[time-nuts] A real world project need for timing accuracy...

Hal Murray hmurray at megapathdsl.net
Tue Nov 2 04:10:18 UTC 2010


mpb45 at clanbaker.org said:
>    However, time-of-flight info (via an RF link requires
>    that I sync the 10MHz clocks at both ends together.
>    Use of GPS receivers seems to be the most likely way
>    to do this but how do I keep the cost and complexity
>    down? 

What level of accuracy do you need?

I think GPS will do what you want.  You can get one of the mouse/hockey-puck 
units for well under $100.   I think you need one at each end so that doubles 
the cost.  Be sure to avoid the USB units because they don't have PPS signals.

Option 2:  How long can you coast?  (holdover mode)  If your crystal is off 
by 1E-6 and you coast for 1000 seconds, you will be off by 1 ms.  Is that 
good enough?  You can get better accuracy with better hardware and/or better 
software.  By software I mean 1) correcting for the initial calibration 
error, and 2) correcting for the temperature.  [It would be fun to see what 
you could do starting with an inexpensive oscillator package.]  You can 
double-check and/or get more data for correction if you bring the unit back 
to the base station at the end of the run and see how far it has drifted.

Option 3:  Use a bi-directional RF link.  Send a "PPS" signal to the gear at 
the target.  I think you will have to calibrate out the delays through the RF 
links but that should be simple if you assume symmetry.

What sort of RF gear are you considering?  Are the normal units both transmit 
and receive, or do you have to buy another pair for the other direction?


Option 4: Send the PPS signal upstream and compensate for the clock drift at 
the upstream end.  Again, I think you may have to calibrate for the delays 
through the RF link which may vary slightly with distance and antenna setup.



PS: How accurately do you measure the distance?


-- 
These are my opinions, not necessarily my employer's.  I hate spam.







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