[ntp:questions] Single GPS/PPS time source gets marked as a falseticker

unruh unruh at wormhole.physics.ubc.ca
Thu Jul 14 18:14:19 UTC 2011


On 2011-07-14, Chris Albertson <albertson.chris at gmail.com> wrote:
> On Thu, Jul 14, 2011 at 7:04 AM, unruh <unruh at wormhole.physics.ubc.ca> wrote:
>
>> Why? That native binary will also suffer from large latency (the serial link
>> is slow no matter what is being sent over it) and probably also fairly
>> large jitter. And it suffers from the huge downside of your having to
>> learn a whole new language.
>
> Those ASCII characters have to be assembled from internal binary data.
>  Binary to ASCII conversion is not a fixed time process.  for example
> the software might convert "3" faster then "57".   And then some
> GPSes, I don't know about yours don't transmit leading zeroes so that
> sometimes the NEMA sentance is longer of shorter depending on the data
> values.
>
> Finally ASCII is MORE BYTES.  so yes both are sent at 1200 bps but
> less data are sent if it does not need to be in ASCII and you avoid
> the conversion time.  For example a floating point value is four bytes
> in binary and 8 or 9 in ASCII and therefor take half the time to send
> and zero time to convert.

You  are talking about a few ms in 600ms. Who cares? And conversion time
is in the usec range. 
 
>
> About the PPS being "dead on".  How did you compare the PPS to the UTC
> second?  I think you meant the period was one second as close as you
> could see by eye on the screen but the phase was unknown.    At some
> point you will need to figure out the "fudge" number of the PPS
> otherwise your NTP time could be off.

That "fudge" number is in the ns range. We are talking ms. 

>
> I'm not surprized that this is hard.  You are setting up a stratum 1
> NTP server in the ocean using some low-end equipment.  That is outside
> of what people normally do and you'd expect to do some problem
> solving.

??? Exactly which of the problems he mentioned had anything to do with
"setting up ntp server in the ocean"? The questions he asked were all
extremely common questions even for a land based ntp stratum 1 (The only
difference he has is that stratum 1 is all he can do. He has no access
to anything else. But then that is also true for someone in a commercial
establishment who is not allowed any internet connections to the outside
world for security reasons.) Ie, his problems are all extremely common
problems that have been solved thousands of times. 
>

He never told us what his timing requirements are, and certainly usec
timing is very possible. (well, the accessibility of GPS in the midst of
a north Pacific hurricane might be doubtful). 
If he is running linux on these buoys, then he might also look at
chrony, which might have better long term stability if you lose
connectivity, but the key problem would be temperature variations in the
computer crystal. If he is worried about long losses of connectivity,
then using a temperature compensating ntp might be a way to go and would
probably give a vastly improved stability with long periods of
disconnection. 





More information about the questions mailing list