[ntp:questions] fudge time1 for gps-18x-LVC?

David Lord snews at lordynet.org
Fri Feb 5 23:10:58 UTC 2010


unruh wrote:
> On 2010-02-05, David Lord <snews at lordynet.org> wrote:
>> Hi
>>
>> Now I have gps-18x-LVC nmea 127.127.20.0 along with
>> parallel port PPS using pps0 at ppbus0.
> 
> What sentences is the nmea sending out?
> Why not use something like gpsd and the shm driver to combine the nmea
> and the PPS into one very accurate time stamp?
>  

I'll check after I've enough data points. Unless there's
some method I can use to check whilst delivering time. I'd set
it for RMC when I used it last year. Ntp.conf still has setting
"server 127.127.20.0 mode 1"

>> This seemed to work ok until pps timestamps suddenly jumped to
>> -350ms then PPS became false then deselected.
>>
>> To correct for this I've set fudge "127.127.20.0 time1 0.65"
>> also "tos mindist 0.4"
> 
> That sounds far too big. Even at the slowest serial spped it takes a lot
> less than that for a single nmea sentence to come it. 

Exactly

> 
>> Somehow I think I might be in sync with wrong edge of PPS.
> 
> Why?
> 

That would put pps out, but I'd have guessed by only 100ms
if that's default.

>> Eventually I've got an ethernet cable running upstairs
>> (3 x long cables + 2 couplers).
> 
> Depends on what you call long. 10m should easily  "get it upstairs" but
> 10 m is not that long. 

More like 20m, 3 x lengths joined with couplers.

> 
>> I've added 3 x local servers sync to MSF + public servers
>> and marked noselect.
>>
>> About 30 min later "ntpq -p" has:
>> source reach offset(ms)
>>
>> + GPS   377  23.247
> 
> Not surprizing. NMEA takes a long time delivering. 
> 
>> o PPS   377  -0.298
>>    serv1 377  -1.295
>>    serv2 INIT  0.000
>>    serv3 377  -0.091
>>
>> I suspect reason for INIT is both serv1 and serv3 have serv2
>> selected. The GPS offset has been varying +/- a lot but pps
>> still seems to be converging.
> 
> ??? What serv1 and 2 do should not matter. What are their stratums?

serv1 = 2, serv2 = 1, serv3 = 2

> 
>> That looks ok to me except only Google hit for gps18x-lvc +
>> "fudge time1" gives setting of 0.0001
>>
>> What does anyone else use for this fudge time1?
> 
> It depends on how many nmea sentences are being collected.
> 
>> I'm using default nmea strings as garmin loses its settings
>> otherwise whilst not powered up.
> 
> That is a mistake. YOu only need the one. And the garmin does not loose
> its settings -- it has an onboard battery. Besides the software should
> reset the system when you power up.

In that case it's still using RMC and mode 1 in server line is
ok. I should have checked output via minicom that it was same
and had also forgotten I'd left the "mode 1" in server line.

I now suspect answer is that nmea uses pps on dcd when present
otherwise my results don't make sense.

Before connecting to parallel port I'd left it for a day with
PPS 127.127.22.0 driver and PPS to DCD at serial port. That
was giving eg.

pps0 -> /dev/tty00

Feb 1, 2010
server 127.127.20.0 mode 1 prefer
server 127.127.22.0

after 30 minutes:

    remote    refid st t  reach offset jitter
+GPS_NMEA(0) .GPS.  0 l   377  -4.802  0.288
oPPS(0)      .PPS.  0 1   377  -4.784  0.286

With same setup last summer PPS eventually converged
to around 2us.

Feb 2, 2010
With /dev/pps0 lpt0 and PPS connected to both serial DCD
and parallel NACK

server 127.127.20.0 mode 1 prefer
server 127.127.22.0

after 30 minutes:

    remote    refid st t  reach offset jitter
+GPS_NMEA(0) .GPS.  0 l   377  -7.597  1.182
oPPS(0)      .PPS.  0 1   377  -7.619  1.188

after 24 hours:

    remote    refid st t  reach offset jitter
+GPS_NMEA(0) .GPS.  0 l   377  -0.016  0.006
oPPS(0)      .PPS.  0 1   377  -0.014  0.002


after 44 hours:

    remote    refid st t  reach offset jitter
+GPS_NMEA(0) .GPS.  0 l   377  -0.004  0.002
oPPS(0)      .PPS.  0 1   377  -0.005  0.002

The nmea driver does take pps on serial dcd into account if present?


Feb 4, 2010
With /dev/pps0 lpt0 and PPS connected only to parallel NACK
tos mindist increased from 0.02 to 0.4 and fudge time1 added.
Ntp pool serv1 and serv2 and fileserver serv3, with noselect.

server 127.127.20.0 mode 1 prefer
fudge 127.127.20.0 time1 0.651
server 127.127.22.0

after 30 minutes
    remote    refid   st t  reach offset jitter
+GPS_NMEA(0) .GPS.    0 l   377  22.581  5.356
oPPS(0)      .PPS.    0 1   377  -0.486  0.145
  serv1       serv2    2 u   377  -1.295  0.741
  serv2       .INIT.  16 u     0   0.000  0.000
  serv3       serv2    2 u   377  -0.091  1.105

after 10 hours
    remote    refid   st t  reach offset jitter
+GPS_NMEA(0) .GPS.    0 l   377  29.189 21.691
oPPS(0)      .PPS.    0 1   377  -0.009  0.004
  serv1       serv2    2 u   377   1.144  0.526
  serv2       .INIT.  16 u     0   0.000  0.000
  serv3       serv2    2 u   377  -0.018  1.740


David (dazed and confused)




More information about the questions mailing list