> > In article <2g7p56$9s9@crl2.crl.com>, lreeves@crl.com (Les Reeves) writes:
> > > The colorburst frequency is not only cast in stone-it is extremely accurate.
> > > It is more accurate as a frequency reference than WWV. This is provided
> > > that you are tuned to a network-supplied program.
>
> > Also, at that time it was stated that the networks used rubidium-clock
> > frequency standards, which are secondary standards: They're awfully good but
> > they still have to be calibrated against something better. NIST (the folks who
> > run WWV) uses cesium-beam clocks, which are primary standards, needing no
> > calibration for frequency. Have the networks since upgraded to cesium-beam
> > clocks? And, given that the local stations probably haven't, does it matter
> > anyway? Even if they have, they're still "only" as good as NIST's clocks, so
> > why should one over-the-air signal be better than another? (propagation
> > changes on shortwave, maybe?)
> >
>
> Colorburst transmit frequency is required to be +/-10Hz (5.5873E-6 or 5.5873ppm).
> PPM means parts per million.
> This requires a good ovenized oscillator(that isn't cheap). Rubidium Oscillators
> go for about $20,000 I think. Cesium Beam clocks are > $200,000.
> The clocks that NIST uses are the best in the world. They have about 10 of them
> that are all averaged together.
>
> WWV, however loses a lot in its method of transmission and to propagation effects.
> Received accuracy (if you have a stable enough PLL to track it without further
> loss of accuracy) is about 1E-7 (0.1ppm) frequency accuracy and 1ms for timing.
I couldn't find an exact figure to back this up, but quoting "NBS Special
Publication 432" (National Bureau of Standards, which is now the
National Institute of Standards and Technology (NIST)) (#432 supersedes
publication 236) for WWV (Fort Collins, Colorado) & WWVH (Kauai,
Hawaii): "Accuracy and Stability: The time and frequency broadcasts
are controlled by the primary NBS Frequency Standard in Boulder,
Colorado. The frequencies as transmitted are accurate to within one
part in 100 billion (1x10^11) at all times. Deviations are normally
less than one part in 1000 billion (1x10^12) from day to day. However,
changes in propagation medium (causing Doppler effect, diurnal shifts,
etc.) result in fluctuations in the carrier frequencies AS RECEIVED by
the user that may be much greater than the uncertainty described above.".
A binary coded decimal (BCD) time code is also transmitted on a 100Hz
subcarrier by WWV & WWVH.
The audio portion of these broadcasts can be heard via telephone (but
not the RF carriers). At (303)499-7111 for WWV and (808)335-4363 for WWVH.
The accuracy received anywhere in the contiguous 48 states is 30ms or
better.
Now if you want real accuracy ( :-) ), try WWVB (Fort Collins,
Colorado). This is a BCD time code only (1 bit per second!) on a
60KHz radio carrier. And to quote publication 432 again: "The frequency
of WWVB is normally within its prescribed value to better than 1 part in
100 billion (1x10^11). Deviations from day to day are less than 5 parts
in 1000 billion (5x10^12). Effects of the propagation medium on received
signals are relatively minor at low frequencies; therefore, frequency
comparisons to better than 1 part in 10^11 are possible using appropriate
receiving and averaging techniques.".
From publication 432: "Frequency Calibration Service Using Network
Television: For those users who require only frequency calibrations,
an alternative to the radio broadcasts is available. This service provides
a means of calibrating oscillators traceable to NBS. It gives the user
the option of calibrating his oscillator quickly at very low cost, with
modest accuracy, or of expending more time and money for higher accuracy.".
"The service is very reliable because the networks use extremely
stable rubidium or cesium oscillators to generate the 3.58MHz color
subcarrier frequency which is transmitted with all color programs. The
color signal is then used as a transfer standard. Any oscillator that
has a frequency of 10/N MHz, where N is any integer from 1 to 100, can
be calibrated.".
"If a user wants to make a calibration, he compares the color signal
coming from the network centers in New York City (or Los Angeles for
those on the West Coast) with his local oscillator. NBS monitors the same
network signals and publishes the difference between the network oscillators
and the NBS Frequency Standard in the monthly NBS Time and Frequency
Services Bulletin. A user then knows two things: (1) the difference
between his oscillator and the network oscillators (by measurement)
and (2) the difference between the networks and NBS (by publication).
With this information, he can easily compute the difference between
his oscillator and NBS. Thus, his calibration is traceable to the NBS
Frequency Standard.".
"NBS has developed two methods for making these frequency calibrations.
Equipment is commercially available for both methods.".
"Color Bar Comparator Method: The color bar comparator is a simple
circuit that connects to a standard color television set. It produces
a colored bar on the screen that changes color or moves across the
screen at a rate that depends on the frequency difference between the
user's oscillator and the TV network signal. By timing these changes
with a stopwatch and referring to the data published by NBS, an oscillator
can be rapidly calibrated to an accuracy of 1 part in 1 billion (1x10^9).".
"Digital Offset Computer Method: The second method, using a digital
offset computer, provides an automatic means of calibrating high-
quality crystal or atomic oscillators. It compares a signal from the
user's oscillator with the TV color signal and displays the frequency
difference on the TV screen as parts in 100 billion (parts in 10^11).
If measurements are averaged over about 15 minutes, a calibration
accuracy of one part in 100 billion can usually be achieved.".
NBS time can also be received via the GOES (Geostationary Operational
Environmental Satellite) satellites of the National Oceanic and Atmospheric
Administration (NOAA).
> Even to keep this accuracy would cost you at least $1000. Stratum 3 oscillators used in non-central office telephone equipment are 4.7ppm and cost at least $2000.
>
> Since the colorburst crystal in your TV is > 100ppm, any PLL that uses that
> crystal to lock onto an external source cannot be any better than 100ppm.
I am building a self-resetting clock which uses WWVB and should have
an accuracy of around 1ms (which is much more than needed for a clock :-) ).
I have a current version which has been running for over 10 years which
as I remember is accurate to around 1/256 of a second. So the important
factor for accuracy while NOT receiving the WWVB signal is the STABILITY
of the crystal (not the frequency accuracy).
For more information on receiving WWVB, see Don Lancaster's articles
in Radio Electronics: July 1972 (pages 54-58), August 1972 (pages 60-62),
August 1973 (pages 48-51 - this starts the WWVB part), and September
1973 (pages 98,101). I only used the design for the receiving antenna
and preamp (which are not my areas of expertise) and threw out the rest
of the design and replaced it with a microprocessor (note the dates on
the articles :-) ). I am lucky that I live in Colorado, so I didn't need
to use any of the fancier receiving methods which Don Lancaster describes
in his articles.
> Incidently, NIST was working on a computer system where you could request time
> and frequency by modem. It would figure out the delay of the telephone path
> and compensate for it. Some manufacturers (True Time was one of them) was working
(303)494-4774 (1200 baud, 8bit, 1stop, no parity).
> > In article <2g7p56$9s9@crl2.crl.com>, lreeves@crl.com (Les Reeves) writes:
> > > The colorburst frequency is not only cast in stone-it is extremely accurate.
> > > It is more accurate as a frequency reference than WWV. This is provided
> > > that you are tuned to a network-supplied program.
>
> > Also, at that time it was stated that the networks used rubidium-clock
> > frequency standards, which are secondary standards: They're awfully good but
> > they still have to be calibrated against something better. NIST (the folks who
> > run WWV) uses cesium-beam clocks, which are primary standards, needing no
> > calibration for frequency. Have the networks since upgraded to cesium-beam
> > clocks? And, given that the local stations probably haven't, does it matter
> > anyway? Even if they have, they're still "only" as good as NIST's clocks, so
> > why should one over-the-air signal be better than another? (propagation
> > changes on shortwave, maybe?)
> >
>
> Colorburst transmit frequency is required to be +/-10Hz (5.5873E-6 or 5.5873ppm).
> PPM means parts per million.
> This requires a good ovenized oscillator(that isn't cheap). Rubidium Oscillators
> go for about $20,000 I think. Cesium Beam clocks are > $200,000.
> The clocks that NIST uses are the best in the world. They have about 10 of them
> that are all averaged together.
>
> WWV, however loses a lot in its method of transmission and to propagation effects.
> Received accuracy (if you have a stable enough PLL to track it without further
> loss of accuracy) is about 1E-7 (0.1ppm) frequency accuracy and 1ms for timing.
I couldn't find an exact figure to back this up, but quoting "NBS Special
Publication 432" (National Bureau of Standards, which is now the
National Institute of Standards and Technology (NIST)) (#432 supersedes
publication 236) for WWV (Fort Collins, Colorado) & WWVH (Kauai,
Hawaii): "Accuracy and Stability: The time and frequency broadcasts
are controlled by the primary NBS Frequency Standard in Boulder,
Colorado. The frequencies as transmitted are accurate to within one
part in 100 billion (1x10^11) at all times. Deviations are normally
less than one part in 1000 billion (1x10^12) from day to day. However,
changes in propagation medium (causing Doppler effect, diurnal shifts,
etc.) result in fluctuations in the carrier frequencies AS RECEIVED by
the user that may be much greater than the uncertainty described above.".
A binary coded decimal (BCD) time code is also transmitted on a 100Hz
subcarrier by WWV & WWVH.
The audio portion of these broadcasts can be heard via telephone (but
not the RF carriers). At (303)499-7111 for WWV and (808)335-4363 for WWVH.
The accuracy received anywhere in the contiguous 48 states is 30ms or
better.
Now if you want real accuracy ( :-) ), try WWVB (Fort Collins,
Colorado). This is a BCD time code only (1 bit per second!) on a
60KHz radio carrier. And to quote publication 432 again: "The frequency
of WWVB is normally within its prescribed value to better than 1 part in
100 billion (1x10^11). Deviations from day to day are less than 5 parts
in 1000 billion (5x10^12). Effects of the propagation medium on received
signals are relatively minor at low frequencies; therefore, frequency
comparisons to better than 1 part in 10^11 are possible using appropriate
receiving and averaging techniques.".
From publication 432: "Frequency Calibration Service Using Network
Television: For those users who require only frequency calibrations,
an alternative to the radio broadcasts is available. This service provides
a means of calibrating oscillators traceable to NBS. It gives the user
the option of calibrating his oscillator quickly at very low cost, with
modest accuracy, or of expending more time and money for higher accuracy.".
"The service is very reliable because the networks use extremely
stable rubidium or cesium oscillators to generate the 3.58MHz color
subcarrier frequency which is transmitted with all color programs. The
color signal is then used as a transfer standard. Any oscillator that
has a frequency of 10/N MHz, where N is any integer from 1 to 100, can
be calibrated.".
"If a user wants to make a calibration, he compares the color signal
coming from the network centers in New York City (or Los Angeles for
those on the West Coast) with his local oscillator. NBS monitors the same
network signals and publishes the difference between the network oscillators
and the NBS Frequency Standard in the monthly NBS Time and Frequency
Services Bulletin. A user then knows two things: (1) the difference
between his oscillator and the network oscillators (by measurement)
and (2) the difference between the networks and NBS (by publication).
With this information, he can easily compute the difference between
his oscillator and NBS. Thus, his calibration is traceable to the NBS
Frequency Standard.".
"NBS has developed two methods for making these frequency calibrations.
Equipment is commercially available for both methods.".
"Color Bar Comparator Method: The color bar comparator is a simple
circuit that connects to a standard color television set. It produces
a colored bar on the screen that changes color or moves across the
screen at a rate that depends on the frequency difference between the
user's oscillator and the TV network signal. By timing these changes
with a stopwatch and referring to the data published by NBS, an oscillator
can be rapidly calibrated to an accuracy of 1 part in 1 billion (1x10^9).".
"Digital Offset Computer Method: The second method, using a digital
offset computer, provides an automatic means of calibrating high-
quality crystal or atomic oscillators. It compares a signal from the
user's oscillator with the TV color signal and displays the frequency
difference on the TV screen as parts in 100 billion (parts in 10^11).
If measurements are averaged over about 15 minutes, a calibration
accuracy of one part in 100 billion can usually be achieved.".
NBS time can also be received via the GOES (Geostationary Operational
Environmental Satellite) satellites of the National Oceanic and Atmospheric
Administration (NOAA).
> Even to keep this accuracy would cost you at least $1000. Stratum 3 oscillators used in non-central office telephone equipment are 4.7ppm and cost at least $2000.
>
> Since the colorburst crystal in your TV is > 100ppm, any PLL that uses that
> crystal to lock onto an external source cannot be any better than 100ppm.
I am building a self-resetting clock which uses WWVB and should have
an accuracy of around 1ms (which is much more than needed for a clock :-) ).
I have a current version which has been running for over 10 years which
as I remember is accurate to around 1/256 of a second. So the important
factor for accuracy while NOT receiving the WWVB signal is the STABILITY
of the crystal (not the frequency accuracy).
For more information on receiving WWVB, see Don Lancaster's articles
in Radio Electronics: July 1972 (pages 54-58), August 1972 (pages 60-62),
August 1973 (pages 48-51 - this starts the WWVB part), and September
1973 (pages 98,101). I only used the design for the receiving antenna
and preamp (which are not my areas of expertise) and threw out the rest
of the design and replaced it with a microprocessor (note the dates on
the articles :-) ). I am lucky that I live in Colorado, so I didn't need
to use any of the fancier receiving methods which Don Lancaster describes
in his articles.
> Incidently, NIST was working on a computer system where you could request time
> and frequency by modem. It would figure out the delay of the telephone path
> and compensate for it. Some manufacturers (True Time was one of them) was working
(303)494-4774 (1200 baud, 8bit, 1stop, no parity).
