Digital TV
A Technical FAQ
Rev 4.0.  June 14th 2000

by Chris Muriel
chris.muriel@analog.com

What is Digital Satellite Television
Part 2) Modulation Schemes?
Modulation & Error Coding

The real world is analogue so we have to find a way of transmitting our mpeg2 transport stream as information on a (non-digital) carrier wave. A traditional analogue satellite transmission varies a carrier FREQUENCY in sympathy with the video signal -"frequency modulation" or FM. Similar to this, one can vary the PHASE instead of the frequency - "Phase Modulation". Now we could code our digital signal (which consists of simply binary or "ones & zeros") directly as phase modulation in which case 0 degrees (our carrier reference frequency) could represent a binary 0, whilst 180 degrees phase shift = binary 1; there would be a practical difficulty in keeping track & always changing 180 degrees as there will be natural phase variations over the transmission path. To solve this we can instead make the phase changes cumulative -i.e. make the phase changes refer to the previously signalled state rather than 0/180 degrees absolute.
This is known as DPSK -differential phase shift keying.

Now those of you who know of schemes used for digital transmission, in modems for example,will know that DPSK is somewhat inefficient. There are various schemes that allow the data rate to be doubled, quadrupled (or more) whilst maintaining the original signalling rate.Thus Quaternary (or Quadrature) phase shift keying uses a 2-bit symbol (instead of previously described 1-bit) based on 4 possible phases. At the same time,
0 degrees is avoided to prevent long periods of unmodulated carrier which could cause problems in part of the circuitry - too complex to discuss here. So we use typically 45,135, 225 & 315 degrees. We now have a greater data rate in bits per second than our actual baud (signalling) rate. This can, in fact be further extended by using a constellation of 8 or 16 phases & beyond - although tolerance to noise (required signal to noise ratio for a suitably low bit error rate) & to interference increases as the data rate rises. [This is one reason why computer modems have trouble at high data rates on a poor line & your 56K modem ends up communicating at 31,200 or less]. Other digital transmission media can still use mpeg2 but change from qpsk to a different modulation scheme.Normally this is QAM for cable (although qpsk can be used for the return path back to the operator) whilst digital terrestrial TV uses CODFM -coded orthoganal digital frequency modulation -with either 2048 or 8192 individual carriers each seperately modulated.

The choice of type of modulation is made based on the sort of problems most prevalent for the medium, e.g. terrestrial is more subject to multipath interference ("ghosting" in analogue TV) & CODFM is fairly resistant to this.The same type of modulation is also used for DAB (Digital Audio Broadcast). The main figure of merit for a qpsk demodulator is the minimum Eb/No that the receiver can tolerate to deliver a specified BER (Bit Error Rate) to the MPEG section. Eb/No is the ratio of energy per bit to the noise available at the demodulator. So since satellite signals are inherently noisy,a low order modulation scheme is used with lots of error correction. In fact the DVB adopted what is known as a "concatenated FEC" scheme which means that multiple error correction types are used together - in this case "convolutional" & "block" coding are both used. Viterbi coding is a form of convolutional coding (also used in modems) & the "code rate" refers to the ratio of the number of bits coming out for a given number of bits going in.

So 3/4 means 3 bits come out for every 4 bits going into the decoder. The DVB uses 1/2 code rate for channels with lots of noise (low Eb/No). The error correction comes from the redundant coding data that is transmitted. The constraint length (k) is the number of bits over which the code is computed ; for DVB K=7. The operator decides which code rate to use & the receiver must either scan for the right rate or be told by the user (manual entry). The block coding scheme used is called Reed-Solomon usually abbreviated to R/S, with additional coding by interleaving blocks of bytes. 204/188 code is used which means 188 bytes come out for 204 in - the remainder being parity bytes.

For a good, more technical description of qpsk etc. try the following URL :- http://www.coolstf.com/mpeg

D2MAC :- I am sometimes asked whether D2MAC transmissions (still used by Nordic countries) are digital. In fact the "A" in MAC stands for "Analogue" -
MAC = Multiplexed Analogue Component.
The (analogue) video luminance & chrominance signals are sent at separate times which avoids the interference seen when they are all sent at the same time in PAL. So the 2 signals cannot mix with each other to generate false alias signals which can show up in PAL - like flashes or stripes of colour often seen on a newsreader's (plain grey or checked) suit or tie. Howebver, there IS a digital part to D2MAC - the sound & the video synchronisation information is sent in digital packets. For the audiophiles, note that the audio is only 14 bit so it isn't quite CD-quality. So a D2MAC signal is a "hybrid", part digital & partly analogue.