Frequency/Time Services
General:
http://www.itu.int/rec/R-REC-TF.768/en
http://www.itu.int/dms_pubrec/itu-r/rec/tf/R-REC-TF.768-6-200305-S!!PDF-E.pdf (latest with station overview)
digilander.libero.it/occultazioni/segnali-di-tempo.pdf
ftp://ftp2.bipm.org/pub/tai/scale/timesignals.pdf
(Note that the Swiss HPG service listed in some of the documents is no longer in operation)
MSF (60 kHz)
http://www.npl.co.uk/science-technology/time-frequency/products-and-services/time/msf-radio-time-signal
http://www.npl.co.uk/science-technology/time-frequency/products-and-services/time/time-and-frequency-bulletins-for-msf-droitwich-and-gps-off-air-signals
http://www.npl.co.uk/science-technology/time-frequency/products-and-services/time/msf-bulletins
WWVB (60 kHz)
Michael A. Lombardi and Glenn K. Nelson: "WWVB: A Half Century of Delivering Accurate Frequency and Time by Radio"
The phase inversion that was introduced to enhance the reception of the signal by domestic radio-controlled clocks left PLL-based WWVB receivers useless. Techniques, such as squaring of the 60 kHz carrier, are now required for receivers to phase lock.
RBU (66 2/3 kHz)
http://en.wikipedia.org/wiki/RBU_%28radio_station%29
DCF77 (77.5 kHz)
http://www.ptb.de/cms/en/fachabteilungen/abt4/fb-44/ag-442/dissemination-of-legal-time/dcf77.html
http://www.ptb.de/cms/fileadmin/internet/fachabteilungen/abteilung_4/4.4_zeit_und_frequenz/pdf/PTBM_50a_DCF77_engl.pdf
Loran / Chayka (100 kHz)
After the switch-off af European Loran-C by December 31, 2015, the only station left is the one at Anthorn (GRI = 67310 µs), apparently to allow the development of e-Loran services. With the previous noise margin figures in the 40 dB range reported by my FS700 receiving the transmitter at Sylt, it was sad to see the noise margin drop to about 18-23 dB when I switched to Anthorn. Also, the FS700 regularly loses lock due to signal conditions. The Russian Chayka system, in particular the chain with GRI = 80000 µs, has on the other hand become more interesting. Here in Denmark, as of July 2017, my FS700 receives Chayka from the Slonim site some 880 km away with about 20-28 dB noise margin. Still, the reception is plagued by recurrent and annoying resync attempts by the FS700.
For background information refer to:
_http://www.uscg.mil/history/docs/LoranCUsersHandbook.pdf
Various proceedings and papers on Loran from the International Loran Association (previously known as the Wild Goose Association), are available at the Royal Institute of Navigation, as the ILA ceased its operations in 2011:
http://ila.rin.org.uk/proceedings.htm
http://ila.rin.org.uk/otherarchives/1992%20Loran-C%20User%20Handbook%20-%20USCG.pdf
http://ila.rin.org.uk/otherarchives/2007%20eLoran%20Definition%20Document-1.0.pdf
European LORAN-C unavailability schedules, were available at http://www.loran-europe.eu/viewpage.php?page_id=8, but the link appears to be a dead after the closure of the European Loran-C service in December 2015.
http://www.chronos.co.uk/index.php/en/news-and-pr/news-page/1605-taviga-welcomes-the-continued-transmission-of-eloran-timing-and-data-signals-from-the-uk
TDF / France Inter (162 kHz)
http://www.metrologie-francaise.fr/fr/references/temps-frequence.asp
http://tvignaud.pagesperso-orange.fr/am/allouis/allouis-heure.htm
http://en.wikipedia.org/wiki/T%C3%A9l%C3%A9_Distribution_Fran%C3%A7aise
The transmitter was taken off air by December 31, 2016, but the time signal service has since then been reassumed.
Droitwich (198 kHz)
http://www.npl.co.uk/science-technology/time-frequency/products-and-services/time/time-and-frequency-bulletins-for-msf-droitwich-and-gps-off-air-signals
http://www.npl.co.uk/science-technology/time-frequency/products-and-services/time/droitwich-bulletins
Solec Kujawski (225 kHz)
This powerful transmitter made for regular AM broadcasts by the Polish Radio SA is not really intended for time/frequency dissemination. However, the accuracy of the carrier frequency, claimed to be controlled by a rubidium reference under continuous surveillance, should be within 2 ⋅ 10-12 (Reference in Polish: "Naziemne i satelitarne (Galileo, GPS) źródła sygnałów wzorcowych częstotliwości i czasu dla potrzeb synchronizacji sieci telekomunikacyjnych oraz lokalizowania wywołań alarmowych", Warszawa, grudzień 2006). Later updates in Wikipedia states that the carrier frequency is generated by "a set of twin high-accuracy thermally-stabilized quartz oscillators." Measurements show that the carrier is way too off for any reference purposes.
Furthermore, the carrier is subject to an asymmetric phase modulation, taking place once per minute. See additional details below and check out:
http://en.wikipedia.org/wiki/Solec_Kujawski_radio_transmitter
http://pl.wikipedia.org/wiki/RCN_Solec_Kujawski
Uklad Antenowy w Radiowym Centrum Nadawczym Solec Kujawski (Polskie Radio S.A.), Daniel Jósef Bem, Politechnika Wroclawska (http://www.ire.pw.edu.pl/~jjarkow/RADIO_DRM/ANtena%20RCN.pdf)
http://www.itu.int/rec/R-REC-TF.768/en
http://www.itu.int/dms_pubrec/itu-r/rec/tf/R-REC-TF.768-6-200305-S!!PDF-E.pdf (latest with station overview)
digilander.libero.it/occultazioni/segnali-di-tempo.pdf
ftp://ftp2.bipm.org/pub/tai/scale/timesignals.pdf
(Note that the Swiss HPG service listed in some of the documents is no longer in operation)
MSF (60 kHz)
http://www.npl.co.uk/science-technology/time-frequency/products-and-services/time/msf-radio-time-signal
http://www.npl.co.uk/science-technology/time-frequency/products-and-services/time/time-and-frequency-bulletins-for-msf-droitwich-and-gps-off-air-signals
http://www.npl.co.uk/science-technology/time-frequency/products-and-services/time/msf-bulletins
WWVB (60 kHz)
Michael A. Lombardi and Glenn K. Nelson: "WWVB: A Half Century of Delivering Accurate Frequency and Time by Radio"
The phase inversion that was introduced to enhance the reception of the signal by domestic radio-controlled clocks left PLL-based WWVB receivers useless. Techniques, such as squaring of the 60 kHz carrier, are now required for receivers to phase lock.
RBU (66 2/3 kHz)
http://en.wikipedia.org/wiki/RBU_%28radio_station%29
DCF77 (77.5 kHz)
http://www.ptb.de/cms/en/fachabteilungen/abt4/fb-44/ag-442/dissemination-of-legal-time/dcf77.html
http://www.ptb.de/cms/fileadmin/internet/fachabteilungen/abteilung_4/4.4_zeit_und_frequenz/pdf/PTBM_50a_DCF77_engl.pdf
Loran / Chayka (100 kHz)
After the switch-off af European Loran-C by December 31, 2015, the only station left is the one at Anthorn (GRI = 67310 µs), apparently to allow the development of e-Loran services. With the previous noise margin figures in the 40 dB range reported by my FS700 receiving the transmitter at Sylt, it was sad to see the noise margin drop to about 18-23 dB when I switched to Anthorn. Also, the FS700 regularly loses lock due to signal conditions. The Russian Chayka system, in particular the chain with GRI = 80000 µs, has on the other hand become more interesting. Here in Denmark, as of July 2017, my FS700 receives Chayka from the Slonim site some 880 km away with about 20-28 dB noise margin. Still, the reception is plagued by recurrent and annoying resync attempts by the FS700.
For background information refer to:
_http://www.uscg.mil/history/docs/LoranCUsersHandbook.pdf
Various proceedings and papers on Loran from the International Loran Association (previously known as the Wild Goose Association), are available at the Royal Institute of Navigation, as the ILA ceased its operations in 2011:
http://ila.rin.org.uk/proceedings.htm
http://ila.rin.org.uk/otherarchives/1992%20Loran-C%20User%20Handbook%20-%20USCG.pdf
http://ila.rin.org.uk/otherarchives/2007%20eLoran%20Definition%20Document-1.0.pdf
European LORAN-C unavailability schedules, were available at http://www.loran-europe.eu/viewpage.php?page_id=8, but the link appears to be a dead after the closure of the European Loran-C service in December 2015.
http://www.chronos.co.uk/index.php/en/news-and-pr/news-page/1605-taviga-welcomes-the-continued-transmission-of-eloran-timing-and-data-signals-from-the-uk
TDF / France Inter (162 kHz)
http://www.metrologie-francaise.fr/fr/references/temps-frequence.asp
http://tvignaud.pagesperso-orange.fr/am/allouis/allouis-heure.htm
http://en.wikipedia.org/wiki/T%C3%A9l%C3%A9_Distribution_Fran%C3%A7aise
The transmitter was taken off air by December 31, 2016, but the time signal service has since then been reassumed.
Droitwich (198 kHz)
http://www.npl.co.uk/science-technology/time-frequency/products-and-services/time/time-and-frequency-bulletins-for-msf-droitwich-and-gps-off-air-signals
http://www.npl.co.uk/science-technology/time-frequency/products-and-services/time/droitwich-bulletins
Solec Kujawski (225 kHz)
This powerful transmitter made for regular AM broadcasts by the Polish Radio SA is not really intended for time/frequency dissemination. However, the accuracy of the carrier frequency, claimed to be controlled by a rubidium reference under continuous surveillance, should be within 2 ⋅ 10-12 (Reference in Polish: "Naziemne i satelitarne (Galileo, GPS) źródła sygnałów wzorcowych częstotliwości i czasu dla potrzeb synchronizacji sieci telekomunikacyjnych oraz lokalizowania wywołań alarmowych", Warszawa, grudzień 2006). Later updates in Wikipedia states that the carrier frequency is generated by "a set of twin high-accuracy thermally-stabilized quartz oscillators." Measurements show that the carrier is way too off for any reference purposes.
Furthermore, the carrier is subject to an asymmetric phase modulation, taking place once per minute. See additional details below and check out:
http://en.wikipedia.org/wiki/Solec_Kujawski_radio_transmitter
http://pl.wikipedia.org/wiki/RCN_Solec_Kujawski
Uklad Antenowy w Radiowym Centrum Nadawczym Solec Kujawski (Polskie Radio S.A.), Daniel Jósef Bem, Politechnika Wroclawska (http://www.ire.pw.edu.pl/~jjarkow/RADIO_DRM/ANtena%20RCN.pdf)
The services as a timing source
If you intend to build an off-air receiver to exploit a broadcast time service for a frequency reference, you need to get familiar with the behavior of the service at your reception site. This is essential to understand the dynamics of the signal, to make educated design decisions and to estimate the level of accuracy you could expect. In other words: Before designing a receiver, find out what you are trying to lock to. Due to the nature of propagation and the influence of the local noise environment the usefulness of a service depends very much on where you are located, and the findings for your reception site may not necessarily apply to another site. As a part of the investigations, you should find out where the optimal location of your antenna is. Switched-mode power supplies are well known to disturb long-wave services. However, another source of noise and much annoyance is the mains cable ducting, often hidden in walls, floors and ceilings, which can make it difficult to find a quiet spot in some buildings.
The two graphs show an example of the accumulated phase (top) and the level (bottom) for these 5 services received at my lab located in the vicinity of Copenhagen, in Denmark:
The services were captured with my E-field antenna, located more than 10 m from the nearest mains cable ducting. The capture started 2014-01-14 at 10:10:18, and was terminated 2014-01-15 at 12:03:14. (New graphs will be uploaded later) |
The signal from the antenna was filtered with a Krohn-Hite 3202 filter set to 200 kHz cut-off, and digitised with National Instruments USB-6366 X-series DAQ using 500 kHz sampling. The data processing was done in LabVIEW, and the phase and level were calculated for each 480 k samples. The reference for the sampling was provided by the Stanford FS700 frequency standard, so all phases are relative to a clock derived from Loran-C.
The large phase disturbances during destructive fading are quite clear to spot. These will definitely kill any off-air receiver with too wide a bandwidth and with no precautions to address fadings and phase jumps. Simple off-air receivers are typically useful for daytime use only. The phase variations are generally larger during the night, and they may become quite nasty around sunrise and sunset. During this particular capture, for instance, only the reception of the DCF77 and Droitwich services made it through the sunrise.
It should be clear that phase jumps are a challenge to every off-air receiver. However, even when these jumps are handled properly, the challenge remains to cope with the slowly varying phase variations if you aim for accuracy. Of particular interest is the diurnal phase variation. Take a look at the curved shape of the phase during daytime, clearly visible during the first hours of this capture. It's also clear that there's a distinct difference between night and day. An important phenomenon that plays a trick is the changing electron density in the atmosphere. The delay changes are in orders of microseconds, all depending on the trajectory, the season, solar activity, etc. In order to improve the accuracy of your off-air receiver you need to compensate for the trajectory delay, or perhaps do comparisons at the same time of the day, preferably at a time where the trajectory is exposed to as much sun as possible.
The large phase disturbances during destructive fading are quite clear to spot. These will definitely kill any off-air receiver with too wide a bandwidth and with no precautions to address fadings and phase jumps. Simple off-air receivers are typically useful for daytime use only. The phase variations are generally larger during the night, and they may become quite nasty around sunrise and sunset. During this particular capture, for instance, only the reception of the DCF77 and Droitwich services made it through the sunrise.
It should be clear that phase jumps are a challenge to every off-air receiver. However, even when these jumps are handled properly, the challenge remains to cope with the slowly varying phase variations if you aim for accuracy. Of particular interest is the diurnal phase variation. Take a look at the curved shape of the phase during daytime, clearly visible during the first hours of this capture. It's also clear that there's a distinct difference between night and day. An important phenomenon that plays a trick is the changing electron density in the atmosphere. The delay changes are in orders of microseconds, all depending on the trajectory, the season, solar activity, etc. In order to improve the accuracy of your off-air receiver you need to compensate for the trajectory delay, or perhaps do comparisons at the same time of the day, preferably at a time where the trajectory is exposed to as much sun as possible.
Anomaly categories - a closer look
I have highlighted typical anomalies to have in consideration when designing an off-air receiver, so that it may take the right precautions when these anomalies occur. To complicate matters, more than one anomaly may appear at the same time. The examples below are taken from real-life acquisitions with my E-field antenna and/or my loop antenna, and an acquisition device, such as the National Instruments USB-6366 X-series DAQ or the USB-6251. Data acquisition and processing are done in LabVIEW. The Z3805A GPS receiver and a frequency divider provide the sampling reference to one of the PFI inputs. A Hewlett Packard 461A amplifier and a Krohn-Hite 3550 low-pass filter are inserted between the antenna and the acquisition device. The acquisition setup essentially measures the phase between the carrier of the broadcast service and a local carrier which has a frequency which is a user-defined integer times the sampling rate divided by the number of samples. With a sampling rate governed by GPS the phase of each carrier being analyzed will be relative to GPS.
(MORE GRAPHS TO BE ADDED)
(MORE GRAPHS TO BE ADDED)
Picture of DCF77 service interruption to be uploaded.
Major overhaul of the MSF resulting in several days of service interruption.
|
Service interruption
Short interruptions now and then should be expected for the broadcast time services, typically for reasons of maintenance work, harsh weather conditions, or perhaps replacement of transmitter gear.
The upper figure shows an example of an incident where the DCF77 service got interrupted for a longer period of time, observed during January 2014. The reason behind the specific interruptions in the figure remains unknown. However, the PTB informs on its web pages that electrical detuning of the antenna resonance circuit due to bad weather is to blame for those of the interruptions that have the longest duration. In case of MSF there are both scheduled off-air periods and unscheduled off-air breaks to look out for. Information about the scheduled periods, which typically will last for a few hours but longer for the annual overhaul, is available on the NPL website. The second figure shows the 60 kHz level (in dBV) from my loop antenna from 1st through 20th of March 2015 while the main scheduled outage for 2015 took place. |
Interference from appliances
One common source of interference is switched-mode power supplies where the switching frequency or the harmonics are uncomfortably close to the reception frequency of your receiver. Powerful sources may even overload sensitive receivers.
The upper graph shows the normal, unpolluted spectrum from the loop antenna up to 250 kHz. The RMS level is close to 10 mV. You can identify the Loran spectrum around 100 kHz, and the services at 60 kHz, 66 2/3 kHz, 77.5 kHz, 162 kHz and 198 kHz. The graph in the middle shows what happens when I turn on a microwave oven in the kitchen while using a loop antenna for the reception of time services. The RMS level is about 30 mV. The reason for the interference is not unwanted microwave emission, but the use of an otherwise fine switching power inverter in the oven. The noise couples partly from components inside the oven (interconnections and the inverter transformer's stray field), and partly through the mains cabling. In order to reduce the level of interference, I decided to make a differential-mode mains filter and build it into the oven. The filter is located between the power inverter and the existing common-mode filter. Thanks to the filter the phase excursions in my test setup became smaller, but are far from gone. If I want less interference from the oven, I would have to switch to my E-field antenna. The lower graph demonstrates the challenge of the induction cookers that have become popular for their speed and for allowing the user to control the heat with ease. They do emit quite a powerful magnetic field, and may inject a strong signal into a loop antenna. When I put a frying pan on my small induction hob and turn it on, this is what I get. The RMS level is about 320 mV, and the HP 461A pre-amplifier between the loop antenna and the digitizer is actually clipping. Making experiments with a loop antenna in the attic while cooking with induction gear in the kitchen do not go together well. Remedies include using the outside E-field antenna or a complete relocation of the loop antenna, likely perpendicular to my noisy kitchen! |
Picture to be uploaded.
|
Station interference
Another reason why we get to see interference is that some stations share the same frequency. For instance, the France Inter and the Turkish TRT4 stations share 162 kHz, and Droitwich and a couple of other stations share 198 kHz. Normally, this is not an issue for the reception in the western parts of Europe, but propagation conditions change, and off-air time receivers may pick up the wrong signal.
The figure shows an example of interference where the 198 kHz Droitwich transmitter seems to compete with another 198 kHz source, perhaps Radio Mayak in St. Petersburg, but the origin of interference was never identified. One means to combat potential interfering stations is to use a loop antenna positioned in such a way that the interferer gets nulled out. Other methods include statistical analysis to govern the off-air receiver. In some cases, depending on where your receiver is located, station interference will simply rule out the use of specific broadcast services that share frequency with another service. |
Picture to be uploaded.
|
Ground wave variations
The standard deviation of the phase is clearly smaller during the daytime than during the nighttime, but the phase does not remain flat, as one should think. The ground wave phase is subject to a phase shift change during the daytime. The exposure of sunlight on the propagation trajectory, resulting in a varying electron density which changes the waveguide profile and thus the propagation speed, is one important mechanism. The phase depends not only on the time of day, but also on the day of the year, the frequency and distance to the transmitter. Attempts to model the propagation in the LF range can be found in the literature, and there's even some research into the pertubations associated with earthquakes. However, for the use of time/frequency dissemination, I think it would be fair to say that the area of LF propagation has not received much attention for some time, likely due to the widespread use of GPS. In order to provide a better basis for designing a multi-frequency off-air receiver, I have conducted experiments for some time to measure and model the propagation phase for the common time services.
|
Picture to be uploaded.
|
Skywave propagation with destructive fading
When the skywave becomes comparable with the ground wave in intensity the level and phase of the signal at the receiver site will jump around. The signal from the skywave and the ground wave may eventually cancel each other out, and the phase information around the event is utterly misleading. Especially the conditions around sunset and sunrise can be extremely challenging for an off-air-receiver. As a consequence, the typical off-air receiver suffers from not only excessive phase variations under these conditions, detrimental to the medium-term stability, but also from cycle slips which effectively destroys the long-term stability. The phenomenon depends on the distance to the transmitter, but in large areas within the reception range, common PLL_based off-air receivers should be restricted to use only during periods with daylight.
A well-designed off-air receiver, intended for 24/7 use, should be able to reassume the correct phase after hours of destructive fading. This means that the receiver should be able to gracefully go in and out of a free-run state, and that there's a limit to the drift rate of the unlocked receiver; For instance, 45 degrees phase difference after one hour for a 77.5 kHz carrier equals 1.6 µs and a relative frequency offset of 4.5 * 10e-10. |
Solar activty
Solar activity influences the atmospehere's ionization level and thereby the propagation conditions. It's easier to check this out than you may think: The Geostationary Operational Environmental Satellites (GOES) include X-ray sensors (XRS) to measure the X-ray flux from the sun, and data sets are available to the public from the US National Geophysical Data Center (http://www.ngdc.noaa.gov/stp/satellite/goes/dataaccess.html).
The upper graph shows the unscaled "XRS long wavelength channel irradiance (0.1-0.8 nm)", for 6:00 - 18:00 local time on 2015-03-11, extracted from the time averaged data file from GOES-15 for March 2015 (g15_xrs_1m_20150301_20150331.csv). The x-axis represents the number of minutes starting from 2015-03-01. The peak, a powerful X-class flare, is located at 16:22 UTC (17:22 local time). The middle graph shows the measured phase for the RBU service at 66 2/3 kHz during the same period of time. The lower graph shows the phase for DCF77 at 77.5 kHz. The x-axis shows the measurement number out of 60,000 for that particular day. We see not only that the phase excursions correlate with the X-ray flux, but also that the phase may jump in either direction. |
Solar eclipse
Solar eclipses are definitely not an everyday event, but they do happen. At 20th of March 2015 a solar eclipse took place over the Northern parts of Europe. In Copenhagen, where my lab is located, a partial eclipse started at 8:42 UTC and lasted until 11:00 UTC. The maximum obscuration in Copenhagen was 80.64% at 9:50 UTC.. The map shows the details of the eclipse (click on it for further information). Eclipse map courtesy of Fred Espenak - NASA/Goddard Space Flight Center. For more information on solar and lunar eclipses, see Fred Espenak's Eclipse Web Site: http://eclipse.gsfc.nasa.gov/eclipse.html The graph in the middle shows the phase on 2015-03-01 between 8:00 and 16:00 Copenhagen time for MSF (White), RBU (Red), DCF77 (Green), France Inter (Blue) and BBC 4 from Droitwich (Purple). This behavior is rather typical for a day where nothing out of the ordinary occurs. The numbers on the x-axis represent the measurement numbers for that particular day. With 720 kS chuncks at 500 kHz sampling rate there are 60000 measurements over 24 h. The next graph shows the phase during the day of the eclipse on the 2015-03-20. The yearly overhaul of the MSF transmitter had just been completed the day before so we get to see the influence on the MSF service, too. The phase excursions situated around the time of the maximum obscuration are quite clear to see, but are still still benign compared to the disruptive phase storms one may encounter around sunset, sunrise and during the night. The RBU phase (Red) wobbles quite much in this graph, but exactly how much the eclipse and other conditions each are to blame is hard to tell. |
Thunder storms
The lightning during a thunderstorm results in pulse noise, but most off-air reference should be able to suppress the short phase excursions associated with pulse noise. The real risk of a thunderstorm is the potential corruption of digital communication (RS-232, USB, etc.), computers freezing, or perhaps glitches in clock references. Of course, a close or direct hit of lightning is another beast alltogether, as this becomes a question of having protection against destruction, not just being annoyed by signal corruption. The figure here shows level and phase spikes for the reception of MSF (60 kHz), DCF77 (77.5 kHz), France Inter (162 kHz) and BBC (198 kHz) during a thunderstorm. The receiver, an NI USB-6251, used a 400 kHz sampling clock derived from a 10 MHz GPS reference. At one point all phases jump, which coincided with a nearby lightning strike. The GPS receiver went into holdover for a couple of minutes, but that does not explain the jumps. A closer look at the phase changes reveal that they equal one jump of the 400 kHz sampling clock. |
Time decoding
The timing formats for the MSF, DCF77 and the RBU services are described in the references listed above and serve as a recipe for how to design a software-defined clock. I have implemented time decoders for MSF and DCF77 in LabVIEW in order to investigate methods and evaulate the robustness of designs that can be implemented later in VHDL.
This is a part of the front panel (with captions in Danish) of the time decoder in LabVIEW that can be used with the National Instruments USB-6251, the USB-6366 X-series or similar acquisition hardware.
The signal is provided by two amplified ferrite antennae, one tuned for 60 kHz, the other for 77.5 kHz, conveniently connected in parallel, made possible by the open collector outputs. |
Solec Kujawski as a frequency reference
Above: Phase drift at 225 kHz carrier of the Solec Kujawski transmitter observed on 2015-05-07 (midnight to midnight).
Examples of the phase modulation of the 225 kHz carrier: A burst, and a group of bursts.
Checking the carrier accuracy on 2021-09-04: The phase (time interval drift) of the 10 MHz signal from a 225 kHz receiver over 12 hours relative to the 10 MHz from the Z3805A GPS receiver. With a slope of about -7e-10 the offset is even larger than back in 2015.
|
This longwave transmitter in Poland does not seem to be mentioned at all in the literature (at least not in German or English) when it comes to time/frequency dissemination, though I found a reference to the accuracy of the transmitter in a Polish report on a national time transmission system (see the reference above) which looked promising. Thanks to one of the translation services on the web, I found out that the transmitter's frequency of 225 kHz appears to be governed by a rubidium reference under "continuous surveillance" claiming an accuracy within 2 * 10e-12.
In reality, the accuracy does not live up to expectations. The graph shows the phase relative to GPS over 24 hours on 2015-05-07. With 480 km to the transmitter site I get a quite stable signal during daytime. The result is almost a textbook example of a linear phase drift, equivalent to a relative error of 6.6 * 10e-11, not even close to the specified accuracy. This makes me wonder how well the "continuous surveillance" procedures of the transmitter actually work. According to later updates on the Solec Kujawski transmitter in Wikipedia the carrier is generated by "a set of twin high-accuracy thermally-stabilized quartz oscillators." If there had ever been any plans of providing a carrier of high frequency accuracy, they seem to be all forgotten. Another item of concern is that the phase is modulated. The modulation is not symmetrical, and consists of a varying number of bursts, starting a few seconds after each turn of a minute, which makes it a potential nuisance. The second figure shows an example of a phase modulation burst. The clock period is 20 ms. The bursts typically come in bundles of 6, though larger bundles are seen. The lower figure shows a bundle of 14 phase bursts. The best suggestion I have right now to mitigate the effects of the phase modulation, before having conducted any experiments, would be to gate the modulation out. I would not try to swamp out the phase excursions by a small loop bandwidth or fancy filtering. One successful attempt of gating out the phase modulation is shown under the repair pages. |