So I have read that you need 4 satellites to be able to get a proper position from your GPS. But how many do you need if you are only interested in the 1pps signal from them? And what effects play a role for the precision of this signal?

(I am using a 1pps input from a GPS receiver for an atomic rubidium clock)

1 Answer 1


If you know the exact position of the receiver then it is possible to do timing with only one satellite. Essentially GPS is solving four unknowns (ECEF position, and receiver clock error) with four (pseudo-range) distance measurements, so if you can provide three unknowns, you only need one measurement.

Not all receivers support this though - it was a standard feature in timing receivers the last time I did a survey, several years ago. If your receiver (or integrated implementation) does not support it, you will need at least four satellites (three for the ECEF unknowns, and one for the receiver clock error).

The biggest contributions to timing error are the same as the contributions to distance errors (since its the same measurement set) - multipath and ionospheric variation. A good antenna, good antenna placement and increasing the masking angle will all help to reduce error.

More GPS measurements (including knowledge of your antenna location, which is really what you are measuring with GPS) can give you an over-determined solution, which can reduce error using least-squares fit. However at this level, you need to work out how much you are really gaining in output (Allen variance), and whether you see time steps when you swap between satellites - that can mess up the integration.

  • "If I know the position of my receiver" Where Would use that in prior information? I know my position ofc, but I have not informed my program the position (And tbh I have no idea how to do that). In the case I dont inform my program the position, how many sattelites is then needed? Is it then also just 3 as with the position?
    – Nillo
    Commented May 22, 2016 at 11:25
  • Whether you can do it, and how you do it, is implementation specific. I updated the answer to specify four satellites for an general GPS solution. To be honest, I assumed this was a theoretical question, because you will almost always have enough satellites for a solution, and you want to "coast" your timing solution on the rubidium clock in any case, because GPS measurements have much higher white noise levels.
    – BradHards
    Commented May 22, 2016 at 22:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.