Discussion:
LEA-6T TCXO measurements
(too old to reply)
Gabs Ricalde
2018-04-26 00:56:07 UTC
Permalink
Raw Message
The NAV-TIMEUTC message of the u-blox LEA-6T has a fractional seconds
field, which is the receiver clock offset estimate and can be used to
measure the internal TCXO. Attached are the ADEV and frequency
difference plots (blue traces). For comparison, I also have PPS logs
from a TL-WR703N NTP server (pink traces) with ntpd stopped.

I'm wondering why the LEA-6T has that kind of frequency jumps.
Bob kb8tq
2018-04-26 12:36:42 UTC
Permalink
Raw Message
Hi

What you are looking at appear to be sawtooth jumps. Simply put, the module is
looking at the closest edge on the TCXO to do it’s timing. When the device drifts,
it can “slip” to another cycle. If you watch the PPS out, there are also artifacts that
result from this process operating in a non-ideal fashion. Lots of details on that in
the archives.

Bob
Post by Gabs Ricalde
The NAV-TIMEUTC message of the u-blox LEA-6T has a fractional seconds
field, which is the receiver clock offset estimate and can be used to
measure the internal TCXO. Attached are the ADEV and frequency
difference plots (blue traces). For comparison, I also have PPS logs
from a TL-WR703N NTP server (pink traces) with ntpd stopped.
I'm wondering why the LEA-6T has that kind of frequency jumps.
<ADEV.png><freq.png>_______________________________________________
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Gabs Ricalde
2018-05-04 09:18:55 UTC
Permalink
Raw Message
Post by Bob kb8tq
Hi
What you are looking at appear to be sawtooth jumps. Simply put, the module is
looking at the closest edge on the TCXO to do it’s timing. When the device drifts,
it can “slip” to another cycle. If you watch the PPS out, there are also artifacts that
result from this process operating in a non-ideal fashion. Lots of details on that in
the archives.
Hi Bob,

The raw data has a sawtooth pattern due to millisecond clock jumps
[1], but TimeLab unwraps it and they happen more frequently than the
frequency jumps.
I redid the the experiment using the same receiver (pink trace) and
another LEA-6T (green trace), there were no frequency jumps. SBAS was
enabled in the old measurements (blue trace). I'm not sure if that
explains it, but the manual does not recommend SBAS for timing.


[1] "GNSS Receiver Clocks" http://www.insidegnss.com/node/2512
Bob kb8tq
2018-05-04 12:56:16 UTC
Permalink
Raw Message
Hi

Two different meanings of the term sawtooth.

The first plot shows a very normal phase rollover process. That (when incorrectly
unwound) can create errors. In some cases those sorts of errors are inevitable and
it is always wise to see if data “noise” occurs right at a wrap around point.

The other meaning of sawtooth in the case of GPS is rooted in the same process.
The GPS time solution and local time have a pattern very much like the first plot.
As long as it is running as shown in the first plot, it averages out (provided there is
averaging). The problem comes in when the plot does *not* look like the first graph.
If the TCXO comes to a flat spot on it’s compensation curve, it can “hang out” to one
side of the error band. Instead of a sawtooth you get a curve. That process of staying
to one side of the error band does not average out and you get a bump in the time
record.

If you dig into the archives, there are lots of plots showing typical data.

Bob
Post by Gabs Ricalde
Post by Bob kb8tq
Hi
What you are looking at appear to be sawtooth jumps. Simply put, the module is
looking at the closest edge on the TCXO to do it’s timing. When the device drifts,
it can “slip” to another cycle. If you watch the PPS out, there are also artifacts that
result from this process operating in a non-ideal fashion. Lots of details on that in
the archives.
Hi Bob,
The raw data has a sawtooth pattern due to millisecond clock jumps
[1], but TimeLab unwraps it and they happen more frequently than the
frequency jumps.
I redid the the experiment using the same receiver (pink trace) and
another LEA-6T (green trace), there were no frequency jumps. SBAS was
enabled in the old measurements (blue trace). I'm not sure if that
explains it, but the manual does not recommend SBAS for timing.
[1] "GNSS Receiver Clocks" http://www.insidegnss.com/node/2512
<phase diff.png><freq2.png>_______________________________________________
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Loading...