Administrative Ruling no. 316/93, of 18 of March

Published in D.R. number 65 (Series I-B) of 18 March 1993

Ministério das Obras Públicas, Transportes e Comunicações (Ministry for Public Works, Transport and Communications)

Administrative Ruling

The existence of various standards for the transmission of digital stereophonic sound for television makes it necessary to adopt a single national standard that is compatible with all television systems adopted in Portugal by Administrative Rules no. 936/81, of October 28th, and no. 1155/91, of November 7th, which established the PAL (Phase Alternation Line) and D2 MAC (Multiplexed Analogue Component) systems, respectively. 

ANNEX
NICAM (Near Instantaneous Companded Audio Multiplex) Standard

1- This standard defines the technical characteristics of the NICAM system which allows for the transmission of two-channel digital sound and/or data services with terrestrial television systems B, G, H and I.

2- Specification of the sound/data multiplex and sound coding methods:

2.1 Baseband format:

2.1.1 Frame structure:

The transmitted serial data stream is partitioned into 728-bit frames which are transmitted continuously without gaps. One frame is transmitted every millisecond; the overall bit-rate is thus 728 kbit/s made up as follows:

8-bit frame alignment word 8 kbit/s (see section 2.2.1 )
5 bits for control information 5 kbit/s (see section 2.2.2)
11 bits for additional data 11 kbit/s (see section 2.2.3)
704 sound, parity or data bits 704 kbit/s (see section s 2.2.4 and 2.2.5)

Total:
728 kbit/s

Diagrams of the frame structures for conveying stereo and mono sound signals are shown in figure 1. The 720 bits which follow the Frame Alignment Word (FAW) shall form a structure identical with that of the first-level protected, companded sound-signal blocks in the systems of the MAC/packet family, so that decoding of the sound signals may be performed by the same type of decoder which is used in the MAC systems.

The first 16 bits after the frame alignment word shall be used to signal control information (see section 2.2.2) and as additional data bits (see section 2.2.3). The corresponding 16 bits in the MAC/packet family have not yet been allocated.

Frame structures for data services use the same frame alignment word, frame flag bit and additional data, with other control bits as described in section 2.2.2.2, but the audio samples are replaced by other data.

2.1.2 - Bit interleaving

Interleaving is applied to the block of 704 bits which follows the frame alignment word, control bits and additional data bits in order to minimise the effect of multiple-bit errors. The bits of each frame are transmitted in the following order:

2.1 .3 - Energy dispersal scrambling:

The transmitted bit-stream shall be scrambled for spectrum-shaping purposes. The scrambling shall be done synchronously with the multiplex frame. The frame alignment word is not scrambled, and is used to synchronise the pseudo-random sequence generator used for descrambling in the receiver. The other parameters shall be as follows:

i. the bit which immediately follows the frame alignment word is the first coded bit and is added modulo-two to the first bit of the pseudo-random sequence;

ii. the bit which immediately precedes the frame alignment word is the last scrambled bit;

iii. scrambling shall take place after interleaving (and descrambling shall, therefore, be performed prior to de-interleaving at the receiver);

iv. the pseudo-random sequence is defined by the following generator polynomial and initialisation word:

Generator polynomial: x9 + x4 + 1

Initialisation word: 1 1 1 1 1 1 1 1 1

The diagram for a possible generator for this sequence is shown in figure 2. Thus, the sequence shall start 0000 0111 1011 1110 0010.

2.2 Coding of information

2.2.1 - Frame alignment word:

The frame alignment word shall be 01001110; the left-most bit shall be transmitted first.

2.2.2 - Control information:

The control information shall be conveyed by the frame flag bit, C0, three application control bits C1 , C2, and C3, and the reserve sound switching flag C4, (see figure 1 ).

2.2.2.1 - The frame flag bit:

The frame flag bit C0, shall be set to 1 for 8 successive frames and to 0 for the next 8 frames thus, it defines a 16-frame sequence (¹). The frames are numbered within the sequence as follows: the first frame (frame no. 1) of the sequence is defined as the first of the 8 frames in which C0 = 1 ; hence, the last frame (frame no. 16) of the sequence is the last of the 8 frames in which C0 = 0. This frame sequence is used to synchronise changes in the type of information being conveyed in the channel.

2.2.2.2 - The application control bits:

The last 704 bits in each frame may be used to convey either sound samples or data. The current application of these bits shall be defined by the three application control bits C1, C2, and C3, as indicated in table 1 below.

When a change to a new application is required, these control bits shall change to define the new application on frame No. 1 of the last 16-frame sequence of the current application. The 704-bit sound/data blocks shall change to the new application on frame No. 1 of the following 16-frame sequence.

2.2.2.3 - The reserve sound switching flag:

Digital sound decoding equipment may be arranged so that it can switch the output of the conventional FM sound demodulator to replace the sound decoded from the digital signal in the event of the failure of the latter. Switching to the output of the FM demodulator shall be acceptable only if the FM carrier is modulated with the same sound programme as the failing digital signal: the reserve sound switching flag provides the means to inhibit such switching, and it shall be incorporated as the fifth bit, C4, of the control information.

The reserve sound switching flag shall be set to 1 when the FM signal is carrying the same sound programme as the digital stereo signal, or the digital mono signal. In the case where two digital mono signals are being transmitted, this refers to the M1 signal only. When the FM signal is not carrying the same programme as the digital sound signal, the reserve sound switching flag shall be set to 0. In this state it can be used to prevent switching to the FM sound.

C4 is not defined in the present document for transparent data channels or for any future data option.
 

(*) The use of the reserve sound switching flag (bit C4)is only specified in the case when the digital signal is carrying sound. It has no meaning in the case of data transmission, when th only sound available id the FM sound signal.
(**) This mode may be used to broadcast three independent sound signals.

2.2.3 Additional data: Eleven additional data bits AD0 to AD10 (see figure 1) are reserved for future applications yet to be defined.

2.2.4 The sound/data block: The last 704 bits in any frame form a block of either sound or data information. When C3 = 0, the two types of information shall not be mixed within one frame. When digital sound signals are being carried, 64 sound samples (D0 to D64) are transmitted within each frame. Figure 1 a) shows the structure of a stereo sound frame, and figure 1 b) shows the mono sound frame. If a stereo pair of sound signals is being transmitted (C1 = C2 = C3 = O), the odd-numbered samples (D1, D3, ...., D63) shall be used to convey the A-channel, and the even-numbered samples (D2, D4, ...,D64) the B-channel (see section 2.2.5.1). Thus 32 samples of each channel shall be transmitted in every frame, corresponding to one complete companding block of each sound channel.

If two independent mono sound signals, M1 and M2, are being transmitted (C1 = 0, C2 = 1, C3 = 0), M1 shall be transmitted in frame Nos. 1, 3, 5,..... (i.e. odd-numbered frames), and M2 in frame nos. 2, 4, 6,..... ( i.e. even-numbered frames). Frame numbering is defined by the 16-frame sequence of the frame flag bit, C0, see section 2.2.2.1 .

lf one mono sound signal, M1, is being transmitted (C1 = 1 , C2 = 0, C3 = 0), it shall be transmitted in the odd-numbered frames and data shall be transmitted in the even-numbered frames.

Thus, for mono sound signals, each frame with sound information in it shall contain 64 consecutive sound samples, which will span 2 complete companding blocks, shown as blocks n and (n + 1 ) in figure 1 b). No format has yet been defined for data signals.

2.2.5 Sound signals

2.2.5.1 Digitisation and near-instantaneous compounding

Sound signals shall be sampled at 32 kHz and coded initially with a resolution of 14 bits per sample. For transmission, the number of bits per sample is reduced to 10, using near-instantaneous companding, and one parity bit shall be added to each 10-bit sample word for error detection and scale-factor signalling purposes.

The near-instantaneous compression process forms the 14-bit digital samples corresponding to each of the sound signals into blocks of 32. Thus, each companding block contains the samples for 1 ms of one sound channel. All of the samples in each companding block shall then be coded, using a 10-bit 2's complement code, to an accuracy determined by the magnitude of the largest sample in the block, and a 3-bit scale-factor code shall be formed to convey the decree of compression to the receiver. Figure 3 illustrates the coding of companded sound signals.

Pre-emphasis to CCITT Recommendation J.17 shall be applied to the sound signals prior to compression. This may be done using analogue pre-emphasis networks prior to digitisation. Alternatively, this pre-emphasis could be applied using digital signal processing that achieves the equivalent result. The phase characteristic of the pre-emphasis is not specified.

For stereophonic transmission the left and right signals shall be sampled simultaneously; the A samples convey the sound signal to be reproduced by the left-hand loudspeaker and the B samples convey the sound signal to be reproduced by the right-hand loudspeaker.

TABLE 2
Summary of sound coding characteristics

Sampling frequency: 32 kHz
Initial resolution: 14 bits/sample
Companding characteristics: Near-instantaneous, with compression to 10 bits/sample in 32-sample (l ms) blocks
Coding for compressed sampIes: 2's complement (see figure 3)
Pre-emphasis: CCITT Recommendation J.17 , gain set to drive reference level defined below

Nominal reference levels:

For a sine wave test-signal at a frequency of 400 Hz, the sound signal alignment level (see Note 1), 0 dBu0s (see Note 2), shall be 22 dB below the maximum of the digital coding range in television systems B, G and H and 24,3 dB below the maximum of the digital coding range in television system I.

(For a sine wave test-signal at a frequency of 2 kHz, the alignment level, 0 dBu0s, is 12,5 dB below the maximum of the digital coding range in television systems B G and H and 14,8 dB below the maximum of the digital coding range in television system I);

With a mono sine wave test-signal at a frequency of 400 Hz, applied at the alignment level, 0 dBu0s, to the modulation input of the FM sound transmitter, the resulting nominal deviation of the FM sound carrier shall be ± 13 kHz in television systems B, G and H and 17 kHz in television system I .

The relationship between the levels of the digital stereo and the FM mono sound signals, for the case where the alignment levels at the input to the digital sound coder and the input to the FM sound modulator are both 0 dBu (i.e. an absolute voltage level of 0,775 V), is as follows: when the digital channels are used to convey a stereo signal, the compatible mono signal is usually derived from the sum of the A and B sound signals attenuated by 6 dB. However, at least one operator uses 3 dB (²) attenuation.

Note 1 - The CCIR Recommendation 645 (vol. x-1 and vol. XII) "Test signals to be used in international sound programme connections" refers to "the alignment level" 0 dBu0s for those indications given for various types of measurer. For example, a sinusoidal wave at the alignment level O dBu0s shows "Test" on a EBU FFM.

Note 2 - The CCIR Recommendation 574-2 (vol. XIII) "Use of the decibel and the neper in telecommunications" defines dBu0s as the "absolute voltage level for 0.775 V, referring to a relative point at the level of 0 in sound programme transmissions."

2.2.5.2 Error protection for sound signals:

One parity bit shall be added to each 10-bit sound sample to check the six most significant bits for the presence of errors. The parity group, thus, formed is even (i.e. the modulo-two sum of the six protected sample bits and the parity bit is zero). Subsequently, the parity bits are modified to signal the 3-bit scale-factor word associated with each sound signal companding block (see section 2.2.5.3). Table 3 shows the coding ranges and protection ranges associated with each 3-bit scale-factor word (R2, R1, R0). The five coding ranges indicate the degree of compression to which each block of samples has been subjected for the near-instantaneous companding process.

In addition to signalling the coding range, the scale factor indicates seven protection ranges. The protection range information signalled by the scale factor may be used in the receiver to provide additional error protection for the most significant bits of the samples.

2.2.5.3 Scale-factor signalling-in-parity for sound signals The three-bit scale-factor R2, R1 , R0 (see table 3), associated with each sound signal companding block, is carried by modification of the parity bits in the samples used to convey that sound signal. When a stereo sound signal is being sent, let FE1( specification ">⁴) be the scale-factor word R2A, R1A, R0A associated with the 32 A samples, and FE2 the scale-factor word R2B, R1B, R0B associated with the 32 B samples. If Pi is the parity bit of the ith sample, then this is modified to P'i , by modulo-two addition of one bit of one of the scale-factor words according to the following relationship:

P'i = Pi R2A for i = 1, 7, 13, 19, 25, 31, 37, 43, 49
P'i = Pi R1A for i = 3, 9, 15, 21, 27, 33, 39, 45, 51
P'i = Pi R0A for i = 5, 11, 17, 23, 29, 35, 41, 47, 53

P'i = Pi R2B for i = 2, 8, 14, 20, 26, 32, 38, 44, 50
P'i = Pi R1B for i = 4, 10, 16, 22, 28, 34, 40, 46, 52
P'i = Pi R0B for i = 6, 12, 18, 24, 30, 36, 42, 48, 54

When a mono sound signal is being sent, FE1 is the scale-factor word R2n , R1n , R0n associated with the first block of 32 samples in the frame, and FE2 is the scale-factor word R2n + 1 , R1n + 1 , R0n + 1 associated with the second block of 32 samples in the frame.

As in the case of stereo sound, the parity bit of the ith sample (P i) is modified (to P i ) by modulo-two addition of one bit of one of the scale-factor words. However, the modification of the parity bits in the mono case relates to the frame structure of the mono signal, as follows:

P'i = Pi R2n for i = 1 , 4, 7, 10, 13, 16, 19, 22, 25
P'i = Pi R1n for i = 2, 5, 8, 11 , 14, 17, 20, 23, 26
P'i = Pi R0n for i = 3, 6, 9, 12, 15, 18, 2 1 , 24, 27

P'i = Pi R2n+1 for i = 28, 3 1 , 34, 37, 40, 43, 46, 49, 52
P'i = Pi R1n+1 for i = 29, 32, 35, 38, 41 , 44, 47, 50, 53
P'i = Pi R0n+1 for i = 30, 33, 36, 39, 42, 45, 48, 51, 54

Note. - Some scale factor information for the second block of samples is transported in the coding of the parity of samples 28 to 32 which are in the first block. This is in accordance with the coding of the specification < 1 > EBU.

Scale-factor, coding range and protection range information are extracted at the decoder by majority-decision logic. Subsequently the original parity is restored for the purposes of error detection and concealment.
The control information as described in Section 6.2.3 of < 1 > is not used. However, the corresponding parity relation may be either odd or even.

3- Specification of the modulation parameters:

3.1 - Analogue signals:

3.2 - Digital signal:

3.2.6 - Transmitted spectrum and differential group delay time for systems B, G and H:

For systems B, G and H, the spectrum of the transmitted digital sound signal should be nominally within 2 dB relative to the ideal response for frequencies in the range 5,85 MHz 250 kHz relative to the vision carrier.
The differential group delay time for frequencies in the range 5,85 MHz 250 kHz should be nominally within 100 ns.

3.3 - Specification of the digitally-modulated carrier
3.3.1 - Type of modulation:
The modulation system is differentially-encoded QPSK ( ), i.e. four-state phase modulation in which each change of carrier phase state conveys two data bits.
3.3.2 - Differential encoding:
The input data stream at the modulator is differentially encoded by the following processes (see fig. 6):

i. Serial to two-bit parallel conversion:
The input data stream is formed into bit-pairs by a serial to two-bit parallel converter.

ii. Coding of transmitted phase changes:

The amounts of the changes of carrier phase which correspond to the four possible values of the input bit-pairs (An, Bn) are:

tabela

Where, as indicated in figure 6, An is the input bit at some arbitrary time, and Bn is the input bit one bit-rate clock period later.

Thus, the carrier phase can dwell in one of four rest-states, spaced 90 apart, as illustrated in figure 7 a). An input bit-pair shall shift the carrier phase into a different rest state by the amount of phase-change assigned to that particular value of bit-pair. The transmitted phase changes and subsequent carrier rest-states for the input bit-pair sequence 00, 10, 11, and 01 are illustrated in figure 7 b).

In the receiver the transmitted data-stream shall be unambiguously recovered by determining the phase-changes between one bit-pair and the next.

4- References:

< 1 > European Broadcasting Union 1986, "Specification of the systems of the MAC/packet family", tec. doc. 3258-L.

Ministry for Public Works, Transport and Communications.

Signed on January 22, 1993.

By the Minister for Public Works, Transport and Communications, Carlos Alberto Pereira da Silva Costa, Secretary of State for Housing.

 
¹ Repetitive false detection of the frame alignment word within the 704 bit sound/data block can be avoided by including of the frame flag bit (C0) in the frame alignment word decoding strategy.
² Receiver manufacturers should assume an attenuation of 6 dB
³ It would be possible to add another range of protection, however, the final code of faactor scale-shows "7th range of protection" (not the 8th) in order to maintain un maximum number of common elements with the MAC / EBU package
⁴ The initials "FE" (facteur d' échaile) were used in accordance with the EBU < 1 > specification
⁵ In the case of cable distribution for adjacent channel operation is recommended to use a feature of 0.75 MHz for PAL I VSB and placement of the attenuation of the VSB filter, the system B, -20 dB at frequency of 0, 95 MHz below the picture carrier
⁶ Four-phase phase shift keying modulation (DPSK), encoded differentially.