDMA signal really
can go through walls that the GPS signal can't.  So I'd say the way to
deside is to bring a hnd held GPS to the proposed antenna ocation and
see what hapens.  Can it get a good view of the sky.   Most GPSes have
a display screen that will answer that question.

Also the OP is asking for +- 5 microseconds.  That is moderately hard.
 It means each box can only have a worst case 2.5 uS error.  To get to
that level some care is required.)    The specs on the GPS 18X say the
pps is "+- 1 microsecond" which is not very good.  It uses to much of
his error budget for my taste,  For the same price you can have at
least 10X better.   and maybe 100X better with a $200 Thunderbolt GPS.
 Then in effect none of his error budget is used up outside of the
computer.

Then we have to know a little about the computer.  Hopefully it has a
real serial port and is running BSD or Linuix and has the pps kernel
API.

This is that kind of thing that will suck up the OP's time.  There are
a million little things that he was to learn and each takes 10 minutes
or an hour.    To me reading this stuff is "entertainment" and a good
way to avoid the TV and learn something.  But to another person it is
just a huge "time sink" (pun intended)

Bottom line:  This is just the classic "build or buy" business decision.

=====
Chris Albertson
Redondo Beach, California



More information about the questions mailing list