[ntp:questions] ESR looking for good GPS clocks

Chris Albertson albertson.chris at gmail.com
Tue Mar 6 19:30:05 UTC 2012


On Tue, Mar 6, 2012 at 10:54 AM, unruh <unruh at invalid.ca> wrote:

> If the report came a set time after the determination, that would not
> matter. Ie, if the nmea sentence started exactly 573.124ms after the
> determination, it would not matter that it reported late. ntpd could
> fudge that offset away. If that time varies by 170ms however, tht is a
> disaster.

The ONLY requirement is that the NMEA sentence be sent during the
second to which it applies.   According to NMEA that is all that is
required.     But in real life one might get lucky and find the
sentence has a fixed offset but this is not required by NMEA.     I
think to many people depend on NMEA being far better than the spec
requires.

the PPS on the other hand is spec's to be right on the UTC second
(within some given 1 sigma error bar)



We don't know how GPSes work internally.   But I assume every so
offten some extra work is required, like, say if a Sat. it was
tracking is blocked by a tree and the GPS needs to make a decision
about what to do.   It might, maybe take a few mS and delay the NMEA
data.

Also the data in the GPS is binary floating point and NMEA is ASCII.
It is likely the time to do the conversion depends on the numeric
values that are being converted

In cases I've seen the engineer was smart enough to pad ASCII number
with leading zeroes so that sentences always take the same time to
transmit down the cable.  But the logic used to assemble the messages
varies based on the numeric value.   The "computer" is not a 4GHZ quad
core.   It is likely a low power uP that is only "just fast enough"

The time to compute a time or nav solution might depend on the
possition of the sats.  the uP likely uses some iterative method that
iterates until  the error is below some threshold.   Then the NMEA
data are output.   the number of required iterations might vary slowly
over a 12 hour period

After all this the only spec the engineer absolutly had to meet was
"the NMEA data shall be output within the second to which it applies".


Chris Albertson
Redondo Beach, California


More information about the questions mailing list