[time-nuts] Best GPS 1PPS Accuracy
Randy at synergy-gps.com
Wed Dec 13 13:27:28 EST 2006
Jason is right. I think the M12M is currently at the top of the heap.
Rick Hambly and Tom Clark have developed some circuitry that knocks the
jitter down quite a lot.
Basically, for a "normal" timing receiver (please don't flame me guys, I
know there is other stuff out there) the jitter is dependent on the
receiver micro-controller's clock frequency. For instance, the old UT+
timing receiver had a jitter spec of +/- 45ns. If you were to compare
the short term 1PPS output from a UT+ to a rubidium, OCXO, or any
"stable" reference oscillator with a 1PPS output you would see that one
pulse might show up about 45ns ahead of the reference 1PPS output, the
next one would be about 45ns behind the reference, etc., etc......
The reason for this particular "jitter" of 90ns is because the UT+ used
a crystal running at about 11MHz, giving a period of about 90ns per
clock cycle. Since the receiver can only place 1PPS pulses with a 90ns
granularity it places the pulse as close to the what it thinks is the
actual UTC time tick. One time it will be on one side of the UTC tick,
the next time it may occur after the tick. This is where the term
"sawtooth" comes from. In one of the timing messages the receiver puts
out its best guess of how far off the NEXT pulse is going to be. If you
plot this data it looks roughly like a sawtooth as on average 50% of the
pulses are ahead of the time tick and the other 50% are behind it. I
have attached a screenshot from Winoncore12 showing this waveform.
Note that in this screenshot the jitter is about +/-15ns. That's because
the receiver I used for this plot is an M12+ timer. The reason for the
increased resolution is that the M12+T uses a 16.384MHz clock, and that
it can place the pulse on both the rising and falling edges of the
clock. This means that the M12+T has a 30ns granularity, resulting in a
+/-15ns amplitude of the sawtooth (jitter).
Note that these are all "perfect world" numbers, which is why they are
expressed as 1 Sigma values. Actual realizable performance ultimately
depends on how good the underlying code running the GPS receiver is.
This is why we all ultimately rely on the experts in the timing
community (many of whom are on this list) to test and publish "real
world" numbers, instead of some marketeer's spin on the performance of
the receiver he happens to hawking at the time.....
Sorry for the long post....
Senior Applications Engineer
Synergy Systems, LLC
randy at synergy-gps.com
From: time-nuts-bounces at febo.com [mailto:time-nuts-bounces at febo.com] On
Behalf Of Jason Rabel
Sent: Wednesday, December 13, 2006 9:39 AM
To: 'Discussion of precise time and frequency measurement'
Subject: Re: [time-nuts] Best GPS 1PPS Accuracy
One problem is that older GPS receiver spec sheets give numbers with SA
on, even though SA has been off for quite some time, so their numbers
will be inherently higher.
Probably the most accurate GPS receiver today would be the Motorola M12M
Timing. I would go into more detail but I'm about to head out the door.
I'm sure others will chime in on this soon.
There were some posts a few days ago about some new papers that were
published, there is some good info on the various recent versions of the
Motorola receivers... Might want to check that out.
> Can someone explain the accuracy numbers that are represented in specs
> I find that Trimble says the Resolution T
> (http://www.trimble.com/resolutiont.shtml) has 15 ns (1 Sigma) like
> So I guess the real question is; are you comparing apples to apples
> standard deviation (Sigma) is specified?
> Another question is; what's the most accurate GPS receiver module
> realize this is only as accurate at the antenna system, but I'll save
> a multipath discussion for later.
time-nuts mailing list
time-nuts at febo.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: M12+T sawtooth example.gif
Size: 8928 bytes
Desc: M12+T sawtooth example.gif
Url : http://www.febo.com/pipermail/time-nuts/attachments/20061213/44095935/attachment-0001.gif
More information about the time-nuts