[time-nuts] Realtime comparing PPS of 3 GPS

Erik Kaashoek erik at kaashoek.com
Sun May 29 14:45:37 UTC 2022


During evaluation of a GPS module there was some concern over the 
stability/correctness of the PPS and, as the NIST paper on PPS accuracy 
mentions there could be substantial variation in PPS accuracy due to 
atmospheric conditions, an experiment was set up to measure the PPS of 3 
different GPS modules simultaneously.
The experiment uses one NEO-7M and two ATGM336H GPS modules
The PPS of each module is send to 3 separate timer capture inputs of one 
timer running at 200 MHz providing 5 ns measurement accuracy.
The 200 MHz timer was clocked using a 10 MHz Rb output up converted to 
200 MHz using a PLL.
A plot showing the timing variation is attached.
Vertical axis is in ns, horizontal axis is in seconds
All GPS modules had separate simple puck antenna all with almost free 
sky view and all using at least 10 GPS satellites.
The differences between the PPS of the 3 modules was surprisingly large, 
two modules where in average 10 ns apart and a 3rd module was in average 
39 ns apart from the first two modules. More investigation is needed as 
the HW paths between PPS output of the GPS modules and the timer capture 
inputs of the 3 timers are not fully identical
One GPS module showed little variation in PPS timing versus the Rb 
reference, only a slow drift of about 15 ns over 400 seconds. The other 
modules showed a periodic variation of up to 20 ns. The worst variation 
was observed from the NEO-7M module.
The measurement is at time of posting of this message still running to 
see if over a longer period the differences reduce.
If the observed short term (< 400 s)  PPS variation in the order of 
10-20 ns are common and unavoidable it would make sense for the 
requirement for phase stability of a GPSDO to be aligned with this level 
of  PPS phase uncertainty.
Erik.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: Comparing_3_GPS_PPS.JPG
Type: image/jpeg
Size: 56615 bytes
Desc: not available
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20220529/68a9e80b/attachment.jpe>


More information about the Time-nuts_lists.febo.com mailing list