[ntp:questions] Sure GPS - Very High Jitter and Offset

Miguel Gonçalves mail at miguelgoncalves.com
Sat Aug 6 16:19:31 UTC 2011


Hi Ken!

2011/8/5 Ken Link <klink at numberzero.org>:
> Hi Miguel,
>
> Thanks for the response. I'm curious about your ntp.conf. I've looked
> at the definitions of the various parameters but I'm not sure how they
> translate to a Windows platform, that's why I didn't set any of them.
> I also see that you have hardcoded a 0.400s offset. How did you
> determine this value?

When I started NTP without this value and with several noselect
servers in the ntp.conf file I saw NTP lock 400 ms away from the
correct time. I am know testing my servers against known stratum 1 NTP
servers to validate and I am getting this:

[root at backup ~]# ntpq -pn
     remote           refid      st t when poll reach   delay   offset  jitter
==============================================================================
+10.0.2.10       .GPS.            1 u   17   64  377    0.154    0.066   0.019
*10.0.2.9        .GPS.            1 u   47   64  377    0.162    0.038   0.036
 17.72.133.54    .GPS.            1 u   14   64  377   67.494    1.206   2.543
 17.72.133.55    .GPS.            1 u    1   64  377   68.454    1.378   1.590
 194.117.9.130   194.117.9.138    2 u    9   64  377   22.638    0.872   2.340
 192.53.103.108  .PTB.            1 u    -   64  377   72.495   -2.460   1.854
 150.214.94.5    .GPS.            1 u   12   64  377   93.473    6.060   0.852
 145.238.203.14  .TS-2.           1 u    6   64  377   53.768    1.514   2.361
 193.204.114.232 .CTD.            1 u    -   64  377   64.828    1.798   0.755
 162.23.41.34    .HBGi.           1 u    5   64  377   72.297   -2.258   1.442

My internal servers are the first ones and they are on the same switch
as this server.

> I might still have ACPI enabled on my machine, I'll check that after work today.

Good.

> The module is great at locking onto the various GPS satellites. I've
> just thrown the antenna on my bed in my apartment, and I can see
> plenty of satellites, even on the first floor of a three story
> apartment complex.

I find it very sensitive also. I am thinking about buying a Resolution
T from Trimble but I am afraid it might be not so sensitive as Sure
is.

> I'm curious, would you be able/willing to disable PPS and restart
> ntpd? I'd just like to see how accurate yours can get, without PPS. I
> didn't expect mine to be as jittery as I'm seeing and I can't imagine
> the PPS even working correctly if there is this much jitter on the
> line already.

As Chris Albertson said, it is pointless as there's no guarantee that
the offset from the NMEA sentence to the start of the second stays the
same. Just check again my validation tests above. Without PPS I guess
you couldn't reach less than 100 us accuracy over a LAN.

> Thanks,
> Ken

No problem!

Hope this helps!

Cheers.
Miguel

>
> 2011/8/5 Miguel Gonçalves <mail at miguelgoncalves.com>:
>> Hi Ken!
>>
>> I am using the same board on a FreeBSD 7.4-p2 machine. I'll tell you
>> about my experience to see if that helps.
>>
>> I made the patches as advised by David Taylor and it is working great.
>> I also disabled the FIFO and ACPI.
>>
>> Here's a startup from loopstats:
>>
>> 55778 43424.602 0.000000000 46.912 0.000008833 0.000004 4
>> 55778 43440.601 0.000157002 46.534 0.000100345 0.133775 4
>> 55778 43456.602 0.000012943 46.444 0.000009039 0.129096 4
>> 55778 43472.602 0.000003100 46.576 0.000001060 0.129432 4
>> 55778 43488.602 0.000000931 46.503 0.000000383 0.123788 4
>> 55778 43504.601 0.000000426 46.514 0.000000497 0.115858 4
>> 55778 43520.602 0.000000515 46.500 0.000000221 0.108485 4
>> 55778 43536.602 0.000000312 46.541 0.000000155 0.102534 4
>> 55778 43552.602 0.000000028 46.541 0.000000255 0.095911 4
>> 55778 43568.601 0.000000254 46.532 0.000000167 0.089774 4
>> 55778 43584.602 0.000000622 46.532 0.000000213 0.083976 4
>> 55778 43600.602 0.000000633 46.559 0.000000198 0.079125 4
>> 55778 43616.602 0.000000794 46.559 0.000000151 0.074015 4
>> 55778 43632.601 0.000000655 46.582 0.000000175 0.069708 4
>> 55778 43648.601 0.000000599 46.582 0.000000254 0.065206 4
>> 55778 43664.602 0.000000598 46.592 0.000000166 0.061091 4
>> 55778 43680.602 0.000000919 46.592 0.000000350 0.057146 4
>> 55778 43696.601 0.000001334 46.592 0.000000184 0.053455 4
>> 55778 43712.601 0.000001513 46.592 0.000000201 0.050003 4
>> 55778 43728.602 0.000001520 46.622 0.000000216 0.047974 4
>>
>> As you can see I quickly get to nano-second performance. I believe 1.5
>> us is normal and to be expected. Heck, I have some screenshots of a
>> Meinberg advertising 18 us offset from UTC.
>>
>> The corresponding clockstats:
>>
>> 55778 43424.603 127.127.20.0
>> $GPGGA,120344.000,4055.2197,N,00829.6109,W,1,8,1.05,254.5,M,51.2,M,,*46
>> 55778 43440.601 127.127.20.0
>> $GPGGA,120400.000,4055.2197,N,00829.6108,W,1,9,0.99,254.6,M,51.2,M,,*46
>> 55778 43456.602 127.127.20.0
>> $GPGGA,120416.000,4055.2197,N,00829.6107,W,1,8,0.95,254.7,M,51.2,M,,*42
>> 55778 43472.602 127.127.20.0
>> $GPGGA,120432.000,4055.2196,N,00829.6106,W,1,8,1.06,254.9,M,51.2,M,,*41
>> 55778 43488.602 127.127.20.0
>> $GPGGA,120448.000,4055.2194,N,00829.6104,W,1,8,1.07,255.2,M,51.2,M,,*47
>> 55778 43504.601 127.127.20.0
>> $GPGGA,120504.000,4055.2190,N,00829.6100,W,1,8,1.00,256.1,M,51.2,M,,*49
>> 55778 43520.602 127.127.20.0
>> $GPGGA,120520.000,4055.2183,N,00829.6094,W,1,8,0.94,257.6,M,51.2,M,,*4B
>> 55778 43536.602 127.127.20.0
>> $GPGGA,120536.000,4055.2172,N,00829.6084,W,1,8,1.00,259.7,M,51.2,M,,*40
>> 55778 43552.602 127.127.20.0
>> $GPGGA,120552.000,4055.2160,N,00829.6073,W,1,9,1.00,262.1,M,51.2,M,,*46
>> 55778 43568.601 127.127.20.0
>> $GPGGA,120608.000,4055.2148,N,00829.6062,W,1,9,1.00,264.4,M,51.2,M,,*43
>> 55778 43584.602 127.127.20.0
>> $GPGGA,120624.000,4055.2149,N,00829.6063,W,1,8,1.09,265.1,M,51.2,M,,*41
>> 55778 43600.602 127.127.20.0
>> $GPGGA,120640.000,4055.2150,N,00829.6064,W,1,8,1.02,265.2,M,51.2,M,,*44
>> 55778 43616.602 127.127.20.0
>> $GPGGA,120656.000,4055.2150,N,00829.6064,W,1,8,1.10,265.2,M,51.2,M,,*40
>> 55778 43632.601 127.127.20.0
>> $GPGGA,120712.000,4055.2151,N,00829.6065,W,1,8,1.10,265.0,M,51.2,M,,*43
>> 55778 43648.602 127.127.20.0
>> $GPGGA,120728.000,4055.2151,N,00829.6066,W,1,8,1.10,264.9,M,51.2,M,,*41
>> 55778 43664.602 127.127.20.0
>> $GPGGA,120744.000,4055.2152,N,00829.6066,W,1,8,1.26,264.8,M,51.2,M,,*4C
>> 55778 43680.602 127.127.20.0
>> $GPGGA,120800.000,4055.2152,N,00829.6067,W,1,8,1.27,264.7,M,51.2,M,,*4C
>> 55778 43696.601 127.127.20.0
>> $GPGGA,120816.000,4055.2153,N,00829.6068,W,1,8,1.10,264.6,M,51.2,M,,*40
>> 55778 43712.602 127.127.20.0
>> $GPGGA,120832.000,4055.2152,N,00829.6068,W,1,8,1.01,264.6,M,51.2,M,,*47
>> 55778 43728.602 127.127.20.0
>> $GPGGA,120848.000,4055.2151,N,00829.6067,W,1,8,1.01,264.6,M,51.2,M,,*46
>>
>> This is at a window that has quite a few obstacles outside. I am
>> getting 8, 9 satellites!
>>
>> By the way, I am using the board at 9600bps and with the PPS signal
>> connected to the DCD pin.
>>
>> My suggestion for you is try to make the patch advised by David Taylor.
>>
>> Here's my NTP configuration:
>>
>> # PPS & NMEA
>> server 127.127.20.0 mode 18 prefer minpoll 4
>> fudge 127.127.20.0 flag1 1 flag2 0 flag3 1 time2 0.400
>>
>> Hope this helps!
>>
>> Cheers,
>> Miguel
>>
>> On 5 August 2011 03:09, Ken Link <klink at numberzero.org> wrote:
>>> My Sure Electronics GPS finally arrived a few days ago, and I'm having
>>> some trouble getting it stabilized. I'm running Windows 7 x64, with
>>> the latest serialpps.sys driver (20110720), NTP 4.2.6p3 compiled on
>>> the same machine. I also compiled and ran the code from an email chain
>>> back in April to reprogram the module to only send $GPGGA - which it
>>> does. It's set to 115200bps now, although it didn't appear to have any
>>> effect on jitter/offset from 9600, 19200, 38400, and 115200 that I can
>>> tell. I've turned off UART buffers to the COM port. I haven't soldered
>>> a wire to get PPS to the serial port yet, but I plan to once I get it
>>> stabilized.
>>>
>>> ntp.conf:
>>> driftfile "C:\Program Files (x86)\NTP\etc\ntp.drift"
>>> server 127.127.1.0
>>> fudge 127.127.1.0 stratum 12
>>> server 127.127.20.1 mode 82 minpoll 4 prefer
>>> server 127.127.22.1 minpoll 4
>>> server 0.us.pool.ntp.org iburst
>>> server 1.us.pool.ntp.org iburst
>>> server 2.us.pool.ntp.org iburst
>>> server 3.us.pool.ntp.org iburst
>>> server ntp.numberzero.org iburst
>>> peer 192.168.1.3 iburst
>>> peer 192.168.1.4 iburst
>>> enable stats
>>> statsdir "C:\Program Files (x86)\NTP\etc\"
>>> statistics loopstats
>>>
>>> The problem is the offset for the Sure GPS consistently varies between
>>> -250ms and -350ms, and the jitter roughly between 5ms and 35ms.
>>> Obviously, ntpd rejects it and picks something from the pool instead.
>>> Any ideas why it could be so unstable/inaccurate? I must be forgetting
>>> something obvious. Here is my ntpq -p: http://i.imgur.com/gNURx.png
>>>
>>> Thanks in advance!
>>> _______________________________________________
>>> questions mailing list
>>> questions at lists.ntp.org
>>> http://lists.ntp.org/listinfo/questions
>>>
>>
>



More information about the questions mailing list