[ntp:questions] Very large jitter and offsets on GPS ref clock after upgrade to "p5"
snews at lordynet.org
Wed Feb 8 17:45:06 UTC 2012
David J Taylor wrote:
> "David Lord" <snews at lordynet.org> wrote in message
> news:gnbbu8-qne.ln1 at me6000g.home.lordynet.org...
>> I've not put a scope on the pps but I added an extra section
>> of cable from my Sure GPS, 18 m of 4 x pair telephone cable.
>> I will need that length to reach from the GPS down to my PCs.
>> In the process the offset jumped from 0.004ms to about 12ms
>> but fairly quickly went back to a few usec.
>> PC = me6000 (VIA ME6000), NetBSD-5, ntpd 4.2.6p5
>> From my ntp-stat file:
>> time offset(ms)
>> 15:06 -0.004
>> 15:12 0.000
>> 15:18 0.001
>> 18:24 0.006 extra 18 m cable added
>> 15:30 -12.938
>> 15:36 -3.672
>> 15:42 -0.814
>> 15:48 -0.186
>> 15:54 -0.054
>> 16:00 -0.012
>> 16:06 -0.006
>> 16:12 -0.002
>> 16:18 -0.002
>> 16:24 -0.001
> I suspect that the offset jump was purely from losing the PPS signal,
> and nothing to do with the extra cable as such. BTW: I think your
> editing accidentally included the time - from 15:24 to 18:24. Must have
> had "18" on the brain! <G>
I didn't spot that until after I'd posted.
> My comments on wiring are as before - use one line from each pair as
> ground, and then the other four "active" lines become TXD RXD PPS and
> +5V power.
Nearly that, but I'm splitting the power and planning on
having +9V and -Ve on two of the lines and a 5V regulator
since some of the GPS modules with interfaces take over
200mA so common OV with the -Ve at the far end should
remove any effect from voltage drop over that length of
I might try rs422 for the PPS (and TxD lines) but they
are 16-pin chips and have unused sections. The 8-pin
versions with single Tx + Rx are much too expensive.
More information about the questions