[gpsd-users] PPS MID 52 (0x34) working?

[Alexander Carver]

I have a question about the SiRF PPS message MID 52 and whether it's 
actually working in gpsd 3.4 (haven't quite updated to 3.6 yet).  I'm 
playing around with turning MID 52 on and off and have noticed that I 
get less wander in the offset to the SHM that is sent over to ntpd.

With MID 52 on, my offset wanders +/-40ms.  With MID 52 off, the wander 
is reduced to about +/-10ms.

I see a long comment in the source regarding MID 52 being output on the 
second which is true of my receiver (observed using the SiRFDemo software):

        /*
         * Carl Carter from SiRF writes: "We do not output on the
         * second (unless you are using MID 52).  We make
         * measurements in the receiver in time with an internal
         * counter that is not slaved to GPS time, so the measurements
         * are made at a time that wanders around the second.  Then,
         * after the measurements are made (all normalized to the same
         * point in time) we dispatch the navigation software to make
         * a solution, and that solution comes out some 200 to 300 ms
         * after the measurement time.  So you may get a message at
         * 700 ms after the second that uses measurements time tagged
         * 450 ms after the second.  And if some other task jumps up
         * and delays things, that message may not come out until 900
         * ms after the second.  Things can get out of sync to the
         * point that if you try to resolve the GPS time of our 1 PPS
         * pulses using the navigation messages, you will find it
         * impossible to be consistent.  That is why I added
         * MID 52 to our system -- it is tied to the creation of the 1
         * PPS and always comes out right around the top of the
         * second."
         */

The function sirf_msg_ppstime() is called when MID 52 is seen on the 
wire but I'm not understanding how that function is adjusting the time 
reported to ntpd and/or the rest of gpsd.  If the message is emitted at 
the second, I would have expected the SHM data to be more stable than 
the navigation time as mentioned in the code comment but it seems to be 
that the opposite is true.

I'm not sure how I can tell if the message is actually working or not. 
The log messages only seem to indicate that this particular section of 
code executed but not whether it was used properly.