[time-nuts] NTP REFCLOCK for a Jackson Fury??

Chris Albertson albertson.chris at gmail.com
Thu Oct 17 14:54:51 EDT 2013 

This isn't NMEA but many NMEA GPSes work just find and if you read the NMEA
specifications the ascci data only has to be valid for the second in it is
output.  This allows for a random 0.9999...  second offset.

If your ref cock is bothered by a .999 second offset I'd argue the ref
clock software is broken because it makes a wrong assumption.   The ASCII
data is analogous to the old voice telephone service what says "at the tone
the time will be xxx ---- BEEP."  As long as she makes the announcement
before the beep and after the last beep it is perfectly in spec.

If the software depends on an EXACT and CONSTANT offset between the
announcement and the UTC second it is really not doing the correct thing
re. the specs.   But NTP ref clocks do make this assumption because many
times it is correct.   Many GPS receivers are predicable and each sentence
has a common offset.

I think there really is a difference between using two drivers, Atom for
PPS and some other for ASCII data vs. using a combined driver.  The
combined drivers are not forced into making the constant offset assumption
and use the PPS for timing and ASCII for numbering.   But if you look at
the stand alone drivers they use the ASCII for both numbering and timing.
So I argue the best model for your device is a combined driver.

Details like this about NTP ref clocks are best posted to the NTP list.
 The true experts live there




On Thu, Oct 17, 2013 at 11:00 AM, Scott Mace <smace at intt.net> wrote:

> I think the Fury does not guarantee that the PTIME:TCODE will be near the
> 1-PPS leading edge.  When I tried it with the HPGPS driver and a net4501,
> ntpd wasn't happy with it.  I also needed leap second processing to work,
> holdover notification, and to collect other data from it.  The GPGGA
> sentence is on-time even if you are asking the Fury to do other things with
> SCPI commands.  Try setting your Fury hpgps driver fudge flag4 to 1 and I
> bet it will stop working well.  This flag logs the system:print output.
>
>         Scott
>
> On 10/17/2013 10:24 AM, Bill Dailey wrote:
>
>> Interesting.. I see something completely different with my Fury.
>>
>> My hard-coded fudge factor is 0.077 yielding:
>>
>> Every 2.0s: ntpq -p                                     Thu Oct 17
>> 10:20:00
>> 2013
>>
>>       remote           refid      st t when poll reach   delay   offset
>> jitter
>> ==============================**==============================**
>> ==================
>> -four2.fairy.mat 64.250.177.145   2 u   12   64  377   72.493   17.784
>> 0.524
>> -66-162-15-65.li 64.236.96.53     2 u   33   64  377   50.226   15.291
>> 1.181
>> +barricade.rack9 209.51.161.238   2 u   10   64  377   22.908    1.075
>> 0.362
>> -ccadmin.cycores 130.207.244.240  2 u   38   64  377   39.218    3.972
>> 1.241
>>   192.168.1.171   .INIT.          16 u    - 1024    0    0.000    0.000
>> 0.000
>> +SHM(0)          .FURY.           0 l    6 16 377 0.000    0.526
>> 0.034
>> *SHM(1)          .PPSF.           0 l    5   64  377    0.000    0.008
>> 0.001
>> -europium.canoni 193.79.237.14 2 u   32   64  377  116.177   -1.573
>> 0.992
>>
>> So... i could fine tune my fudge some but the point is the jitter is
>> fairly
>> low.  (fury is from the sentences and ppsf is pps from fury).
>>
>> My soekris (soekris is the .171) numbers were much better but it is
>> inoperable for some reason.
>>
>> Doc
>>
>>
>> On Thu, Oct 17, 2013 at 9:59 AM, Tim Shoppa <tshoppa at gmail.com> wrote:
>>
>>  Back in 2003 or so, in the Z3801A with hpgps ntpd refclock driver, I had
>>> to
>>> add a negative offset of -0.98 seconds to the driver's decoding of
>>> PTIME:TCODE? to get it to be right in combination with PPS refclock.
>>>
>>> The documentation in the Z3801A manual correctly described the actual
>>> behavior.... "PTIME:TCODE? Provides timecode message 980 to 20 ms prior
>>> to
>>> 1 PPS of indicated time".
>>>
>>>
>>>
>>> On Thu, Oct 17, 2013 at 9:40 AM, Scott Mace <smace at intt.net> wrote:
>>>
>>>  I posted a patch for the Fury to work with ntpd in Oct 2008.  It uses
>>>> the
>>>> GPGGA output.  The PTIME:TCODE? command is not "on-time" with the 1-PPS
>>>> output on the Fury, so the HPGPS driver does not work.
>>>>
>>>> http://www.febo.com/pipermail/****time-nuts/2008-October/**
>>>> 033901.**html<http://www.febo.com/pipermail/**time-nuts/2008-October/033901.**html>
>>>> <
>>>>
>>> http://www.febo.com/pipermail/**time-nuts/2008-October/033901.**html<http://www.febo.com/pipermail/time-nuts/2008-October/033901.html>
>>> >
>>>
>>>>
>>>> the ntp-fury.diff:
>>>> http://www.febo.com/pipermail/****time-nuts/attachments/**<http://www.febo.com/pipermail/**time-nuts/attachments/**>
>>>> 20081020/846021fe/attachment-****0001.bin<
>>>>
>>> http://www.febo.com/pipermail/**time-nuts/attachments/**
>>> 20081020/846021fe/attachment-**0001.bin<http://www.febo.com/pipermail/time-nuts/attachments/20081020/846021fe/attachment-0001.bin>
>>>
>>>>
>>>>
>>>>
>>>>          Scott
>>>>
>>>>
>>>>
>>>> On 10/16/2013 04:44 PM, Frank Hughes wrote:
>>>>
>>>>  Hi,
>>>>>
>>>>> What NTP REFCLOCK can be used for a Jackson Fury?
>>>>>
>>>>> I know that the Jackson Fury docs suggest using:
>>>>> http://www.realhamradio.com/****gpscon-info.htm<http://www.realhamradio.com/**gpscon-info.htm>
>>>>> <
>>>>>
>>>> http://www.realhamradio.com/**gpscon-info.htm<http://www.realhamradio.com/gpscon-info.htm>
>>> >
>>>
>>>>
>>>>> But that means I would have to put up a windows server
>>>>> to replace the FreeBSD ntpd server I built for use w/ the Trimble TB.
>>>>>
>>>>> Looking for open source options, if possible.
>>>>>
>>>>> Thanks,
>>>>> Frank
>>>>> KJ4OLL
>>>>> ______________________________****_________________
>>>>> time-nuts mailing list -- time-nuts at febo.com
>>>>> To unsubscribe, go to https://www.febo.com/cgi-bin/****<https://www.febo.com/cgi-bin/**>
>>>>> mailman/listinfo/time-nuts<
>>>>>
>>>> https://www.febo.com/cgi-bin/**mailman/listinfo/time-nuts<https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts>
>>> >
>>>
>>>> and follow the instructions there.
>>>>>
>>>>>   ______________________________****_________________
>>>>>
>>>> time-nuts mailing list -- time-nuts at febo.com
>>>> To unsubscribe, go to https://www.febo.com/cgi-bin/****<https://www.febo.com/cgi-bin/**>
>>>> mailman/listinfo/time-nuts<
>>>>
>>> https://www.febo.com/cgi-bin/**mailman/listinfo/time-nuts<https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts>
>>> >
>>>
>>>> and follow the instructions there.
>>>>
>>>>  ______________________________**_________________
>>> time-nuts mailing list -- time-nuts at febo.com
>>> To unsubscribe, go to
>>> https://www.febo.com/cgi-bin/**mailman/listinfo/time-nuts<https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts>
>>> and follow the instructions there.
>>>
>>>
>>
>>
>>  ______________________________**_________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/**
> mailman/listinfo/time-nuts<https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts>
> and follow the instructions there.
>



-- 

Chris Albertson
Redondo Beach, California

More information about the time-nuts mailing list