[ntp:questions] Help: fudge time2 value for NMEA driver

Brian Inglis Brian.Inglis at SystematicSw.ab.ca
Fri Nov 4 03:59:34 UTC 2016

On 2016-11-03 19:27, Paul wrote:
> On Thu, Nov 3, 2016 at 2:27 PM, Brian Inglis wrote:
>>What do your refclock ntp.conf lines look like?
> I'm not sure why you're asking but:

For the reference and edification of the OP.

> # PPS (ATOM)
> server minpoll 3
> fudge refid GPPS
> # NMEA  @19200
> server minpoll 3 mode 65570 prefer
> fudge time2 0.039 refid FURY

To be fair, I should include mine:

# NMEA driver
#server	127.127.20.?	mode 0=all 1=$GPRMC 2=$GPGGA 4=$GPGLL 8=$GPZDA|$GPZDG 256=$PGRMF
#server	127.127.20.?	mode 0=4800 16=9600 32=19200 48=38400 64=57600 80=115200 bps
#server	127.127.20.?	mode 128=sub-second receive clockstat 65536=extra clockstats
#fudge	127.127.20.?	flag1 pps flag2 fall flag3 kern flag4 noll time1 pps-delay time2 \n-offset

# RPi, GT MKT 3339, kernel PPS
server	prefer minpoll 4 maxpoll 4 mode 17	# mode 17 == $GPRMC 9600bps
fudge	flag1 1 flag3 1 time2 0.500		# pps kernel \n-offset
#fudge	127.127.20.?	3339 pps-delay m*cableE-9+receiverE-9s \n-offset 650/9600s=68ms

# W10, Garmin 18x LVC, user mode PPS
server    prefer minpoll 4 maxpoll 4 mode 17	# mode 17 == $GPRMC 9600bps
fudge    flag1 1         time2 0.500		# pps \n-offset
#fudge	127.127.20.?	18x pps-delay 5*18.2E-9+122.2E-9s=213.2ns \n-offset 680/9600s=71ms

>> Also please see driver comments in [time-nuts] NTP refclock JLT Fury:
> Using the NMEA (like) output is sufficient to number the seconds --
> using an HP SCPI driver is an odd choice for non-HP gear.

The JLT Fury emulates the HP/Symmetricom/Agilent/Keysight 58503 which is a
newer variant of the venerable HP38xx GPS-DOs, and works with compatible
software like HP SatStat, Ulrich Bangert's Z38xx, Graham Baxter's GPScon,
etc. which can provide functions and displays similar to those provided
graphically in Marks Sims' Lady Heather for Trimble Thunderbolt compatibles,
if you are more familiar with those than HP or Motorola series.

NMEA output is just one command which produces both $GPRMC and $GPGGA messages,
with no other selection available, at any requested interval on the serial

>> For decent results you also have to switch off output of other messages.
> That's not my ublox experience although I did arrange that
> all the output per tick happened in less than a second.

Single message output may be required for NMEA PPS handling to work, and why it
did not work with your Fury or uBlox.

IIRC early Fury models use the Motorola Oncore compatible M12+ timing GPS,
and later models use the Motorola/iLotus M12M, compatible with the M12+.
JLTs own latest M12M compatible uses the uBlox M8T, so the LEA-6T performance
can be on par with the Fury, but the latter uses the available Motorola TRAIM
to do PPS sawtooth correction in its uC, and I don't believe that is available
for any NTP driver, nor kernel PPS, either as patches or via config parameters.

Take care. Thanks, Brian Inglis, Calgary, Alberta, Canada

Take care. Thanks, Brian Inglis, Calgary, Alberta, Canada

More information about the questions mailing list