[ntp:questions] NTP on local networks
snews at lordynet.org
Thu Aug 2 14:17:47 UTC 2012
Will Shackleford wrote:
> I should have provided a number of additional details.
> The lab is an interior lab on the ground floor of a 3 story building
> with no windows.
> The work is not classified but security rules do make things difficult.
> GPS doesn't seem like an easy solution.
> The computers log data from an experiment(s). If the time stamps in the
> log files for
> one computer are off by 20-30ms from the time stamps in the log files of
> another computer
> then the results of the experiment could be impacted.
> However if all computers were off by the same offset from official time
> of a few seconds, it wouldn't matter.
> Off by more than a few seconds wouldn't make the data unusable but might
> cause some confusion when we go to find the right log files.
> The rubidium units I see on E-bay look like they were designed to go in
> a CD player. Would I need
> additional hardware to connect it to a computer and software to read it
> as a clock source?
I have an LPRO-101 that I've not yet powered up. It's about
size of a CDROM drive but I will be fitting mine with a large
heatsink to avoid temperature gradients. Peak power usage is
above 24V at 2A but that drops to maybe 0.5A when unit has reached
operating temperature. With a lower current supply the unit
will take longer to warm up or may not even reach operating
Next problem is that the 10 MHz output needs dividing to give
1 PPS and that needs synchronising against an accurate 1 PPS
signal, eg from GPS.
For level of accuracy you need, you could synchronise vs ntp
but I've no idea how to get a 1 PPS from ntp. A radioclock
would probably be ok, eg in UK I receive MSF which can be
within around +/- 1ms. DCF from Germany can also be received
but is subject to propogation effects and fading. The DCF
signal is phase modulated at the second so if near to the
transmitter you have usec accuracy.
More information about the questions