[time-nuts] 1 pps accuracy to UTC

[Tom Van Baak]

Martyn,

1)
GPS receiver timing accuracy is usually quoted as RMS,
so you need to average for a while before you worry about
matching any claimed accuracy or comparing raw 1pps
outputs. At least 100 samples; maybe 5 minutes or more.

I assume you've done the obvious thing like set both cable
delays to the same ns value. Or zero just to be sure. What
software are you using to configure your M12+ and M12M?

Using that software also check your SV list and signal levels
to verify that both receivers are seeing the same or almost
the same satellites and at similar signal levels.

Make sure both boards have the exact same gps position.

Make sure both boards are in zero-D timing mode.

If the boards aren't configured exactly the same and reporting
almost the same SV signals, look into that first before you
use your SR620.

You should not be seeing 50 ns of jitter out of an M12. To
me this sounds like a M12 configuration or an error in your
test setup.

See typical M12 1PPS (sawtooth) plots:
http://www.leapsecond.com/pages/m12/sawtooth.htm

With averaging the jitter for a plain M12 should go down
to 9 ns rms; with sawtooth correction, down to 3 ns. See:
<http://www.mail-archive.com/time-nuts@febo.com/msg00172.html>
<http://www.leapsecond.com/pages/gps4/cns-1pps.gif>

2)
When you're down at the ns level, have a look at the 1PPS
signal going into the SR620 inputs. If you're not careful
with buffers and loads you'll gets tens of ns of error just in
mis-setting your trigger levels. Put BNC tee's on the SR620
inputs, use a fast 'scope with high impedance input (so
as not to further load the M12/620 signal) and peak at the
risetime. Make sure the unfiltered, unattenuated, DC, 50R,
manual trigger level that you picked is ideal.

Here's a good example of just how important rise times
and trigger levels are for GPS receivers:
    http://www.leapsecond.com/pages/gpsdo-rise/
As you can see, even a fancy HP GPS receiver can be
10 ns off for every volt you set your trigger level.

3)
Just as a double check of your test setup, split the output
of *one* GPS receiver, with a 3 meter cable and see what
your SR620 TI and stddev values are. Expect this simple
test to give you about 10 ns (mean) and 25 ps (jitter). Try
it with the other GPS receiver too. If all this works you have
better confidence that your test set-up is working.

4)
Finally, have a look at Rick's papers on M12 timing at his:
    http://www.gpstime.com

/tvb


> Hello,
> 
> I may have asked this before, but still haven't got an answer.
> 
> The Motorola M12+ or M12M quote 20 ns accuracy to UTC.
> 
> I have two receivers running off the same aerial through a GPS splitter.
> 
> If I measure the 1 pps signals on my SR620 counter (single shot), they are 
> up to 100 ns out.
> 
> They are never within 20 ns.
> 
> In fact even just 1 pps signal has about 50 ns jitter on it.
> 
> I've tried this on many GPS receivers.
> 
> So how can they quote 20 ns, when two of them are up to 100 ns out.
> 
> Has anything got good results from the 1 pps signal output?
> 
> Any ideas
> 
> Martyn