[ntp:questions] Re: Linux PPS triggering with NMEA

Ulrich Windl Ulrich.Windl at RZ.Uni-Regensburg.DE
Mon May 9 12:39:42 UTC 2005


"MarkO" <orzech at nps.edu> writes:

> We're currently running ntp version 4.1.1 at 1.786 on linux kernel version
> 2.4.18-4.  We have a Garmin NMEA GPS connected to the serial port and
> sending a (default) 100ms PPS pulse on pin #1.
> 
> I have a couple very basic questions.  First, do we need to install
> PPSkit in order to synchronize with the Garmin's PPS pulse?  And, if
> so, can we use PPSkit 2.0.2 (for linux kernel 2.4.19) with our current
> kernel, or do we need to upgrade the kernel as well?

I think you'll need the PPSkit, and a new kernel is strongly recommended for a
variety of security reasons.

> 
> I've run a few tests that lead me to think the answer to both above
> questions is "Yes", but since I'm such a NTP novice, I was hoping to
> get a confirmation.  In trying to get the PPS synchronization running,
> I've set up the ntp.conf file as follows:
> 
> server 127.127.20.1 mode 1
> fudge 127.127.20.1 flag3 1
> fudge 127.127.20.1 time1 0.042
> driftfile /etc/ntp/drift
> 
> Ten minutes or so after starting ntpd, I get the following output from
> ntpq -p:
> 
> remote    refid      st t when poll reach   delay   offset  jitter
> ===================================================================
> *GPS_NMEA(1)  .GPS.   0 l    1   64   37    0.000   31.270  27.697
> 
> and from ntptime, I get:
> 
> ntp_gettime() returns code 0 (OK)
>   time c60147cc.bd2c9000  Fri, Apr  8 2005 11:11:24.738, (.738961),
>   maximum error 725913 us, estimated error 19588 us
> ntp_adjtime() returns code 0 (OK)
>   modes 0x0 (),
>   offset 26811.000 us, frequency 153.481 ppm, interval 4 s,
> maximum error 725913 us, estimated error 19588 us, status 0x1(PLL),
>   time constant 2, precision 1.000 us, tolerance 512 ppm,
>   pps frequency 0.000 ppm, stability 512.000 ppm, jitter 200.000us,
>   intervals 0, jitter exceeded 0, stability exceeded 0, errors 0.

This is the nano-free version of the software!

> 
> 
> However, the time on our system seems to be consistently about 1/2
> second behind the correct time (possibly triggering on the NMEA text
> line rather than the PPS pulse?).  In a previous post to this mailing
> list, Steven ("rtxo") suggested that you should see the line PPSSIGNAL
> in the ntptime output if PPS is running properly.
> (ref:
> http://groups-beta.google.com/group/comp.protocols.time.ntp/browse_thread/thread/419f18b3a5c58e52/9ca0e3e5470738cd?q=pps&rnum=21#9ca0e3e5470738cd
> )
> 

You may play with some PPS API offset adjustments or NTP driver fudge flags.

> Though our ntptime output does refer to the "pps frequency", it doesn't
> have a "ppssignal" line.  

Possibly due to being 0.5s off.

Sorry for the late reply.

Ulrich



More information about the questions mailing list