Time in Broadcasting: The BBC measurement and technical receiving station at Tatsfield
28 October 2022 tbs.pm/76351
It is not, perhaps generally realised that, in one sense, time is the basis of all broadcasting, for time is the reference against which is measured the frequency of the electromagnetic waves used for radio transmission. Broadcasting frequencies are expressed as so many kilocycles (thousands of cycles) per second, or so many megacycles (millions of cycles) per second.
Time has, of course, many other applications in broadcasting and this little booklet, produced by the British Broadcasting Corporation especially for the Centenary Exhibition of the British Horological Institute, describes some of these. It is hoped that it will be of interest to horologists and others visiting the exhibition.

From ‘Time in Broadcasting’, published in October 1958 as part of the centenary exhibition of the British Horological Institute
Generation and checking of standard frequencies
Among the various activities at the Tatsfield receiving station, the frequencies of BBC and other transmitters are measured with an accuracy as high as 1/10⁸ (one part in 100,000,000). Frequency is a function of time and its unit is the cycle per seconds thus the accuracy with which it can be measured depends upon the accuracy of time-determination. The number of cycles can usually be determined by methods equivalent to direct counting, and readers will know how even the most popular and inexpensive timepieces achieve an accuracy which is high compared to other measuring devices. As an example, a cheap clock which gains or loses one minute per day is measuring time with an accuracy of one part in 1440, or better than 0.1 per cent, while a voltmeter with this accuracy would be well within the most accurate category for electrical measuring instruments (BS.89 sub-standard). Conversely, a clock having an accuracy similar to that of transmitter frequency-measuring equipment – i.e. 1/10⁸ – would not have gained or lost more than 1 second in a period of approximately 3¼ years.
Recent advances in time measurement have reached an accuracy better than 1/10⁹ (one part in 1,000,000,000) by such methods as the use of a natural resonance in the caesium atom for a source of reference.
The daily rotation period of the earth has hitherto been accepted as the fundamental unit of time, but this is now known to be subject to variations of as much as ± four parts in 100,000,000, and is also increasing progressively at about one millisecond per century as a result of tidal friction. This criterion is therefore not quite acceptable for these very accurate measurements, and the more constant time of revolution of the earth around the sun is becoming accepted as the new unit. The reference time-period generally chosen is that of the sidereal year 1900.
The Tatsfield frequency-measuring equipment is checked against various time standards, and the accuracy with which these checks can be made is aided by several factors which make precise measurements of time easier than other physical measurements such as mass or distance. One of these is that it is quite simple to produce an electrical oscillation whose frequency is an exact multiple or sub-multiple of the frequency of another oscillation. If the frequency of the original oscillation is known accurately, the frequencies of the multiples or sub-multiples will
be known with the same accuracy. Another factor is that when two or more oscillations are mixed, frequencies equal to the sums or differences of any of the component frequencies can easily be extracted electrically from the mixture. A third factor is that the frequencies of two independently variable oscillations can be made exactly equal by observing when the frequency difference or beat tone between them is zero. These factors enable any “unknown” frequency to be measured with the help of a limited number of fixed reference frequencies.
The three frequency standards at Tatsfield are in the form of quartz crystals maintained in a state of piezo-electric oscillation by valve circuits. Each crystal is separately mounted in a temperature-controlled compartment or “oven”, and the ovens themselves are in an underground standards room which is thermostatically stabilized at approximately 29°C. In this way the temperature of the quartz is maintained within a few thousandths of a degree centigrade. The two older standards have rectangular quartz plates oscillating at 1 Mc/s [1 Mc/s means 1 megacycle per second] and 5 Mc/s respectively. The latest standard is an Essen ring oscillator. This type was introduced at the National Physical Laboratory in 1934, and afterwards developed by the Post Office until it became accepted as a standard of time measurement with a short-term and long-term accuracy second only to the caesium resonator at the National Physical Laboratory. The quartz is in the form of a ring of 5-cm mean diameter, oscillating circumferentially on its sixth overtone at 100 kc/s, and suspended on threads at nodal points in order to minimize the damping. The ring is contained in a chamber from which the air is removed, also in order to reduce damping, and it is maintained in oscillation by a valve circuit of a type devised by G.G. Gouriet of the BBC Engineering Division Research Department, which has the minimum possible effect on its natural resonant frequency.
Various methods are used for checking the local frequency standards against the standard signals incoming to Tatsfield. In making these comparisons, the signal to be measured and/or the reference signal are generally converted to the same nominal frequency by multiplication or division. Thus, if one of the local crystal standards is to be compared with the Uniform Time Signals, which are transmitted from Rugby as five-millisecond pulses recurring once per second, the crystal frequency is divided by electronic dividers, phonic wheel with gearing, or a combination of both methods, down to pulses at a nominal 1 c/s. The two sets of pulses are then fed to a counter chronoscope which measures and indicates the time difference between the pulses to the fifth decimal place of a second. The chronoscope’ is reset automatically after each reading, making it possible to take forty readings in five minutes. The arithmetic mean of the forty readings is calculated, and the whole procedure of taking a mean of forty readings is repeated at weekly intervals. Any variations in the successive weekly readings can be converted to a frequency variation and expressed as parts in 10⁸. The geometric mean of the variations in each individual set of forty readings gives the “standard error” of the daily time comparison, which is of the order of ± 0.5 milliseconds or better, in normal reception conditions. It should, of course, be remembered that a time signal received by radio can be subject to quite appreciable variation, especially if the signal has come over a long-distance path including one or more reflections from the ionosphere. As the height of the ionosphere alters, the distance travelled by a radio signal reflected from it must vary.
The 200-kc/s carrier of the Droitwich long-wave transmitter is itself a sub-standard frequency guaranteed to 1 part in 10⁶, although the actual working tolerances are considerably closer than this. The signal is checked at Tatsfield by multiplying the received carrier up to 1 Mc/s, and then comparing it with a standard of the same nominal frequency derived from one of the Tatsfield crystals. The two signals are fed to the horizontal and vertical deflecting plates of a cathode-ray oscilloscope, thereby producing an ellipse which rotates slowly if there is any frequency difference. In practice, the 1-Mc/s signals are not fed directly to the oscilloscope plates, but are separately multiplied to 20 Mc/s before connection to the oscilloscope. This speeds up the rotation of the ellipse so that small frequency differences can be more rapidly seen and measured.
Former transmitting and receiving stations outside of the the Tatsfield site
Measurement of Transmitted Frequencies at Tatsfield
The standard frequencies from the Tatsfield oscillators are used for checking the frequencies of BBC and other transmissions with an accuracy of 1 part in 10⁷ or even better. This high accuracy is genuinely necessary for a number of reasons. For example, when two or more medium-wave transmitters (whose carrier frequencies are of the order of 1 Mc/s) have to share the same nominal wavelengths, some listeners will hear unpleasant beats if the carriers differ by as little as one-tenth cycle per second. Experiments carried out in the early days of synchronized working of transmitters (i.e. two or more transmitters on the same frequency) showed that optimum conditions could be maintained provided that individual carrier frequencies did not exceed their nominal value by more than ± 0.1 cycles per second. In other words the difference frequency between any two transmitters on the same frequency should not exceed 0.2 cycles per second. Means are therefore available at each transmitter for checking the frequency of all transmissions daily against a known reference. The error can be measured and if necessary an adjustment made. Once a day is quite sufficient as the crystal oscillators in use in the BBC are very stable and often do not require resetting for some days.
The frequencies of medium-wave, long-wave, television and v.h.f. transmitters can be measured to one-hundredth of a cycle by a most interesting stroboscopic method. A frequency of 1 Mc/s is derived from one of the standards. By dividing or multiplying the reference frequency in decade steps (i.e. with a division or multiplication ratio of ten at each step), and by making use of harmonics, a series of standard frequencies at 100-c/s intervals can be produced, including a frequency which is more than 300 c/s, but not more than 400 c/s below the frequency to be measured. This frequency is selected, and an oscillator is locked into synchronization with it. The oscillator frequency is mixed with the frequency to be measured, producing a difference frequency somewhere between 300 c/s and 400 c/s.
The standard frequency at 100 c/s is made to drive a phonic motor, and this is geared down to drive a shaft at 1 revolution per second. This shaft carries a transparent disc with ten concentric rings of stroboscopic markings on it, in a similar fashion to the stroboscopic speed testers for use on gramophone turntables, except that these are opaque, whereas the clear spaces on this disc are transparent. By selecting the appropriate stroboscopic ring and passing light through it, a beam of light flickering at any rate between 380 c/s and 470 c/s, in 10-c/s steps, can be produced. This flickering light is passed to a photocell and produces an alternating potential with a frequency determined by the particular stroboscopic ring chosen, and with the same accuracy as the standard from which it has been derived. A selected frequency from this photo-tone generator, as it is called, is now mixed with the 300-400 c/s difference frequency already mentioned, to produce a further difference in the region of 70-80 c/s. This remaining difference is matched and measured by a visual stroboscopic process.
To do this it must be fed to a stroboscopic lamp which flashes once for each cycle of the signal incoming to it. Thus an alternating voltage has been produced by photo-electric means, and after mixing this voltage with another alternating voltage, the difference-frequency voltage is being reconverted to a flickering light at a lower frequency. At this final stage, another transparent rotating disc, on the same shaft as the photo-tone generator, is used. On this disc there are eleven concentric rings of figures, with 80 equidistant figure “0s” on the outermost ring, 79 equidistant figure “1s” on the next ring, 78 figure “2s” on the next ring, and so on down to 70 dots representing 10 on the innermost ring. Note that the number of figures on each ring decreases as the figure itself increases. This is because the difference frequency which is now being measured decreases as the frequency of the transmission increases. The disc is placed so that the light from the stroboscopic lamp can be looked at through all the rings of moving figures. Since the shaft is rotating once per second, the figure “3s”,of which there are 77, will sweep past a given point 77 times per second, the figure “2s” 78 times, and so on. If, therefore, the lamp is flashing, say, exactly 77 times per second, the figure 3 will appear stationary, and all the other figures will appear to creep clockwise or anticlockwise. In this particular case, 3 is the units digit of the frequency being measured.
The other digits are already known from the settings of the oscillator and the photo-tone generator. If the frequency being measured is fractionally different from an exact number of cycles, none of the figures will remain quite stationary. The slowest-moving figure will be the one nearest to the exact value, and the exact frequency can be deduced to one decimal of a cycle by counting the number of times this figure passes the viewing window in ten seconds. If the second decimal place is required, the number of transits of the figure during a hundred-second period must be counted. Both the ten-second and the hundred-second periods are indicated by pips derived from the standard frequency.
The visual method of interpolation is the most accurate, but it cannot be used under conditions of severe interference. In such cases, aural interpolation is used, since a normal ear can recognise a minimum heterodyne note through very heavy interference. Other sets of frequency-measuring apparatus at Tatsfield work on this principle, and enable an accuracy of better than one cycle per second to be achieved.
Frequency Measurements at the Transmitting Stations
Although the Tatsfield measurements just described are the final check of the frequencies of transmitting stations, very accurate measurements are also carried out at the stations themselves.
The Droitwich transmission is readily received in any part of the United Kingdom and either this, or a Post Office 1-kc/s reference tone distributed by line to certain stations, can be used to check transmitter frequencies. When the Droitwich carrier is used at medium-wave stations it is passed through dividers which bring the frequency down to 1 kc/s. This is then applied to the horizontal and vertical plates of a cathode ray tube, producing a triangular picture. The time taken for the electron beam to traverse once round the triangle is one-thousandth of a second and corresponds to one cycle of reference frequency. The transmitter frequency to be measured is fed to the cathode ray tube grid so that the electron beam is only allowed to reach the screen of the tube for a short portion of each cycle of the transmitter frequency. The triangular display therefore appears to be made up of a number of dots.
As all BBC medium-wave transmitter frequencies are exact multiples of 1 kc/s, the dots will appear stationary when the frequency being measured is correctly aligned against the reference frequency. If the transmitter frequency is high or low the dots will appear to move in a clockwise or anti-clockwise direction respectively. If a marker line is drawn across the face of the tube then the time taken for one dot to be replaced by the next is a measure of the transmitter frequency error in cycles per second.
If the frequency to be measured is 1457 kc/s then for a stationary picture there must be 1457 dots round the triangle with a 1—kc/s reference frequency. Not all these are displayed and the picture is so arranged that a large number of these are obscured by the brighter spots at the corners of the triangle.
The Droitwich signal is also used for checking the frequencies of short-wave stations. In this case, a 100-kc/s crystal oscillator is kept aligned against the 200-kc/s carrier. The 100-kc/s signal is divided down to give 5 kc/s which is then arranged to provide spectrum frequencies between 3,000 and 26,000 kc/s at a spacing of 5 kc/s. The multiples of 5—kc/s spacing is chosen as all BBC short-wave frequencies are multiples of 5 kc/s.
This frequency spectrum is then mixed with the transmitter frequency to be measured. An audible beat note is obtained between the frequency to be measured and the appropriate 5-kc/s point in the spectrum. The transmitter frequency can then be adjusted until the beat is zero.
Frequency tolerances of BBC transmissions
International Tolerance | BBC Tolerance | |
---|---|---|
Long Wave | ± 10 cycles/second | Guaranteed ± 0.02 c/s |
Generally better than ± 0.05 c/s | ||
Medium Wave | ± 10 cycles/second | Guaranteed ± 0.1 c/s |
Generally better than ± 0.05 c/s | ||
International Common Frequency | ± 20 cycles/second | Guaranteed ± 1 c/s |
Short Wave | ± 30/10⁶ | Guaranteed ± 30/10⁶ |
Generally better than ± 15/10⁶ | ||
Television | ± 30/10⁶ | Guaranteed ± 30/10⁶ |
Generally better than ± 10/10⁶ | ||
VHF | ± 30/10⁶ | Guaranteed ± 30/10⁶ |
Your comment
Enter it below