Re: [gpsd-dev] gpsd for time sync on Ubuntu remix too hard for normals - how to improve?

[Greg Troxel]

Hal Murray <address@hidden> writes:

>> I am seeing well within 10 ms with a cheap recent USB dongle, after
>> calibrating the offset. 

To be fair, I have not been rigorous.   But, I calibrated offset, and
left a machine with just the GPS, no network, for two days.  "jitter" as
reported by NTP was low, when checked on 5ish times a time, less than 5
ms.

I then plugged it to the net after the 2 days, and observed the offsets
obtained as it acquired net peers.  They were in a clump from maybe -6
to +1, and the VFAN/gpsd offset was in the middle of that.

> What brand/model of device?

VFAN Model BG-803.

Attaches (NetBSD) as:

  umodem0 at uhub10 port 3 configuration 1 interface 0
  umodem0: u-blox AG - www.u-blox.com (0x1546) u-blox GNSS receiver (0x1a8), 
rev 1.10/3.01, addr 8, iclass 2/2
  umodem0: data interface 1, has CM over data, has no break
  umodem0: status change notification available
  ucom0 at umodem0

ubxtool -p VER returns:

  UBX-MON-VER:
   swVersion: ROM CORE 3.01 (107888)
   hwVersion: 00080000
   extension: FWVER=SPG 3.01
   extension: PROTVER=18.00
   extension: GPS;GLO;GAL;BDS
   extension: SBAS;IMES;QZSS

marketing description says it has "UBX-M8030".


Here's a line where I didn't have the offset dialed in:

+SHM(0)          .GPS.            5 l    1   16  377    0.000   13.707   0.703

But you can see low dispersion for 8 continuous samples.


I get it that it could be low dispersion locally in time and then have
mode changes when something crosses over a boundary and really be
drifting slowly with a 100 ms sawtooth or worse.  But so far I have not
observed that.


Given the speed/latency of USB, with care by the chipset manufacturer to
emit messages at the same time relative to when PPS would have been,
several-ms accuracy should be possible.