[ntp:questions] Home Time Server

John Ackermann N8UR jra at febo.com
Thu Oct 11 12:27:48 UTC 2007


David L. Mills said the following on 10/10/2007 08:26 PM:

> The Spectracom WWVB radios we have used to do an excellent job; however, 
> the claimed accuracy was only 0.1 ms, mostly because the computer used 
> for demodulation was so chintzy. Sadly, WWVB is dead here because of EMI 
> generated by hundres of UPS systems all over campus. I'd be curious to 
> know if this has also affected DCF-77 users.

I'm still running two Spectracoms here in Ohio (one a WWVB timecode
receiver, the other a WWVB disciplined oscillator) and they work very
well, though not as well as either LORAN-C or GPS receivers.  One or the
other of the receivers loses lock momentarily perhaps every two or three
weeks, but otherwise they are solid.  When lock is lost, there is a
short blip of about -30 milliseconds, with rapid recovery to within a
millisecond or two.  I suspect that blip could be minimized by either
realigning the VCO in the unit, or using an external 10 MHz source
instead of the simple crystal in that VCO.  There was a factory option
for external reference input, but mine doesn't have that.  Yet.

The 0.1 microsecond accuracy Dave mentions is also tied the the method
Spectracom uses for steering; it only adjusts time in 100 microsecond
steps, so that inherently limits the accuracy.

One reason I like keeping the WWVB receiver around is that it give me
something I can easily say is "NIST traceable" without going through
post-correction.

John



More information about the questions mailing list