w3iwi at toad.net
Wed Aug 3 13:15:49 EDT 2005
Rick Putz asked
Has anyone done any studies on the Nav Man Jupiter T and Pico T series receiver
s? They use the SiRF chip set and provide a 10 Khz output that is said to be co
herent with the 1PPS.
I did a lot of testing on the Jupiter-T that has an ONCORE VP physical
footprint and MOT compatible command set.
First of all, it is NOT the SiRF chipset. As I understand it,
Collins/Rockwell sold their semiconductor operation lock, stock &
silicon to the C/R spinoff Connexant, and it was Connexant that built
the Jupiter-T with C/R parts and knowledge. Then Connexant's GPS was
sold to SiRF, except that the Jupiter C/R series was sold to NavMan in
So, under the assumption that the Jupiter-T you mention is the ONCORE
footprint that I tested, here are a few answers. As a GPS receiver,
the Jupiter-T is quite good, and better in fact than the old ONCORE
VP. It is a 12 channel unit (ONCORE=8) so it can track more satellites
and seemed to me to have better RF sensitivity. The Jupiter-T 1PPS
timing is derived from a ~40 MHz clock, so that the sawtooth is only
25 pk-to-pk nsec in amplitude (ONCORE=9.5 MHz with 104 nsec pk-to-pk).
The 10kPPS is coherent with the 1PPS and has the same 25 nsec pk-to-pk
sawtooth jitter. Both of these differences were reduced in the more
However, the Jupiter-T lacks the serial data output that tells the
sawtooth error on the next pulse (ONCORE has message with +/- 127 nsec
in 1 nsec steps). As I understand it, the MOT's generate 1PPS via a
looonnnggg (> 1 second) counter that counts the rcvr clock and the
quantization error can be reported. The Jupiter-T generates its 1 &
10k PPS thru an NCO, and the starting phase is not stored in the
receiver; for this reason they never tried to reconstruct the error.
I built a 10 MHz locker using the Jupiter-T and TvB's PIC-based
synchronous counter with the idea that the 10 kHz jitter could be
averaged out. The only problem came when the Jupiter-T went thru one
of its clock zero-beat times (TvB calls them "hanging bridges") when
the oscillator being locked got pulled off frequency. Alas, I
abandoned that project!
More information about the time-nuts