[time-nuts] 1PPS accuracy of commercial GPS receivers
Magnus Danielson
magnus at rubidium.dyndns.org
Wed May 13 21:33:54 UTC 2009
Philip Pemberton skrev:
> Just out of idle curiosity, is there any significant difference in 1PPS
> accuracy between different GPS modules?
>
> I've got a pair of Trimble SVeeSix CM3 boards (firmware 4.13 if memory
> serves, have to be reflashed to change the comm protocol, which can be
> either TSIP or NMEA) and an Axiom Sandpiper (SiRFStar II, RAM only, no
> onboard NVM, SiRF Binary or NMEA switchable on-the-fly) which are
> specified at 1us and 40ns accuracy respectively. As a comparison point,
> I've also been looking at the Fastrax iTrax321 (IT321) which is a
> solder-down "micro-GPS" module based on the "20-channel" SiRFStar III
> and is -- like the Trimble -- specced at 1us accuracy. This is one of
> the newer SiRFStar III based design.
>
> Am I missing something blindingly obvious here, or is there really that
> much spread in 1PPS accuracy on commercial receiver boards?
>
> Is 1us jitter really that good for a GPS module?
The 1 us figure is a historic figure relating to a worst case degrades
GPS constellation situation when the 24 sat constellation has degraded
significantly etc. This number comes out of ICD-200. A more commonly
referred figure is 340 ns which is what the GPS constellation with SA
enabled. In the SA-disabled world seeing lower numbers as 60-40 ns is
not unreasnoble. Old 6 or 8 channel receivers was adequate for the older
constellation situation, but seing 10-12 sats in todays world is not
unreasnoble and naturally will the bias effects and other noise
processes be lower. In addition has receiver technology advanced to
better suppress various imperfections such as multi-path, weak signals
and quick locking. In the other end, awareness of how the PPS is being
used have improved how the PPS signal is generated and producing
"sawtooth corrections" enables lower time quantization noise.
Old receivers can perform better in todays world, so we could modernize
the specs by comparing them with newer boards in todays environment.
I am not sure if you really got a real answer, but hopefully may some of
the difference become explainable to some degree.
Cheers,
Magnus
More information about the time-nuts
mailing list