Re: [time-nuts] Phase noise from Allan Deviation ?
An initial gnuradio GRC flowgraph for measuring Time Intervals, and dumping raw binary to a file is now on github. There are two helper Python files: one to convert from binary to floating-point-ASCII-text, and another to take that from raw measurements to Time-Intervals. That file can be sent straight into TVB's ADEV.C program. The TEXT-to-TimeInterval python program pretty aggressively defines outliers and throws them out. This is due to a bug in the particular SDR being used which seems to insert random large glitches into the 1-Hertz output. The flowgraph and python code are quite simple, undoubtly one would want to tweak it. I've measured an external disciplined signal input to the SDR running on it's internal crystal oscillator, and another measurement of a GPSDO that is both the external clock reference for the SDR and simultaneously the unknown. The purpose is this later is to characgterize what the background ADEV looks like. The GPSDO compared to itself is of course a lot better and pretty straight as 1/tau for an hour or so becausue it's just measuring anything bad the SDR does to it. https://github.com/Tom-McDermott/SDR-time-interval -- Tom, N5EG On Mon, Dec 14, 2015 at 10:01 PM, Tom McDermottwrote: > Hi Iain, > > I'll publish a flowgraph soon. I built a Out-Of-Tree module to decimate > the output > down to one sample per reading to keep the output file small for even long > runs. > On the order of about 48000:! decimation. Strange thing is - it works > when the > QT Time Sink is enabled, but gives very wrong outputs when the QT Time Sink > is Disabled (the only change). > > So I suspect my custom OOT is doing something wrong, but not sure why > enabling/disabling the scope display changes the gnuradio behavior so much. > I'd like to get more to the bottom of this before publishing the code. > > Have done some short runs throwing 48,000 samples/sec to the output file, > then post > processing that in Python, and it gives the same results as my OOT module > when the > QT GUI is showing. Very strange, but the data file is much too verbose: > 700 Megabytes/hour. > > -- Tom, N5EG > > > > On Mon, Dec 14, 2015 at 1:15 AM, Iain Young wrote: > >> Hi Tom, >> >> On 14/12/15 03:15, You wrote: >> >> I've constructed a homebrew setup to measure time intervals using a >>> software defined radio. Basically a single-channel downconversion to >>> about one hertz, then count samples from the SDR clock to time stamp >>> the zero crossings. This is done in gnuradio and saved to a file for >>> post processing. The resolution is theoretically good, but the accuracy >>> is unknown. >>> >> >> Very interesting. Would you consider making your flowgraph available ? >> >> I have done similar things with just thresholding and looking for the >> start of second (or minute) marker of various distant radio clocks, and >> then graphing how far apart they were, as well as feeding NTP. >> >> >> 73s >> >> Iain >> >> >> ___ >> 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. >> > > ___ 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.
Re: [time-nuts] Phase noise from Allan Deviation ?
Hi Iain, I'll publish a flowgraph soon. I built a Out-Of-Tree module to decimate the output down to one sample per reading to keep the output file small for even long runs. On the order of about 48000:! decimation. Strange thing is - it works when the QT Time Sink is enabled, but gives very wrong outputs when the QT Time Sink is Disabled (the only change). So I suspect my custom OOT is doing something wrong, but not sure why enabling/disabling the scope display changes the gnuradio behavior so much. I'd like to get more to the bottom of this before publishing the code. Have done some short runs throwing 48,000 samples/sec to the output file, then post processing that in Python, and it gives the same results as my OOT module when the QT GUI is showing. Very strange, but the data file is much too verbose: 700 Megabytes/hour. -- Tom, N5EG On Mon, Dec 14, 2015 at 1:15 AM, Iain Youngwrote: > Hi Tom, > > On 14/12/15 03:15, You wrote: > > I've constructed a homebrew setup to measure time intervals using a >> software defined radio. Basically a single-channel downconversion to >> about one hertz, then count samples from the SDR clock to time stamp >> the zero crossings. This is done in gnuradio and saved to a file for >> post processing. The resolution is theoretically good, but the accuracy >> is unknown. >> > > Very interesting. Would you consider making your flowgraph available ? > > I have done similar things with just thresholding and looking for the > start of second (or minute) marker of various distant radio clocks, and > then graphing how far apart they were, as well as feeding NTP. > > > 73s > > Iain > > > ___ > 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. > ___ 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.
Re: [time-nuts] Phase noise from Allan Deviation ?
Hi ADEV to phase noise can be full of gotcha’s. What you have out of your system is a series of phase samples. You don’t *have* to convert those samples to ADEV. There are a whole bunch different variances you can calculate. The best thing to do (by far) is to save the phase samples. That way you can post process them with whatever variance you happen to be interested in. As an example, you may find that modified ADEV and Hadamard help you sort out various noise processes. There also is the option of doing an FFT on the phase samples. Bob > On Dec 13, 2015, at 10:15 PM, Tom McDermottwrote: > > I've constructed a homebrew setup to measure time intervals using a > software defined radio. Basically a single-channel downconversion to > about one hertz, then count samples from the SDR clock to time stamp > the zero crossings. This is done in gnuradio and saved to a file for > post processing. The resolution is theoretically good, but the accuracy > is unknown. > > The result produces ADEV vs. Tau charts with reasonably sane looking > results. > The 'unknown' is a Rubidium oscillator locked to CDMA pilot (TS2700), > and the internal crystal oscillator in the SDR radio as the reference. > Thus, ADEV probably is mostly measuring that internal crystal rather than > the TS2700. Later on a GPSDO will be tried as the reference clock to > see if the Adev results are better. > > It brings up a question: Is it possible to estimate the phase noise of that > internal crystal from the ADEV measurements? There are a bunch of > papers that go the other way: from Phase Noise to Adev. Searching > brings up only one paper that goes from ADEV to Phase Noise but it's text > does not seem to be readily available. It apparently models the oscillator > as a couple of well known error models. > > -- Tom, N5EG > ___ > 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. ___ 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.
Re: [time-nuts] Phase noise from Allan Deviation ?
Yes. I was not overly pleased with the performance of my 2700's. I ended up pulling the PRS10 out of one of them and purchased the breakout connector board from SRS and used the resulting 10 MHz and 1 PPS outputs.From time to time I would sync the unit to one of my GPSDO's using the 1 PPS input. I haven't powered up my 2700's in several years. Your mileage may vary. > On Dec 14, 2015, at 1:10 PM, Bob Campwrote: > > Hi > > One of the reasons the TS2700’s went out of favor is the “quality” of the > CDMA signals > available. The design assumption was that the CDMA carriers provided timing > as good > as GPS on their over the air systems. After the units had been in the field > for a while it > became apparent that the 2700’s were not performing up to expectations. > Further investigation > turned up a range of issues that degraded the CDMA timing relative to GPS. A > lot of it > boiled down to “we are a phone service not a time service”. System wise, > CDMA gets > into trouble at the 10us level. GPS is in trouble at the 100 ns level…. > > Yes there are all sorts of rules and regulations. In the end it’s “this works > fine” that trumps > a lot of them. > > Bob > >> On Dec 14, 2015, at 1:15 PM, Charles Steinmetz wrote: >> >> Tom wrote: >> >>> The 'unknown' is a Rubidium oscillator locked to CDMA pilot (TS2700) >> >> There is very little information publicly available on the Symmetricom >> "BesTime" engine ("BTE"), but after playing with a few TS2700s for quite >> some time, including monitoring a number of internal signals, several things >> became apparent. First, the 2700 does not seem to discipline the PRS10. >> The rubidium runs open loop and the BTE keeps track of the offset and the >> drift rate from "BTE time" (which is synthesized from all available sources >> -- however many CDMA signals it is receiving, plus any wireline telco timing >> signals and the PRS10 -- using a proprietary algorithm to estimate the >> reliability of each source and outputting BTE time and frequency using DDS). >> Hobby users won't be feeding the unit any telco timing signals, so the BTE >> has only the CDMA signals to work from. During holdover (and assuming no >> telco timing signals), the Rb is the sole input to the BTE, which uses the >> stored offset and drift to calculate BTE time. >> >> I found that the TS-2700 is more than an order of magnitude less stable than >> a Trimble Thunderbolt, even with a full complement of rock-solid CDMA >> sources. This may vary somewhat, depending on the CDMA equipment in use at >> any particular location and the diligence of the CDMA operator. >> >> Best regards, >> >> Charles >> >> >> ___ >> 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. > > ___ > 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. > ___ 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.
Re: [time-nuts] Phase noise from Allan Deviation ?
Rick, On 12/14/2015 07:04 AM, Richard (Rick) Karlquist wrote: On 12/13/2015 7:15 PM, Tom McDermott wrote: It brings up a question: Is it possible to estimate the phase noise of that internal crystal from the ADEV measurements? There are a bunch of papers that go the other way: from Phase Noise to Adev. Searching brings up only one paper that goes from ADEV to Phase Noise but it's text does not seem to be readily available. It apparently models the oscillator as a couple of well known error models. -- Tom, N5EG] In general, you cannot determine phase noise from ADEV, even though you can determine ADEV from phase noise. This is just a mathematical reality. Mike Fischer (of HP) presented papers at the 1977 and 1978 that show conversions between PN and ADEV for individual noise processes, where each process has a specific slope of amplitude vs frequency. The only time you can go from ADEV to PN is if you can isolate a process. 1976 PTTI paper by Fischer: http://tycho.usno.navy.mil/ptti/1976papers/Vol%2008_38.pdf 1977 PTTI paper by Chi (Fischer being adviser): http://tycho.usno.navy.mil/ptti/1977papers/Vol%2009_34.pdf 1978 PTTI paper by Fischer: http://tycho.usno.navy.mil/ptti/1978papers/Vol%2010_14.pdf Things have happen since those papers, in many ways. In the specific case of crystal oscillators, in general, they follow a flicker noise of frequency process model close to the carrier (within 100 Hz). You can often assume that ADEV is dominated by this process and therefore translate it to an equivalent PN. The way to tell if ADEV is dominated by this process is that it will be independent of tau, for tau of 0.1 sec or less. On the IEEE UFFC site, there is a tutorial on crystal oscillator design by Mike Driscoll (you don't have to be a member to access it). I believe this covers this topic. You go from ADEV to "frequency noise" and then to phase noise http://www.ieee-uffc.org/frequency-control/learning/2003_IEEE_Tutorial.PDF Other tutorials is at: http://www.ieee-uffc.org/frequency-control/tutorials.asp In particular, check out Francois Vernottes presentation: http://www.ieee-uffc.org/frequency-control/learning/pdf/Vernotte-Varience_Measurements.pdf He covers a lot of ground there, and in the end he also covers curve-fitting to estimate noise levels. You can "practice" this calculation on any crystal oscillator that has published ADEV and phase noise. It is of course extremely easy to screw it up :-) What I have found is that most crystal oscillators seem to obey the flicker model. I have been able to measure flicker noise on crystals that were not installed in an oscillator, and then install them in an oscillator and the ADEV turned out to be what was predictable from the phase noise. It really works! Indeed. However, one has to be careful with ones methods, or one will not estimate the noise-levels properly. Regardless if you use the frequency method or time method, doing meterologically sound and traceable measurements remains a difficult task indeed. 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.
Re: [time-nuts] Phase noise from Allan Deviation ?
Hi One of the reasons the TS2700’s went out of favor is the “quality” of the CDMA signals available. The design assumption was that the CDMA carriers provided timing as good as GPS on their over the air systems. After the units had been in the field for a while it became apparent that the 2700’s were not performing up to expectations. Further investigation turned up a range of issues that degraded the CDMA timing relative to GPS. A lot of it boiled down to “we are a phone service not a time service”. System wise, CDMA gets into trouble at the 10us level. GPS is in trouble at the 100 ns level…. Yes there are all sorts of rules and regulations. In the end it’s “this works fine” that trumps a lot of them. Bob > On Dec 14, 2015, at 1:15 PM, Charles Steinmetzwrote: > > Tom wrote: > >> The 'unknown' is a Rubidium oscillator locked to CDMA pilot (TS2700) > > There is very little information publicly available on the Symmetricom > "BesTime" engine ("BTE"), but after playing with a few TS2700s for quite some > time, including monitoring a number of internal signals, several things > became apparent. First, the 2700 does not seem to discipline the PRS10. The > rubidium runs open loop and the BTE keeps track of the offset and the drift > rate from "BTE time" (which is synthesized from all available sources -- > however many CDMA signals it is receiving, plus any wireline telco timing > signals and the PRS10 -- using a proprietary algorithm to estimate the > reliability of each source and outputting BTE time and frequency using DDS). > Hobby users won't be feeding the unit any telco timing signals, so the BTE > has only the CDMA signals to work from. During holdover (and assuming no > telco timing signals), the Rb is the sole input to the BTE, which uses the > stored offset and drift to calculate BTE time. > > I found that the TS-2700 is more than an order of magnitude less stable than > a Trimble Thunderbolt, even with a full complement of rock-solid CDMA > sources. This may vary somewhat, depending on the CDMA equipment in use at > any particular location and the diligence of the CDMA operator. > > Best regards, > > Charles > > > ___ > 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. ___ 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.
Re: [time-nuts] Phase noise from Allan Deviation ?
On 12/13/2015 7:15 PM, Tom McDermott wrote: It brings up a question: Is it possible to estimate the phase noise of that internal crystal from the ADEV measurements? There are a bunch of papers that go the other way: from Phase Noise to Adev. Searching brings up only one paper that goes from ADEV to Phase Noise but it's text does not seem to be readily available. It apparently models the oscillator as a couple of well known error models. -- Tom, N5EG] In general, you cannot determine phase noise from ADEV, even though you can determine ADEV from phase noise. This is just a mathematical reality. Mike Fischer (of HP) presented papers at the 1977 and 1978 that show conversions between PN and ADEV for individual noise processes, where each process has a specific slope of amplitude vs frequency. The only time you can go from ADEV to PN is if you can isolate a process. In the specific case of crystal oscillators, in general, they follow a flicker noise of frequency process model close to the carrier (within 100 Hz). You can often assume that ADEV is dominated by this process and therefore translate it to an equivalent PN. The way to tell if ADEV is dominated by this process is that it will be independent of tau, for tau of 0.1 sec or less. On the IEEE UFFC site, there is a tutorial on crystal oscillator design by Mike Driscoll (you don't have to be a member to access it). I believe this covers this topic. You go from ADEV to "frequency noise" and then to phase noise You can "practice" this calculation on any crystal oscillator that has published ADEV and phase noise. It is of course extremely easy to screw it up :-) What I have found is that most crystal oscillators seem to obey the flicker model. I have been able to measure flicker noise on crystals that were not installed in an oscillator, and then install them in an oscillator and the ADEV turned out to be what was predictable from the phase noise. It really works! Rick Karlquist N6RK ___ 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.
Re: [time-nuts] Phase noise from Allan Deviation ?
Tom, On 12/14/2015 04:15 AM, Tom McDermott wrote: I've constructed a homebrew setup to measure time intervals using a software defined radio. Basically a single-channel downconversion to about one hertz, then count samples from the SDR clock to time stamp the zero crossings. This is done in gnuradio and saved to a file for post processing. The resolution is theoretically good, but the accuracy is unknown. The result produces ADEV vs. Tau charts with reasonably sane looking results. The 'unknown' is a Rubidium oscillator locked to CDMA pilot (TS2700), and the internal crystal oscillator in the SDR radio as the reference. Thus, ADEV probably is mostly measuring that internal crystal rather than the TS2700. Later on a GPSDO will be tried as the reference clock to see if the Adev results are better. You will see the SDR clock noise, as it most likely will dominate. It brings up a question: Is it possible to estimate the phase noise of that internal crystal from the ADEV measurements? There are a bunch of papers that go the other way: from Phase Noise to Adev. Searching brings up only one paper that goes from ADEV to Phase Noise but it's text does not seem to be readily available. It apparently models the oscillator as a couple of well known error models. Yes. Allan Deviation was invented in order to reliably estimate the strengths of the different noise-types and separate them. Today we can do this in the phase-noise domain, but that was hard to do in the 60thies so they had to use counters. Look at the Allan Deviation wikipedia article, where I have included the formulas for various forms of noises and their ADEV curve. These are the formulas you need. In order to separate white phase noise from flicker noise, ADEV isn't a good tool, so you need to use the MDEV instead. David Allan recommends MDEV for this purpose, as the lack of separation was nagging him and it took some additional 15 years before the problem was solved. In order to estimate the noise levels, you do a curve fit of the AVAR or MVAR to the noise-slopes. I recommend the writings of Francois Vernotte. He has study the field and understood the matching process needed. As always, remember that the system bandwidth B will be important to the estimation of the noise-level of a source. The sigma-counters for instance have much lower system bandwidth to reduce the white phase noise, but as you estimate you will can get the wrong estimated value unless you use the correct value. For omega-counters, Vernotte showed how the proper formulas to be used looks, great work there. The PDEV is so far the optimum processing-tool, but MDEV (which is easier to get right) is not far behind. No one using omega-counters can get the PDEV right to this day, but we are discussing how to correctly remedy that. 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.
Re: [time-nuts] Phase noise from Allan Deviation ?
Hi Tom, On 14/12/15 03:15, You wrote: I've constructed a homebrew setup to measure time intervals using a software defined radio. Basically a single-channel downconversion to about one hertz, then count samples from the SDR clock to time stamp the zero crossings. This is done in gnuradio and saved to a file for post processing. The resolution is theoretically good, but the accuracy is unknown. Very interesting. Would you consider making your flowgraph available ? I have done similar things with just thresholding and looking for the start of second (or minute) marker of various distant radio clocks, and then graphing how far apart they were, as well as feeding NTP. 73s Iain ___ 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.
Re: [time-nuts] Phase noise from Allan Deviation ?
>From PN to ADEV you integrate under the PN-curve with a certain weighting-function. This paper shows the weighting function: https://www.researchgate.net/publication/6308454_Considerations_on_the_measurement_of_the_stability_of_oscillators_with_frequency_counters to go the other way around you would indeed have to assume the shape of the PN-curve, such as the Leeson model or similar. This is quite an assumption and probably best avoided. What you want to do is rather than mix down to 1Hz instead mix down to maybe 1 kHz...200 kHz or so - stream all those phase-samples to the computer, and compute the PN as the PSD of the phase data. As an aside I've found that FFT (e.g. Stable32 uses this) sometimes gives funny PSD results, and the Welch-method (periodogram) is more robust. Anders On Mon, Dec 14, 2015 at 5:15 AM, Tom McDermottwrote: > I've constructed a homebrew setup to measure time intervals using a > software defined radio. Basically a single-channel downconversion to > about one hertz, then count samples from the SDR clock to time stamp > the zero crossings. This is done in gnuradio and saved to a file for > post processing. The resolution is theoretically good, but the accuracy > is unknown. > > The result produces ADEV vs. Tau charts with reasonably sane looking > results. > The 'unknown' is a Rubidium oscillator locked to CDMA pilot (TS2700), > and the internal crystal oscillator in the SDR radio as the reference. > Thus, ADEV probably is mostly measuring that internal crystal rather than > the TS2700. Later on a GPSDO will be tried as the reference clock to > see if the Adev results are better. > > It brings up a question: Is it possible to estimate the phase noise of > that > internal crystal from the ADEV measurements? There are a bunch of > papers that go the other way: from Phase Noise to Adev. Searching > brings up only one paper that goes from ADEV to Phase Noise but it's text > does not seem to be readily available. It apparently models the oscillator > as a couple of well known error models. > > -- Tom, N5EG > ___ > 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. > ___ 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.
Re: [time-nuts] Phase noise from Allan Deviation ?
Tom wrote: The 'unknown' is a Rubidium oscillator locked to CDMA pilot (TS2700) There is very little information publicly available on the Symmetricom "BesTime" engine ("BTE"), but after playing with a few TS2700s for quite some time, including monitoring a number of internal signals, several things became apparent. First, the 2700 does not seem to discipline the PRS10. The rubidium runs open loop and the BTE keeps track of the offset and the drift rate from "BTE time" (which is synthesized from all available sources -- however many CDMA signals it is receiving, plus any wireline telco timing signals and the PRS10 -- using a proprietary algorithm to estimate the reliability of each source and outputting BTE time and frequency using DDS). Hobby users won't be feeding the unit any telco timing signals, so the BTE has only the CDMA signals to work from. During holdover (and assuming no telco timing signals), the Rb is the sole input to the BTE, which uses the stored offset and drift to calculate BTE time. I found that the TS-2700 is more than an order of magnitude less stable than a Trimble Thunderbolt, even with a full complement of rock-solid CDMA sources. This may vary somewhat, depending on the CDMA equipment in use at any particular location and the diligence of the CDMA operator. Best regards, Charles ___ 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.
[time-nuts] Phase noise from Allan Deviation ?
I've constructed a homebrew setup to measure time intervals using a software defined radio. Basically a single-channel downconversion to about one hertz, then count samples from the SDR clock to time stamp the zero crossings. This is done in gnuradio and saved to a file for post processing. The resolution is theoretically good, but the accuracy is unknown. The result produces ADEV vs. Tau charts with reasonably sane looking results. The 'unknown' is a Rubidium oscillator locked to CDMA pilot (TS2700), and the internal crystal oscillator in the SDR radio as the reference. Thus, ADEV probably is mostly measuring that internal crystal rather than the TS2700. Later on a GPSDO will be tried as the reference clock to see if the Adev results are better. It brings up a question: Is it possible to estimate the phase noise of that internal crystal from the ADEV measurements? There are a bunch of papers that go the other way: from Phase Noise to Adev. Searching brings up only one paper that goes from ADEV to Phase Noise but it's text does not seem to be readily available. It apparently models the oscillator as a couple of well known error models. -- Tom, N5EG ___ 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.