Re: [time-nuts] Accuracy results with Trimble Thunderbolt?
Tom, Thanks very much, that's a great list and will keep me busy for some time. At the hobbyist level, are there time interval counters or frequency counters that are particularly popular, and/or that pair well with the listed software? Frank On 3/25/20 2:30 PM, Tom Van Baak wrote: Frank, The software that most of us use to create Allan deviation plots is free and not that hard to use. You will need ascii numerical phase or frequency data, e.g., from a time interval or frequency counter. 1) Stable32, by Bill Riley See: https://ieee-uffc.org/frequency-control/frequency-control-software/stable32/ Documentation and a goldmine of T information: http://www.wriley.com/ 2) TimeLab, by John Miles http://www.ke5fx.com/timelab/readme.htm Documentation and tutorials: http://www.miles.io/PhaseStation_53100A_user_manual.pdf 3) Allan deviation for Python, by Anders Wallin https://pypi.org/project/AllanTools/ 4) Command line tools for ADEV adev_lib.c, adev4.c, adev5.c in my http://leapsecond.com/tools/ directory. 5) Excel, gnuplot, etc. Any software that generates log-log plots can be used to make ADEV plots. You first calculate the statistics using CLI tools and then use the GUI to make the plots. This gives maximum flexibility in plotting clarity and style but requires more work than canned packages like Stable32 or TimeLab. Normally I recommend TimeLab to new users, but I'm run into VBA and Excel wizards who are proficient in that environment. 6) Plotter, by Ulrich Bangert Still used by some time nuts, maybe "not recommended for new design", since Ulrich is no longer with us. /tvb On 3/25/2020 1:44 PM, Frank wrote: Big thanks to Attila, Frank, and Taka for your responses to my questions on my Trimble Thunderbolt. I'm inclined to follow Taka's advice and reset the antenna elevation mask angle with LH's FE command to something between 10 and 20 degrees. I'm guessing that this is entered in degrees (i.e. "15"), but I can't find anything in the LH documentation to verify this, so if it's wrong I'm happy to be corrected. Also, as noted I set the satellite signal level mask to 1 with LH's FL command, but short of doing a full reset of the Thunderbolt I'm having trouble figuring out what the default value is for this. I appreciate Attila's and Frank's suggestions on topics to study such as Allan deviation to dip more toes into metrology. I've started in on this, though I imagine it will take a while to work through. For now, I wanted to throw out one follow-up question. I see that ADEV diagrams plotting Allan deviation against time seem to be a primary tool for evaluating GPSDO performance. If I wanted to compare the Thunderbolt to another GPSDO (for example, I also have a Bodnar unit), is there software not wildly beyond a hobbyist budget that would allow me to compile and display similar data? Thanks again, Frank ___ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
Re: [time-nuts] Accuracy results with Trimble Thunderbolt?
Frank, The software that most of us use to create Allan deviation plots is free and not that hard to use. You will need ascii numerical phase or frequency data, e.g., from a time interval or frequency counter. 1) Stable32, by Bill Riley See: https://ieee-uffc.org/frequency-control/frequency-control-software/stable32/ Documentation and a goldmine of T information: http://www.wriley.com/ 2) TimeLab, by John Miles http://www.ke5fx.com/timelab/readme.htm Documentation and tutorials: http://www.miles.io/PhaseStation_53100A_user_manual.pdf 3) Allan deviation for Python, by Anders Wallin https://pypi.org/project/AllanTools/ 4) Command line tools for ADEV adev_lib.c, adev4.c, adev5.c in my http://leapsecond.com/tools/ directory. 5) Excel, gnuplot, etc. Any software that generates log-log plots can be used to make ADEV plots. You first calculate the statistics using CLI tools and then use the GUI to make the plots. This gives maximum flexibility in plotting clarity and style but requires more work than canned packages like Stable32 or TimeLab. Normally I recommend TimeLab to new users, but I'm run into VBA and Excel wizards who are proficient in that environment. 6) Plotter, by Ulrich Bangert Still used by some time nuts, maybe "not recommended for new design", since Ulrich is no longer with us. /tvb On 3/25/2020 1:44 PM, Frank wrote: Big thanks to Attila, Frank, and Taka for your responses to my questions on my Trimble Thunderbolt. I'm inclined to follow Taka's advice and reset the antenna elevation mask angle with LH's FE command to something between 10 and 20 degrees. I'm guessing that this is entered in degrees (i.e. "15"), but I can't find anything in the LH documentation to verify this, so if it's wrong I'm happy to be corrected. Also, as noted I set the satellite signal level mask to 1 with LH's FL command, but short of doing a full reset of the Thunderbolt I'm having trouble figuring out what the default value is for this. I appreciate Attila's and Frank's suggestions on topics to study such as Allan deviation to dip more toes into metrology. I've started in on this, though I imagine it will take a while to work through. For now, I wanted to throw out one follow-up question. I see that ADEV diagrams plotting Allan deviation against time seem to be a primary tool for evaluating GPSDO performance. If I wanted to compare the Thunderbolt to another GPSDO (for example, I also have a Bodnar unit), is there software not wildly beyond a hobbyist budget that would allow me to compile and display similar data? Thanks again, Frank ___ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there. ___ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
Re: [time-nuts] Accuracy results with Trimble Thunderbolt?
Big thanks to Attila, Frank, and Taka for your responses to my questions on my Trimble Thunderbolt. I'm inclined to follow Taka's advice and reset the antenna elevation mask angle with LH's FE command to something between 10 and 20 degrees. I'm guessing that this is entered in degrees (i.e. "15"), but I can't find anything in the LH documentation to verify this, so if it's wrong I'm happy to be corrected. Also, as noted I set the satellite signal level mask to 1 with LH's FL command, but short of doing a full reset of the Thunderbolt I'm having trouble figuring out what the default value is for this. I appreciate Attila's and Frank's suggestions on topics to study such as Allan deviation to dip more toes into metrology. I've started in on this, though I imagine it will take a while to work through. For now, I wanted to throw out one follow-up question. I see that ADEV diagrams plotting Allan deviation against time seem to be a primary tool for evaluating GPSDO performance. If I wanted to compare the Thunderbolt to another GPSDO (for example, I also have a Bodnar unit), is there software not wildly beyond a hobbyist budget that would allow me to compile and display similar data? Thanks again, Frank ___ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
Re: [time-nuts] Accuracy results with Trimble Thunderbolt?
On Sat, 21 Mar 2020 14:24:48 +0100 Frank Stellmach wrote: > you're welcome to pose any questions here. Definitely! We all started as beginners and learned through asking questions. > Concerning your question, there are basically two different parameters > called 'stability' and 'uncertainty'. > I personally don't use the term 'accuracy' any more, because that's > somewhat misleading. I recommend here reading [1] and [2]. The terms used in metrology are confusing at first and one needs time to digest them. A quite a few things that we are used to do in "normal" life do not work for metrology anymore. Mostly because we assume that our standard (e.g. the calipers, the gauges, weights, etc) do not change. Once you enter metrology, your standard isn't stable anymore and you need to evaluate not only how accurate it is, but how stable. This all then ends up in an uncertainty of your measurement, parametrized by a specific set of conditions (temperature, how long you are measuring, etc). John Vig's Tutorial is also a good source of information to get started: http://www.ko4bb.com/getsimple/index.php?id=manuals=02_GPS_Timing/John_Vig_Tutorials_on_Crystal_Oscillators > Btw.: The GPS system delivers an in-official uncertainty, because the > D.O.D. clock is not participating in the S.I. representation of the UTC. This is not true. While BIPM only allows a single NMI per country to contribute to TAI/EAL these days, this wasn't case in the past. And for historic reasons there are a few countries where two entities contribute to TAI/EAL. The USNO, master over GPS time, is one of those non-NMI entities contributing. They also used to be in the past the one single organization that had the most atomic clocks running, though that's slowly changing now. They still are one of of the organisations that have the most stable clock ensambles contributing to EAL, though, and will stay so for the forseeable future. As for how far they are off, have a look at their circular T entry: https://webtai.bipm.org/database/canvas.html?utclab=ok=usno=57078=58919 And compare it to, e.g. PTB: https://webtai.bipm.org/database/canvas.html?utclab=ok=ptb=57078=58919 Attila Kinali [1] "Characterization of Clocks and Oscillators" NIST Technical Note 1337, by Sullivan, Allan, Howe, Walls, 1990 http://tf.nist.gov/general/pdf/868.pdf [2] "Handbook of Frequency Stability Analysis" NIST Special Publication 1065, by Riley, 2008 http://tf.nist.gov/timefreq/general/pdf/2220.pdf -- Science is made up of so many things that appear obvious after they are explained. -- Pardot Kynes ___ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
[time-nuts] Accuracy results with Trimble Thunderbolt?
Hello Frank,
you're welcome to pose any questions here.
Concerning your question, there are basically two different parameters
called 'stability' and 'uncertainty'.
I personally don't use the term 'accuracy' any more, because that's
somewhat misleading.
The Stability is a characteristic of oscillators, let it be OCXOs, a
naked GPS receiver, or a combined system of a GPS receiver which
disciplines an OCXO. This stability statistics (Allan Deviation, or
ADEV, e.g. inside LH) describes, how much fluctuation / jitter you
encounter on different time scales (also of your counter) when you use
your TB as a time base, on different Gate Times of your frequency
measurements.
You might study such ADEV diagrams for different GPSDO, OCXO, Rb-, Cs-
and MASER clocks on several time-nuts pages to get a better idea:
http://www.ke5fx.com/gpscomp.htm
For the Trimble TB, these fluctuations are on the order of 10^-10 ..
10^-11 for a short Gate Time of maybe 10msec.. 1sec, and prevent that
you get frequency measurements more precise ('accurate') than that.
If you use averaging, or a longer Gate Time, then these fluctuations go
further down due to the good short term stability of the OCXO inside the
TB, then increase at around the time constant you've chosen (500sec?)
due to the big jitter of the GPS signals, and then go down again, into
the 10^-12.. 10^-13 region at averaging times of hours or days, because
the GPS satellite system is synchronized to a Cs master clock at the
D.O.D. Therefore you can achieve an uncertainty (~ 'accuracy') of about
10^-13 also, but that depends also on the oscillator / clock you have in
your house.
Btw.: The GPS system delivers an in-official uncertainty, because the
D.O.D. clock is not participating in the S.I. representation of the UTC.
This LH parameter 'OSC' will give you an estimate, how close the
internal OCXO is currently synchronized to GPS time, but you always have
to take into account the typical ADEV jitter for your specific
measurement setup.
Frank
___
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
Re: [time-nuts] Accuracy results with Trimble Thunderbolt?
Antenna mask of zero degrees is actually NOT good. You get satellites going in and out of your antenna's view too often (because satellite can pop in and out of horizon, and other obstructions), cause switches in satellite used, and that contribute to fluctuations. 10 to 20 degree, in most cases, will give you better results. I'd suggest seeing what it is set to and adjust accordingly. "Auto" anything will assume a lot of things, and what it arrives isn't always the best. Also, what T-bolt outputs and LH displays isn't an actual measurement of the output. It's a computed figure. I don't actually look at it... as long as it's locked, I'm pretty happy. By the way... I have a lot of GPSDO in my lab. Short term fluctuation of output vary quiet a bit. That is quite normal. It is designed to give you a good average over time. (hours to days) --- (Mr.) Taka Kamiya KB4EMF / ex JF2DKG On Friday, March 20, 2020, 5:51:27 PM EDT, Frank O'Donnell wrote: I hope you all don't mind a novice-level question, but I'd be interested in any feedback on what I'm seeing with the accuracy of the 10 MHz signal out on my Trimble Thunderbolt. I bought the Thunderbolt last year used on eBay, and it appears to be of about 2001 vintage. Most recently it's been running continuously for about 2 1/2 months, attached to a roof-mounted Lucent PCTEL 26db twist antenna with a clear view of the sky, with the receiver located in a room with a relatively stable temperature. Lady Heather and Thunderbolt Monitor both appear to report it to be well settled-in. My main use of the Thunderbolt is to supply a 10 MHz reference signal to an HP3586B and HP3336B for frequency measurement purposes. At the beginning of this month I remember typically seeing about 20 ppt accuracy for the 10 MHz reference as reported in Heather. I then used Heather commands to initiate an autoset of oscillator parameters (set antenna elevation mask angle to 0 with FE keyboard command and set signal level mask to 1 with FL command in order to allow collection of signal level data across full sky; clear signal level history with CM keyboard command; let run at least 6-12 hours to build up new satellite signal level map; issue “” autotune command, allow time to complete). Now I seem to be seeing about 60 ppt accuracy on average. For my purposes, these differences are probably academic -- if I'm doing my math right, 60 ppt in a 10 MHz signal is 0.0006 Hz, far down in the noise of frequency variation due to Doppler etc. Still, I guess I've drunk enough of the Koolaid to wonder about the accuracy level and any ways to improve it. So, some questions. Is it possible that the increase in the error seen could be due to the oscillator parameter autoset sequence that I ran? If so, is there a way to remedy this? What would be typical accuracy for the 10 MHz reference on a Thunderbolt? Is there anything else I can do with the Thunderbolt itself to increase the accuracy? If I want to consider an alternative to the Thunderbolt that might offer better accuracy, is there a logical next step? Thanks for any suggestions, Frank ___ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there. ___ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
[time-nuts] Accuracy results with Trimble Thunderbolt?
I hope you all don't mind a novice-level question, but I'd be interested in any feedback on what I'm seeing with the accuracy of the 10 MHz signal out on my Trimble Thunderbolt. I bought the Thunderbolt last year used on eBay, and it appears to be of about 2001 vintage. Most recently it's been running continuously for about 2 1/2 months, attached to a roof-mounted Lucent PCTEL 26db twist antenna with a clear view of the sky, with the receiver located in a room with a relatively stable temperature. Lady Heather and Thunderbolt Monitor both appear to report it to be well settled-in. My main use of the Thunderbolt is to supply a 10 MHz reference signal to an HP3586B and HP3336B for frequency measurement purposes. At the beginning of this month I remember typically seeing about 20 ppt accuracy for the 10 MHz reference as reported in Heather. I then used Heather commands to initiate an autoset of oscillator parameters (set antenna elevation mask angle to 0 with FE keyboard command and set signal level mask to 1 with FL command in order to allow collection of signal level data across full sky; clear signal level history with CM keyboard command; let run at least 6-12 hours to build up new satellite signal level map; issue “” autotune command, allow time to complete). Now I seem to be seeing about 60 ppt accuracy on average. For my purposes, these differences are probably academic -- if I'm doing my math right, 60 ppt in a 10 MHz signal is 0.0006 Hz, far down in the noise of frequency variation due to Doppler etc. Still, I guess I've drunk enough of the Koolaid to wonder about the accuracy level and any ways to improve it. So, some questions. Is it possible that the increase in the error seen could be due to the oscillator parameter autoset sequence that I ran? If so, is there a way to remedy this? What would be typical accuracy for the 10 MHz reference on a Thunderbolt? Is there anything else I can do with the Thunderbolt itself to increase the accuracy? If I want to consider an alternative to the Thunderbolt that might offer better accuracy, is there a logical next step? Thanks for any suggestions, Frank ___ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
