[time-nuts] GPS Timing receivers

[Magnus Danielson]

From: "Tom Van Baak" <tvb at leapsecond.com>
Subject: Re: [time-nuts] GPS Timing receivers
Date: Thu, 6 Oct 2005 14:29:58 -0700
Message-ID: <000e01c5cabd$1b2582c0$3915f204 at computer>

> > Hi everybody,
> >
> > I'm a final year student in TLC engineering, from the Polytechnic of
> Turin.
> > The subject of my thesis is the GPS Timing Receivers in general.
> 
> Congratulations.
> 
> > My main interests are:
> > 1) Evalutation of GPS receivers performances, therefore I'm interested
> > in test procedures.
> > In this connection I've read the articles about the Test Bed at USNO
> > in order to evaluate the M12+.
> 
> Yes, that's a good one.
> 
> > My questions are:
> > - it is correct to evaluate a gps timing receiver just on the basis of
> >their 1pps?
> 
> It depends very much on the application. Sometimes
> evaluations include things like weight, size, power
> requirements, acquisition time, price, TRAIM quality,
> temperature coefficient, etc. But plain 1PPS is a good
> start since it is the key feature that distinguishes a
> GPS timing receiver from a normal GPS receiver.

I agree. There are *many* different issues to consider, but assuming that other
operating conditions is fair, good or even optimum, the PPS will be a good
place to start.

> Then there are a host of issues with evaluating
> antennas, which although not part of the GPS
> receiver, can be important to the quality of the
> 1PPS you are measuring.
> 
> > - it is possible to test a 1PPS using a TIC without a reference clock?
> > - can you suggest me material (articles, proceedings and so on) about
> > this kind of tests.
> 
> I think you need a reference. But for most 1PPS
> analysis a rubidium standard and a 1 ns TIC is
> sufficient.

You don't need a timing reference per se, but it simplifies the analysis alot.
You can compare the signal against itself using either a delayline or by
using the TIC timebase as your resolving time delay. For both solutions the
result will depend on the stability of the delay or the time base for the
region of interest. For longer measurement times, having a good reference
signal is really the best solution in reality, unless you find a stable delay
you can count on.

> > 2) Timing Receivers architectures: I'm mainly interested in
> > understanding how the 1pps is
> > generated within the receiver. I've read a lot of IEEE docs, patents,
> > manuals and application
> 
> This is further complicated since there are many
> different receiver architectures and methods to
> recover the timing. Look for subjects like single vs.
> dual frequency (L1/L2), code vs. carrier phase,
> real-time vs. post processed, precise orbits, all
> in view vs. common view, etc.

Indeed.

> > notes, but I haven't understood yet some details (maybe they are
> > proprietary information!).
> > Could you indicate to me any material that can be useful to my research?
> 
> If anyone finds some good articles let me know too!

Well, the first book I should recommend is really "Understanding GPS - 
Principles and Applications" with Elliott D. Kaplan as editor. Read it.
This is not the authorative text on all aspects (I can give further reading
tips which is) but it gives a very good and actually quite advanced level of
GPS knowledge in the size there is. Now, in there (to tired to dig up the
detailed reference position) is the description of how the PPS is generated
and really from what basis in the solution. I think one should not really dig
into the more advanced material until one have read this book fairly thorowly,
it is well spent time, since things intertangle in an intricate dance which at
first takes some time to apprechiate even if one has got the initial overview
message.

Even if I have more advanced texts, I find that I use this one as a basic
reference for a quick refresh on the basics.

However, to come back to your problem. What the GPS receiver do is really
measure the phase state of the signal as received from a number of satelites at
its (or actually, the antennas) position. The carrier phase is tracked in a
loop, the C/A or P code is tracked in a code loop and the 50 bps message code
helps to create a channel for additional information as well as system time for
the GPS signal as it was transmitted from the satellite. We can then cancel out
the three unknowns in the physical position and the unknown in the local time
offset. Since we correlate the measurements to the local clock, we can now use
the time-position given to correct time such that the timing reference internal
to the node tracks the GPS system. From this it is trivial to have a corrected
PPS generated. This is the big wavey arm kind of description, there is more in
the book, but in the end there is not very much magic in it. You have a clock
phase which is running and you correct its frequency and phase to align. It is
samples in the hardware receiver so we can correlate the system time with the
received signal. The GPS solution gives the errors and we correct from that.

Cheers,
Magnus