Re: [time-nuts] More quartz short term stabilities
Short term stabilities of a bunch of quartz oscillators.
I had posted this some time ago but this is an updated version.
In answer to Ulrich's earlier question on the FTS 1200 #1 I retested it
in
the same setup as the FTS1200 #2 and it still showed the results
below. The Piezos are the 10811 clones. The Motorola units are
The large can units you sometimes see on eBay. The DOXO is the
clone of the Oscilloquartz dual oven 8663.
Seconds 1.0 10 100 1K
10811-60109 6.11-12 2.52-12 8.78-12
10811-60109 1.65-12 1.91-12 9.79-12
10811-60109 1.65-12 1.99-12 1.11-12
10811-60109 5.47-12 1.16-12 2.38-12
105 STYLE #T 7.60-13 2.01-12 3.94-12
10811-60111 1.27-12 2.65-12 1.93-12
105 STYLE #2 1.14-12 1.57-12 2.26-12
105 STYLE #3 1.31-12 1.58-12 2.68-12
10811-60111 3.04-12 5.54-12 1.47-12
105 STYLE #4 1.38-12 3.49-12 4.84-12
10811-60109 2.05-12 2.89-12 3.24-12
10811-60111 1.36-12 1.17-12 1.75-12
105 STYLE #5 1.14-12 2.09-12 2.51-12
FTS 1200 #1 7.14-13 1.48-12 1.92-12
10543A 2.86-11 1.05-11 1.63-11
FTS 1200 #2 2.99-13 4.30-13 7.69-13 8.65-13
10544A 4.08-12 2.95-12 7.79-12
10811-60111 1.24-12 2.10-12 2.09-12
10811-60111 1.48-12 1.59-12 1.54-12
10544A 1.89-12 2.99-12 5.01-12
105 style #6 2.88-12 2.88-12 7.37-12
105 style #7 9.69-13 2.30-12 3.49-12
10544-60511 8.20-13 2.11-12 2.51-12
Piezo clone 2.57-12 2.23-12 2.17-12
5060A osconly 4.04-12 5.94-12 8.42-12 1.30-11
10811-60111 2.59-12 3.87-12 5.77-12
10811-60109 5.08-13 1.42-12 6.10-13
Motorola 1 1.70-12 2.42-12 1.99-12
Motorola 2 2.01-12 1.71-12 1.75-12
1250B 1.22-11 3.04-12 6.54-12
FTS1000A-100 8.26-13 1.22-12 1.03-12
Piezo X 1.93-12 2.53-12 3.44-12
Piezo 1 1.71-12 3.50-12 3.00-12
Piezo 2 1.34-12 1.94-12 3.08-12
Piezo 3 1.41-11 9.50-12 2.00-12
FTS 9110 1.44-12 4.96-13 3.40-13
Wenzel 4 port 7.34-12 1.11-11 2.61-11
DOXO8663clone 2.68-12 3.52-12 3.16-12
Purify your water with professional water treatment. Click now!
http://thirdpartyoffers.juno.com/TGL2141/fc/PnY6rw2uXr2KCGbK1miow624gNyYsxN11dFkW3otSdIjJEjbgpH1X/
Short term stabilities of a bunch of quartz oscillators.
I had posted this some time ago but this is an updated version.
Corby,
Nice set of measurements.
Here are ADEV plots of your oscillators:
http://www.leapsecond.com/pages/corby/adev1.gif
http://www.leapsecond.com/pages/corby/adev2.gif
http://www.leapsecond.com/pages/corby/adev3.gif
Raw data for above:
http://www.leapsecond.com/pages/corby/adev.txt
/tvb
If one turns of the disciplining of the Thunderbolt's internal OCXO,
then one can log various parameters such as the 1 pps error.
A plot of the resulting OADEV (all tau) using such data is attached.
For short tau the GPS receiver noise dominates.
For large tau the OCXO noise dominates.
If the OADEV plot is symmetric about the minimum then the value of tau
at the Allan intercept coincides with the value of tau at the minimum in
the OADEV plot.
In this case the plot has greater slope after the minimum than before so
the value of Tau at the Allan intercept is about 500 sec rather than the
400sec corresponding to the minimum in the OADEV plot.
The optimum loop time constant for this particular Thunderbolt is thus
around 500sec.
In the absence of any other means of measuring the performance of the
Thunderbolt as a function of the loop time constant, this technique
allows the loop time constant to be set near the optimum for each
particular Thunderbolt.
Bruce
Hi Bruce,
Couple of questions. To me the plot looks like it is dominated
with measurement noise (negative slope on the left) and OCXO
frequency drift (positive slope on the right).
You mention "GPS receiver noise" but I don't see that in the
plot. Is this a free-running TBolt? If so, what 1pps source are
you comparing it to?
The title of the plot says, "Undisciplined Thunderbolt OCXO vs.
Thunderbolt GPS receiver"; so is this maybe a plot between
two different Thunderbolts; one locked, the other unlocked?
What TIC was used to collect the data?
/tvb
----- Original Message -----
From: "Bruce Griffiths" bruce.griffiths@xtra.co.nz
To: "Tom Van Baak" tvb@leapsecond.com; "Discussion of precise time and frequency measurement" time-nuts@febo.com
Sent: Tuesday, January 20, 2009 6:04 PM
Subject: Setting loop TC using thunderbolt internal data.
If one turns of the disciplining of the Thunderbolt's internal OCXO,
then one can log various parameters such as the 1 pps error.
A plot of the resulting OADEV (all tau) using such data is attached.
For short tau the GPS receiver noise dominates.
For large tau the OCXO noise dominates.
If the OADEV plot is symmetric about the minimum then the value of tau
at the Allan intercept coincides with the value of tau at the minimum in
the OADEV plot.
In this case the plot has greater slope after the minimum than before so
the value of Tau at the Allan intercept is about 500 sec rather than the
400sec corresponding to the minimum in the OADEV plot.
The optimum loop time constant for this particular Thunderbolt is thus
around 500sec.
In the absence of any other means of measuring the performance of the
Thunderbolt as a function of the loop time constant, this technique
allows the loop time constant to be set near the optimum for each
particular Thunderbolt.
Bruce
Tom
I turned of the OCXO discipling using Thunderbolt monitor menu:
Control:Disable Discipling
Then I used the Window: Logging Dialog box to enable logging of
- Time of week (sec)
- PPS offset (ns)
- 10MHz offset (ppb)
- DAC voltage
- Temperature
to a data file for about 40K sec.
The Thunderbolt keeps the DAC voltage constant and continues to measure
the oscillator offset and PPS offset even whilst the OCXO disciplining
is disabled.
The measurement noise on the LHS is that of the Thunderbolt receiver etc
itself.
The RHS is essentially OCXO drift whilst discipling is turned off.
The thunderbolt has its own internal TIC or equivalent thereto.
To increase the certainty that the receiver is behaving as I believe it
is when configured like this, it would be nice to simultaneously measure
the ADEV of the undisciplined OCXO output by other means.
Unfortunately I am not at present able to do this even though I do have
another GPSDO availble.
Bruce
Tom Van Baak wrote:
Hi Bruce,
Couple of questions. To me the plot looks like it is dominated
with measurement noise (negative slope on the left) and OCXO
frequency drift (positive slope on the right).
You mention "GPS receiver noise" but I don't see that in the
plot. Is this a free-running TBolt? If so, what 1pps source are
you comparing it to?
The title of the plot says, "Undisciplined Thunderbolt OCXO vs.
Thunderbolt GPS receiver"; so is this maybe a plot between
two different Thunderbolts; one locked, the other unlocked?
What TIC was used to collect the data?
/tvb
----- Original Message -----
From: "Bruce Griffiths" bruce.griffiths@xtra.co.nz
To: "Tom Van Baak" tvb@leapsecond.com; "Discussion of precise time and frequency measurement" time-nuts@febo.com
Sent: Tuesday, January 20, 2009 6:04 PM
Subject: Setting loop TC using thunderbolt internal data.
If one turns of the disciplining of the Thunderbolt's internal OCXO,
then one can log various parameters such as the 1 pps error.
A plot of the resulting OADEV (all tau) using such data is attached.
For short tau the GPS receiver noise dominates.
For large tau the OCXO noise dominates.
If the OADEV plot is symmetric about the minimum then the value of tau
at the Allan intercept coincides with the value of tau at the minimum in
the OADEV plot.
In this case the plot has greater slope after the minimum than before so
the value of Tau at the Allan intercept is about 500 sec rather than the
400sec corresponding to the minimum in the OADEV plot.
The optimum loop time constant for this particular Thunderbolt is thus
around 500sec.
In the absence of any other means of measuring the performance of the
Thunderbolt as a function of the loop time constant, this technique
allows the loop time constant to be set near the optimum for each
particular Thunderbolt.
Bruce
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Ah, OK. So you keep the antenna connected and you keep the
GPS receiver in position hold mode, still receiving fixes.
All you're doing is disabling the software disciplining algorithm.
That sounds like a good test. Let me try this too and see if I
agree with your conclusions.
I'll try to simultaneously measure the 1PPS or 10 MHz outputs
as you suggest.
Has anyone hacked a TBolt yet to find which internal pin has
a raw 1PPS from the GPS engine (as opposed to the 1PPS
divided down from the OCXO)?
/tvb
----- Original Message -----
From: "Bruce Griffiths" bruce.griffiths@xtra.co.nz
To: "Tom Van Baak" tvb@leapsecond.com; "Discussion of precise time and frequency measurement" time-nuts@febo.com
Sent: Tuesday, January 20, 2009 9:04 PM
Subject: Re: [time-nuts] Setting loop TC using thunderbolt internal data.
Tom
I turned of the OCXO discipling using Thunderbolt monitor menu:
Control:Disable Discipling
Then I used the Window: Logging Dialog box to enable logging of
- Time of week (sec)
- PPS offset (ns)
- 10MHz offset (ppb)
- DAC voltage
- Temperature
to a data file for about 40K sec.
The Thunderbolt keeps the DAC voltage constant and continues to measure
the oscillator offset and PPS offset even whilst the OCXO disciplining
is disabled.
The measurement noise on the LHS is that of the Thunderbolt receiver etc
itself.
The RHS is essentially OCXO drift whilst discipling is turned off.
The thunderbolt has its own internal TIC or equivalent thereto.
To increase the certainty that the receiver is behaving as I believe it
is when configured like this, it would be nice to simultaneously measure
the ADEV of the undisciplined OCXO output by other means.
Unfortunately I am not at present able to do this even though I do have
another GPSDO availble.
Bruce
Tom Van Baak wrote:
Hi Bruce,
Couple of questions. To me the plot looks like it is dominated
with measurement noise (negative slope on the left) and OCXO
frequency drift (positive slope on the right).
You mention "GPS receiver noise" but I don't see that in the
plot. Is this a free-running TBolt? If so, what 1pps source are
you comparing it to?
The title of the plot says, "Undisciplined Thunderbolt OCXO vs.
Thunderbolt GPS receiver"; so is this maybe a plot between
two different Thunderbolts; one locked, the other unlocked?
What TIC was used to collect the data?
/tvb
----- Original Message -----
From: "Bruce Griffiths" bruce.griffiths@xtra.co.nz
To: "Tom Van Baak" tvb@leapsecond.com; "Discussion of precise time and frequency measurement" time-nuts@febo.com
Sent: Tuesday, January 20, 2009 6:04 PM
Subject: Setting loop TC using thunderbolt internal data.
If one turns of the disciplining of the Thunderbolt's internal OCXO,
then one can log various parameters such as the 1 pps error.
A plot of the resulting OADEV (all tau) using such data is attached.
For short tau the GPS receiver noise dominates.
For large tau the OCXO noise dominates.
If the OADEV plot is symmetric about the minimum then the value of tau
at the Allan intercept coincides with the value of tau at the minimum in
the OADEV plot.
In this case the plot has greater slope after the minimum than before so
the value of Tau at the Allan intercept is about 500 sec rather than the
400sec corresponding to the minimum in the OADEV plot.
The optimum loop time constant for this particular Thunderbolt is thus
around 500sec.
In the absence of any other means of measuring the performance of the
Thunderbolt as a function of the loop time constant, this technique
allows the loop time constant to be set near the optimum for each
particular Thunderbolt.
Bruce
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Tom,
Tom Van Baak skrev:
Ah, OK. So you keep the antenna connected and you keep the
GPS receiver in position hold mode, still receiving fixes.
All you're doing is disabling the software disciplining algorithm.
That sounds like a good test. Let me try this too and see if I
agree with your conclusions.
We discussed this a little while back. I proposed a three-cornered hat
using two receivers and one counter... as the Thunderbolts has a builtin
TIC to GPS which can be logged. We also discussed just running a single
Thunderbolt in holdover. Others had already tried that approach. Bruce
and I discussed to some degree the effect of steering on the result.
I'll try to simultaneously measure the 1PPS or 10 MHz outputs
as you suggest.
Has anyone hacked a TBolt yet to find which internal pin has
a raw 1PPS from the GPS engine (as opposed to the 1PPS
divided down from the OCXO)?
Hmm... I have not even "repaired" (in the sense, that they are broken
until opened and "repaired", i.e. just looking under the hood for sake
of curiosity). any of mine...
Cheers,
Magnus
Magnus Danielson wrote:
Tom,
Tom Van Baak skrev:
Ah, OK. So you keep the antenna connected and you keep the
GPS receiver in position hold mode, still receiving fixes.
All you're doing is disabling the software disciplining algorithm.
That sounds like a good test. Let me try this too and see if I
agree with your conclusions.
We discussed this a little while back. I proposed a three-cornered hat
using two receivers and one counter... as the Thunderbolts has a builtin
TIC to GPS which can be logged. We also discussed just running a single
Thunderbolt in holdover. Others had already tried that approach. Bruce
and I discussed to some degree the effect of steering on the result.
I'll try to simultaneously measure the 1PPS or 10 MHz outputs
as you suggest.
Has anyone hacked a TBolt yet to find which internal pin has
a raw 1PPS from the GPS engine (as opposed to the 1PPS
divided down from the OCXO)?
Hmm... I have not even "repaired" (in the sense, that they are broken
until opened and "repaired", i.e. just looking under the hood for sake
of curiosity). any of mine...
Cheers,
Magnus
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Magnus
There may not even be a real PPS signal produced by the GPS engine which
is compared with the PPS generated by dividing down the OCXO output.
With the GPS engine LO derived from the OCXO, having a real PPS signal
generated by the GPS engine for such a comparison isnt necessary in
order to measure the phase error of the PPS signal generated from the
OCXO output.
Bruce
Hej Magnus
Magnus Danielson wrote:
Tom,
Tom Van Baak skrev:
Ah, OK. So you keep the antenna connected and you keep the
GPS receiver in position hold mode, still receiving fixes.
All you're doing is disabling the software disciplining algorithm.
That sounds like a good test. Let me try this too and see if I
agree with your conclusions.
We discussed this a little while back. I proposed a three-cornered hat
using two receivers and one counter... as the Thunderbolts has a builtin
TIC to GPS which can be logged. We also discussed just running a single
Thunderbolt in holdover. Others had already tried that approach. Bruce
and I discussed to some degree the effect of steering on the result.
The 3 cornered hat technique only works well (even in the extended form
allowing finite correlations between the 3 sources) when the ADEV of the
sources being compared aren't too disparate.
Consequently this comparison will only work well for tau values in the
vicinity of the Allan intercepts which in turn will need to be not too
disparate.
I'll try to simultaneously measure the 1PPS or 10 MHz outputs
as you suggest.
Has anyone hacked a TBolt yet to find which internal pin has
a raw 1PPS from the GPS engine (as opposed to the 1PPS
divided down from the OCXO)?
Hmm... I have not even "repaired" (in the sense, that they are broken
until opened and "repaired", i.e. just looking under the hood for sake
of curiosity). any of mine...
Cheers,
Magnus
Bruce
Bruce Griffiths skrev:
Magnus Danielson wrote:
Tom,
Tom Van Baak skrev:
Ah, OK. So you keep the antenna connected and you keep the
GPS receiver in position hold mode, still receiving fixes.
All you're doing is disabling the software disciplining algorithm.
That sounds like a good test. Let me try this too and see if I
agree with your conclusions.
We discussed this a little while back. I proposed a three-cornered hat
using two receivers and one counter... as the Thunderbolts has a builtin
TIC to GPS which can be logged. We also discussed just running a single
Thunderbolt in holdover. Others had already tried that approach. Bruce
and I discussed to some degree the effect of steering on the result.
I'll try to simultaneously measure the 1PPS or 10 MHz outputs
as you suggest.
Has anyone hacked a TBolt yet to find which internal pin has
a raw 1PPS from the GPS engine (as opposed to the 1PPS
divided down from the OCXO)?
Hmm... I have not even "repaired" (in the sense, that they are broken
until opened and "repaired", i.e. just looking under the hood for sake
of curiosity). any of mine...
Cheers,
Magnus
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Magnus
There may not even be a real PPS signal produced by the GPS engine which
is compared with the PPS generated by dividing down the OCXO output.
With the GPS engine LO derived from the OCXO, having a real PPS signal
generated by the GPS engine for such a comparison isnt necessary in
order to measure the phase error of the PPS signal generated from the
OCXO output.
Actually, the PPS out may very well be the PPS output of the engine...
But yes, I agree. There is no point in realizing the PPS twice when you
dicipline the OCXO anyways, and the time-difference is never measured by
a TIC, the GPS solution gives the time-error and the PPS only need
re-assignment when it is too big.
Cheers,
Magnus