9. Teletext measurements

Video measurements

The teletext signal is carried by the normal programme video signal and therefore any teletext equipment such as data inserters, data bridges or regenerators which are in the main video signal path must have video performance reaching or exceeding normal broadcast standards so that the video signal is not perceptibly degraded. Furthermore, any crosstalk between the fast logic in the data processing circuits and the video channel itself must be at least 70 dB below the video signal levels to ensure that there is no possible interference in the video channel.

Typical performance figures and requirements for the inserter section of any teletext equipment are as follows:

  1. Video input: 1 volt pk-pk into 75 ohms, with a picture to sync ratio of 70/30.
  2. Return loss: better than 30dB to 6MHz.
  3. Frequency response: +/- 0.05dB to 6MHz.
  4. Luminance-to-chrominance delay: better than /- 5 nanoseconds.
  5. Pulse-to-bar ratio with 2T pulse: +0.25 K.
  6. Pulse shape performance (2T +/- 0.2 T pulse): +/- 0.25 K.
  7. Bar overshoot: less than 1% of bar amplitude.
  8. 50 Hz square wave tilt: less than 1% with respect to 0.7 volts.
  9. Differential gain: 0.5% maximum.
  10. Differential phase: 0.5 degrees maximum.
  11. Non-linear distortions are to be measured with a test signal of three full bar lines and a five-step staircase waveform.

(The interested reader is referred to [20] for the principles of K-factor measurements.)

The inserter must also maintain similar performance when tested with a ‘bump’ waveform consisting of three lines of peak white alternating with three lines of black, followed by the staircase waveform.

The performance of the video clamps have to be such that the conversion noise does not degrade the noise performance below the required figure. Conversion noise is produced by the action of the clamps on signal noise and its effect is a low frequency variation of the line-to-line black level.

A most important requirement for any inserter is that it must not insert its signal onto an incorrect line, due for example, to some distortion in the input signal. This requires the sync separator to perform without error, in the presence of the types of distortions found in complex television networks. For example, on very long lines, very low frequency ‘bounce’ can cause the simple type of sync separator to produce spurious outputs.

Care must also be taken in installation lest reflections or echoes occur due to poor terminations, distorting the digital signal and resulting in poor teletext performance.

Teletext data waveform measurements

For the maximum teletext service area from a transmitter the data must be transmitted to full specification. Normal video tests, however, do not check the quality of the teletext data signal and special measurements are therefore necessary. The position of the data signal on the video waveform, and its amplitude, can be measured by normal oscilloscope techniques using the white ITS bar as reference. If the data amplitude is low, reception of teletext will be impaired. If, on the other hand, the amplitude of the data is higher than specified then teletext reception might be improved but ‘sound buzz’ could be produced in the sound channel on certain receivers. Sound buzz is caused by high frequency video components of the data signal interfering with the inter-carrier sound IF signal. It is therefore essential that the transmitted data amplitude be correct. The data levels and timing are detailed in Figures 2.3 and 2.4.

The teletext data signal contained in the FBI is very difficult to measure accurately using a normal oscilloscope, particularly as the data is not locked to the video signal. To examine the data the oscilloscope time-base must be triggered from the data clock run-in at the start of the data line, but this requires a special trigger circuit sensitive only to the 6.9 MHz clock frequency. Examination of the 2T pulse shape might provide some indication of the likely teletext performance but this cannot be relied upon. In general a poor group delay characteristic causes asymmetry and increased overshoots, as illustrated in Figure 9.1.

Figure 9.1 Transmitter phase response and 2T pulse shape

Some useful information about the data waveform can be obtained from a Lissajous figure displayed on an oscilloscope. The video signal is applied to the vertical deflection amplifier and a sub-multiple of the data clock (usually a quarter of clock frequency) is used for for the horizontal deflection. A ‘data-bright-up’ pulse is also necessary to exclude all television information. Figure 9.2 illustrates the arrangement necessary to produce such a Lissajous display. The resulting display resembles an eye. The difference between the 0 and 1 levels is the ‘eyeheight’ and is normally expressed as a percentage of the true data amplitude. As the signal is degraded, the eyeheight falls. In the limit decoding becomes virtually impossible. Interpretation of the eye display is illustrated in Figure 9.3.

Figure 9.2 Lissajou eyeheight display facility

Figure 9.3 Eye patterns

The internationally agreed definitions [4g] for the properties of the data waveform are as follows:

  1. All-noughts level: the signal level resulting from a continuous stream of 0 pulses.
  2. All-ones level: the signal level resulting from a continuous stream of 1 pulses.
  3. Midlevel: the signal level corresponding to a level midway between the all-0 and the all-1 levels.
  4. Data amplitude: a voltage corresponding to the difference between the all-0 and the all-1 level voltages.
  5. Noughts overshoots: a voltage corresponding to the amount by which the peak value of signal voltage in the direction 1 to 0 exceeds the all 0-level signal voltage.
  6. Ones overshoots: a voltage corresponding to the amount by which the peak value of the signal voltage in the direction 0 to 1 exceeds the all-1 level signal voltage.
  7. Peak-to-peak amplitude: a voltage corresponding to the sum of the data amplitude, the 0 and the 1 overshoots.
  8. Eyeheight: for a noise-free signal the eyeheight represents the smallest difference between any 1 pulse and any 0 pulse for sampling positions equally spaced at the data rate and positions chosen to maximize the quantity. It is expressed as a percentage of the data amplitude.
  9. Decision levels: the level defining the decoding margin in the 1 region is the 1 decision level and the corresponding level in the 0 region is the 0 decision level.
  10. Decoding margin: for a teletext signal referred to the clock run-in timing, the decoding margin represents the greatest difference between the 1 and the 0 decision levels for a given rate at which errors occur. The data samples are spaced at the data rate. It is expressed as a percentage of the normal data amplitude of the signal which should be measured in the absence of noise.
  11. Eyewidth: for a noise-free teletext signal, the eyewidth represents the time interval over which error-free data results when the signal is compared with a given decision level chosen to maximize the quantity. It is expressed as a percentage of the bit period.
  12. Decoding threshold: for equipment generating a graphic display from an applied teletext signal, the decoding threshold represents the smallest value of the decoding margin for a given character-failure rate of teletext signal composed of characters in a specified arrangement and degraded in a specified manner.
  13. Data asymmetry: the amount by which the mean level of the eye pattern at the eyeheight measuring position on the time axis differs from the midlevel expressed as a percentage of the data amplitude.
  14. Noise interference: the level of noise interference in this standard is defined as the RMS level of white wideband noise added to the baseband video signal before modulation, expressed in decibels, relative to a black-to-white level transition.
  15. Co-channel interference: for performance testing of teletext equipment, co-channel interference may be simulated by the addition of a sine wave signal to the baseband video signal. The equivalent level of co-channel interference produced is taken as 11 dB greater than the RMS value of such a sine wave signal.
  16. Proportional jitter: the percentage of the bit period not occupied by the eyewidth.

The Lissajous figure method of measurement suffers a fundamen-tal disadvantage in that the reference data clock must be extracted from the distorted data stream. Any consequent phase jitter of the clock will cause phase jitter of the horizontal scan waveform and reduce the apparent height of the eye opening. This difficulty increases with low values of eyeheight as extraction of a jitter-free clock reference is then more difficult. Typical distorted eye displays are shown in Figure 9.4, and the difficulty of accurate measurement is obvious.

Figure 9.4 Distorted eye displays

The data quality may be estimated by direct inspection, on a normal oscilloscope, of the data stream display of the clock cracker page, Figure 9.5. As can be seen, this test page is composed of two symbols only. The binary codes for these symbols (see code table, Figure 5.1) have the minimum number of transitions. This page therefore tests the decoder clock recovery circuit and also provides the least confused bit stream for visual estimation of eyeheight or decoding margin. Errors on the displayed page of text are, of course, very easily seen.

Figure 9.5 Clock cracker page

Decoding margin

The most meaningful parameter of received teletext data quality is the decoding margin. This measurement takes into account the transmitted eyeheight, the noise and reflections which further degrade the data and jitter in the clock regenerator. Measurements made on an oscilloscope tend to be inaccurate since the parameters must often be estimated from very distorted eye displays, typical examples are shown in Figure 9.4.

To measure the decoding margin it is ncessary to determine the effective separation between the worst 0 and the worst 1 level over many data lines. A special instrument designed for this purpose by the BBC [21] is called a decoding margin meter, Figure 9.6. The instrument provides a digital display of decoding margin expressed as a percentage of the true data amplitude. The instrument is automatic in operation and a comparison technique is used so that the measurement is independent of data information or format.

Figure 9.6 Decoding margin meter

A functional diagram of the instrument is shown in Figure 9.7. The input signal is automatically adjusted to a standard level by an AGC system using the VITS bar for reference so that the data amplitude at the inputs the measuring system is at is a standard value and, therefore, the measurement of the decoding margin is not dependent on amplitude. The clamped data signal is then applied to three slicing circuits operating in parallel. The output signals from the first and the third are compared with the output of the second slicer, which is used for reference purposes and has either a fixed or adaptive slicing level. The slicing levels of the first and third circuits are progressively changed, positively towards the 1 level and negatively towards the 0 level with respect to the reference slicing level. A difference signal generated by the comparitors indicates the level of the worst 0 or 1. The difference in potential between the slicing levels is then used for computing the decoding margin which is displayed as percentage of true data amplitude on a digital display.

Figure 9.7 Functional diagram of an automatic decoding margin meter

The measurement can be averaged over 10 or 1000 data lines and hence takes into account the effects of noise and co-channel interference. A fixed or adaptive slicing level can be used for the reference slicer. Measurement of the adaptive slice level provides an indication of data symmetry (quadrature distortion). Consistent monitoring of the data signal quality is therefore possible and an output is provided for a remote display or for an automatic data logger. Also an alarm signal is generated if the data is lost. To assist in detailed analysis of distorted data using an oscilloscope bright-up pulses are generated, corresponding to the worst 0 and 1 in the bit stream.

Television signals are often distributed by widespread networks containing video links, main transmitters fed by re-broadcast links, and transposers [22]. Although equipment is maintained to video standards such networks can cause degradation of the teletext data signal. This is illustrated in Figure 9.8, which shows the improvements in radiated data quality (decoding margin) which results from data regeneration at the main transmitters (Chapter 6, page 65). For simplicity, the degradation in the decoding margin has been assumed to add arithmetically but in practice this is not the case and measurements must be made.

Figure 9.8

Decoder measurements

Decoders and equipment employing decoding circuits such as data bridges and regenerators must operate reliably even with distorted input signals. To measure and accurately assess the performance of such equipment requires a source of signals with controlled levels of distortion. The design of teletext receivers for domestic use also requires such signals so that the performance of IF amplifiers, tuning systems and decoders can be assessed and compared. A source of teletext signals with various types of controlled distortion is therefore essential for design and product quality control purposes. The functional diagram of a calibrated distortion unit designed for this purpose is shown in Figure 9.9 and a photograph of the instrument is shown in Figure 9.10.

Figure 9.9 Teletext calibrated distortion unit

Referring to Figure 9.10, a teletext test page is generated to full broadcast specification and added to a locally generated composite sync signal. The data can be distorted independently in a number of ways. Firstly, the ‘Delphi’ principle (Defined Eye Loss with Precision Held Indication), a technique developed by the IBA [23], is used to generate reflections which can then be added to the data signal. The reflection amplitude can be controlled between 0 and 100% in fixed or variable steps. Secondly, co-channel interference of two different frequencies can be added to the signal and, thirdly, white noise can be added. Both of these additional distortions can be separately added in controlled amounts.

The video signal itself represents a grating, with the vertical bar waveform corresponding to the shape of an elemental data pulse. The combination of the horizontal and vertical lines therefore produces a ‘pulse-and-bar’ waveform which allows the video characteristics of the receiver to be checked. A complete teletext page can thus be generated, with individually controlled levels of reflections (eyeheight), co-channel interference and white noise. This signal may be fed to a receiver, together with the video pulse-and-bar (grating) waveform. The signal can be fed either at video, direct to the teletext decoder, or, using an RF modulator, to the receiver aerial input. Very searching tests may be made of both decoder and receiver performance, and precise distortion levels at which received errors start to be produced in the page may be determined.

If a teletext page is received with errors on the first acquisition, the errors are normally corrected when the page is received a second time. However, to assess the performance of a decoder or receiver it is necessary to count the errors that occur on the first acquisition. The ‘update’ bit is therefore set in the header row of the transmitted page so that the memory is erased just before each new reception of the page.
The page repetition rate may be set to 1, 10 or 100 pages per second. The latter rate enables virtually a continuous page to be displayed, which shows up errors on a continuous basis. This is a useful technique when making adjustments to decoder clock circuits or receiver alignment. Alternatively, the slower rates give a longer time elapse between each page transmission which allows the particular errors on a page to be examined, or the mechanism which produces the errors to be further investigated, before the page is retransmitted.
The accepted design target for teletext receivers is for operation with input signals having decoding margins down to 25%, with two or three errors occurring when a page is first acquired. Allowing for degradation in the tuner and IF circuits, it follows that the decoder needs to operate with input signals having a decoding margin of about 20%.

When testing receiver performance in development or production it is also necessary to generate a magazine of special pages so that all the various aspects of a receiver’s decoder and remote control system can be checked thoroughly and evaluated in a non-ambiguous manner. A teletext test page generator for this purpose must therefore meet the full broadcast specification and have appropriate pages programmed to meet the full broadcast requirements of the particular countries in which the receivers will eventually be used. It is particularly important to ensure that the transmission of ‘multi-level’ pages, that is, pages which contain additional data packets for language requirements, are configured to exactly the same specification as the final broadcasting system on which the receiver will be used.

Decoders for data reception

A teletext system can provide a data distribution service to specific groups of users, called ‘closed user groups’ [24]. The decoder for such applications might provide an RS232 output for driving a personal computer (PC), or the decoder might be contained in the PC itself. With a normal teletext transmission the magazine of pages is continuously repeated. If an error occurs in the reception of a page it is normally corrected when the page is next received. When a teletext system is used for data transmission this error correction arrangement cannot normally be used. A broadcaster will have several different closed user groups and furthermore the data will be continuously changing. It will therefore be transmitted only once or twice and it may well contain eight bit codes consisting of all 0 or 1 levels. The performance requirements of the decoder are therefore more onerous and a typical criterion would be that at least 100 megabytes of data must be received without error.

A decoder for data reception must first be checked as a normal teletext decoder using a calibrated distortion unit followed by checks for possible error rates. The error rate for data reception can be checked by using a teletext test generator producing a repeating data stream containing the full range of codes to be used, together with a cyclic redundancy check word (CRC). The RS232 output from the decoder is fed to a PC which contains special monitoring software. All errors in the received data and the continuity index are counted using the CRC check to confirm correct data reception. Details of all errors are displayed.

In data transmission systems for closed user groups the data is encrypted by the originator so that only receivers that are provided with the appropriate software can make use of the data, the teletext system itself being transparent. An alternative method of 'conditional access' can be used on an individual decoder basis. In such a system an individual user code has to be entered into the decoder to initiate reception. To allow a decoder to be tested without a particular 'user-code', an 'open' code can be used. If the decoder functions correctly with this code, no individual user-code tests are necessary. The test programme must be designed as part of the data management system.

The decoder forms only one component of a complete data transmission system. The decoder test programme or the test generator can also be used at the data source to check the transmission system, including the distribution network and transmitters, for potential error-producing faults.

Next