EP0797323A1 - Transmitter for digital radio broadcast signals - Google Patents

Abstract

The transmitter includes a coder which passes in-phase and quadrature signals to a limiting circuit (T). The signals are passed through an interpolation circuit (M1,M2) which reduces the coding bandwidth and reduces output amplifier distortion. The signals then pass through anti-aliasing filters (Fb1,Fb2), D/A convertors and filters and are applied to a QPSK modulator. The signal is then applied to a pre-amplifier (PO) and two further amplifiers (P1,P2) which are connected in parallel. The signal is transmitted after summing and filtering.

Description

The present invention relates to the emission of digital radio signals more commonly called DAB signals according to the acronym of their Anglo-Saxon designation: Digital Audio Broadcast; it relates more particularly to the power amplification of such signals in the output stages of the transmitters.

When amplifying a DAB signal by a module whose transfer characteristic must be used in its linear part, it must be ensured that the peak voltages of the DAB signal applied to this module are processed in the linear part of the characteristic. transfer. With a power amplifier, since the DAB signal applied to it generally has a difference of 11 dB between its average and peak powers, this means that, in order to amplify the largest peaks in the linear part of the characteristic of the amplifier, the average output power must not exceed the maximum power of the amplifier minus 11 dB; for example, a 100 W amplifier will deliver an amplified DAB signal, the average power of which should not exceed 100 W - 11 dB, i.e. around 8 W, in order to be able to amplify power peaks linearly. 100 W.

Knowing that power amplifiers are expensive and often represent more than half the price of a DAB power transmitter, it appears that the use of an amplifier to obtain an average power at most equal to 8% of its possibilities , constitutes a serious handicap on the financial level.

The object of the present invention is to reduce this drawback.

This is obtained by processing the DAB signal, before power amplification, intended to reduce the difference between its peak power and its average power.

According to the invention there is proposed a transmitter of digital radio signals, comprising, coupled in a serial arrangement, coding means for receiving a signal to be transmitted and providing in exchange a pair of quadrature signals representing, in Cartesian coordinates, a complex number of inputs, a modulator with four phase states, amplification means, characterized in that that it comprises, inserted in series between the coding means and the amplification means, clipping means followed by low-pass filtering means, the clipping means performing clipping at a predetermined value of the module of the number entrance complex.

• la figure 1, un émetteur selon l'art connu,
• la figure 2, un émetteur selon l'invention
• les figures 3 à 5, des oscillogrammes de signaux rencontrés lors de l'étude qui a conduit à passer de l'émetteur selon la figure 1 à l'émetteur selon la figure 2.

The present invention will be better understood and other characteristics will appear from the following description and the figures relating thereto which represent

• FIG. 1, a transmitter according to the known art,
• Figure 2, a transmitter according to the invention
• FIGS. 3 to 5, oscillograms of signals encountered during the study which led to passing from the transmitter according to FIG. 1 to the transmitter according to FIG. 2.

In Figures 1 and 2 the corresponding elements are designated by the same references.

FIG. 1 represents a DAB signal transmitter according to the known art in the case of an embodiment without passing through an intermediate frequency, the baseband signal directly modulating the transmission frequency.

, respectivement dans deux filtres F1, F2. Les deux signaux en quadrature issus des filtres F1, F2 sont appliqués sur un modulateur à quatre états de phase, 3, dit modulateur QPSK dans ce qui suit, d'après le sigle des mots anglo-saxons Quadrature Phase Shift Keying qui signifient modulation à quatre états de phase. Dans le modulateur 3 les signaux en quadrature modulent un signal porteur fourni par un oscillateur non représenté ; la fréquence de ce signal porteur est de 1472 MHz dans l'exemple décrit.The transmitter according to FIG. 1 receives a useful audio signal on an input E of a DAB coder, 1; the encoder 1 delivers, on two outputs, respectively two digital signals in baseband, in quadrature, I and Q, sampled at 2.048 MHz in the example described. These two digital signals are converted to analog, respectively in two converters 21, 22, then filtered in $\frac{\text{x}}{\text{sin x}}$

, respectively in two filters F1, F2. The two quadrature signals from the filters F1, F2 are applied to a modulator with four phase states, 3, called the QPSK modulator in what follows, according to the acronym of the English words Quadrature Phase Shift Keying which mean modulation at four phase states. In the modulator 3 the quadrature signals modulate a carrier signal supplied by an oscillator not shown; the frequency of this carrier signal is 1472 MHz in the example described.

The output signal of the modulator 3 is amplified in a preamplifier PO followed by five amplifiers P1 to P5 mounted in parallel. An adder circuit, Ad, sums the output signals from amplifiers P1 to P5; its output is coupled through a bandpass filter Fs to a transmitting antenna, S.

In what follows we talk about bit error rate and crest factor; the meaning of these terms is recalled below.

The bit or BER error rate of a transmitted signal is the ratio of the number of false bits of the signal to the total number of bits of the signal.

The crest factor of a signal is equal to the ratio of its peak power to its average power; it is generally expressed in decibels.

In the case of FIG. 1, a DAB signal having, for example, a crest factor of 11 dB requires a power drop, back off in the Anglo-Saxon literature, of 11 dB during amplification in order to limit the distortions non-linear and keep the nature of the DAB signal. This means that by using, for example, an amplifier with a power of 100 W defined at 1 dB of compression, the average power at the output cannot exceed 100 W - 11 dB or approximately 8 W; the power drop is 92 W.

By clipping the DAB signal at 6 dB relative to a peak power considered at 11 dB above the average power, it has been observed, in the example which will be described below, that the crest factor of the signal considered passes from 11 dB to 7 dB; with such clipping, the bit error rate or TEB undergoes a slight deterioration but, despite this degradation, its value remains within the standards.

This clipped DAB signal requires a power drop of 7 dB during amplification in order to limit non-linear distortions and to preserve the nature of the DAB signal. This means that using, for example, an amplifier with a power of 100 W defined at 1 dB of compression, the average power at the output can reach up to 100 W - 7 dB, or approximately 20 W.

The gain of 11-7 = 4 dB on the power recoil allows, compared to the previous case, to use 2.5 times less amplifiers (4 dB≃10 log 2.5) to obtain an amplified signal similarly power and comprising intermodulation products out of bands, intermodulation products or shoulders in Anglo-Saxon literature, equivalent; this is true provided, in order not to have more significant intermodulation products, to take certain specific precautions upstream of the amplifiers; these precautions will be indicated later.

The clipping mentioned above, for the DAB signal, can be performed anywhere, between the encoder and the output amplifiers, in the DAB signal transmitter: just at the output of encoder 1 or at output digital-analog converters 21, 22 or at the output of the modulator 3 or even at the output of the mixer in the case where the signal to be transmitted is modulated in intermediate frequency before being transposed, by mixing, in transmission frequency; depending on whether this clipping will be done upstream or downstream of the modulator, it will therefore relate to a DAB signal made of two quadrature signals or of a single signal. It is the clipping at the output of the encoder 1 which was chosen in the embodiment according to FIG. 2 for its relative ease of implementation and the results which it gives.

• calcul du module : $$\text{ρ =}\sqrt{{\text{I}}^{\text{2}}{\text{+ Q}}^{\text{2}}}$$
  
• calcul du module, ρ', après écrêtage à une valeur d'écrêtage ρ_e : $${\text{ρ'=ρ}}_{\text{e}}{\text{\hspace{1em}\hspace{1em}\hspace{1em}si ρ ≥ ρ}}_{\text{e}}$$
  
  $${\text{ρ'=ρ \hspace{1em}\hspace{1em}\hspace{1em}si ρ < ρ}}_{\text{e}}$$
  
  il est à noter que cet écrêtage à la valeur ρ_e représenté en décibels par 20 log (ρ_m/ρ_e), où ρ_m est la valeur maximum prise par le module de (I, Q)
• calcul du signal complexe écrêté (I', Q'), en coordonnées cartésiennes, en lui conservant l'argument du signal complexe d'origine (I, Q) : $$\mathit{\text{I}}\text{'=ρ'.sin[}\mathit{\text{Arc tg}}\text{(}\mathit{\text{I}}\text{/}\mathit{\text{Q}}\text{)]}$$
  
  $$\mathit{\text{Q}}\text{'=ρ'.cos[}\mathit{\text{Arc tg}}\text{(}\mathit{\text{I}}\text{/ Q}\mathit{\text{)}}\text{]}$$
  

This involves clipping on the DAB signal made of a pair of quadrature signals, I and Q, representing the Cartesian coordinates of a complex number (l, Q). For this a clipping circuit which is, in fact, made up of calculation means performs the following operations

• module calculation: $$\text{ρ =}\sqrt{{\text{I}}^{\text{2}}{\text{+ Q}}^{\text{2}}}$$
  
• calculation of the modulus, ρ ', after clipping at a clipping value ρ _e : $${\text{ρ '= ρ}}_{\text{e}}{\text{if ρ ≥ ρ}}_{\text{e}}$$
  
  $${\text{ρ '= ρ if ρ <ρ}}_{\text{e}}$$
  
  it should be noted that this clipping at the value ρ _e represented in decibels by 20 log (ρ _m / ρ _e ), where ρ _m is the maximum value taken by the module of (I, Q)
• calculation of the clipped complex signal (I ', Q'), in Cartesian coordinates, keeping the argument of the original complex signal (I, Q): $$\mathit{\text{I}}\text{'= ρ'.sin [}\mathit{\text{Arc tg}}\text{((}\mathit{\text{I}}\text{/}\mathit{\text{Q}}\text{)]}$$
  
  $$\mathit{\text{Q}}\text{'= ρ'.cos [}\mathit{\text{Arc tg}}\text{((}\mathit{\text{I}}\text{/ Q}\mathit{\text{)}}\text{]}$$
  

FIG. 2 is a diagram of a transmitter according to the invention which is distinguished from the transmitter according to FIG. 1 only by the introduction, between the encoder 1 and the digital-analog converters 21, 22, of additional circuits, and by the absence of the amplifiers P3 to P5, that is to say three out of five of the amplifiers.

The signals 1 and Q delivered by the encoder 1 are not, in reality, directly clipped but are respectively applied to the inputs of two interpolation circuits M1, M2. A clipping circuit, T, has two inputs respectively connected to the outputs of the two interpolation circuits and two outputs respectively coupled to the inputs of the converters 21, 22 through low-pass filters Fb1, Fb2; thus the converters 21, 22 respectively receive signals l ′ and Q ′ which come to it from the clipping circuit only after having been filtered.

The role of the interpolation circuits M1, M2 is to increase the useful bandwidth of the signal compared to the signal coming from the coder in order to correct the phenomenon of aliasing, aliasing in the Anglo-Saxon literature, caused by the circuit d 'clipping.

The role of the clipping circuit T is to carry out the three operations which have been discussed above in order to reduce to a predetermined value, ρ _e , the modules of those of the complex signals DAB which have a value greater than ρ _e , but without changing the argument of these signals.

The low-pass filters F1 and F2 are, in the example described, filters with 64 coefficients coded on 12 bits. Their role is to eliminate the intermodulation products caused by the clipping circuit.

Figures 3, 4 and 5 are oscillograms obtained at the output of modulator 3, that is to say after modulation of the carrier signal at 1472 MHz; the scale of these oscillograms is 300 kHz per square in horizontal and 10 dB per square in vertical; the frequency of 1472 MHz was placed in the middle of the oscillogram.

The oscillogram in FIG. 3 shows the spectrum of the signal at the output of the modulator 3 in a circuit configuration according to FIG. 1, that is to say without clipping; as it appears on figure 3 the products of intermodulation are with -56dB. Furthermore, the measurement of the crest factor of this modulator output signal gives a value of 11dB. As for the measurements made by placing a receiver in series with the transmitter, they show that a bit error rate of 10 ^-4 leads to an increase in the carrier-to-noise ratio, C / N, of 0.1 dB compared to the C / N ratio that would be obtained with a perfect transmitter and receiver.

It should be noted that this way of presenting the C / N ratio as a function of the bit error rate, and not the opposite which would seem more logical, is due to the fact that it is a value of bit error rate which is sought. .

The oscillogram of FIG. 4 shows the spectrum of the signal at the output of the modulator 3 in a circuit configuration which is distinguished from that of FIG. 2 only by the absence of the low-pass filters Fb1, Fb2, replaced by short circuits; in this configuration and for a clipping at 6 dB of the peak value, the intermodulation products are at -26 dB.

The low-pass filtering performed with the filters Fb1, Fb2 according to FIG. 2 makes it possible to lower these intermodulation products to -55 dB, as it appears in FIG. 5. The measurement of the crest factor of this signal of the modulator 3 , in the configuration according to FIG. 2, gives a value of 7 dB; and the bit error rate, always 10 ^-4 , is obtained with a C / N ratio which drops to 0.4 dB, ie a degradation of 0.3 dB with respect to the transmitter according to FIG. 1.

This value of 0.4 dB is the maximum value currently authorized, according to standards; it allows to go from a crest factor of 11 dB to a crest factor of 7 dB and thus, as explained above, allows to reduce the number of output amplifiers from five to two for average power unchanged transmitter output. The standard which fixes the maximum degradation to carrier-to-noise ratio, C / N at 0.4 dB is not final and the maximum degradation value will probably be increased to 0.8 dB in the future, more clipping severe will then be possible, which will lead to a reduction in the crest factor greater than that of the example described using FIGS. 2 and 5.

It has been stated in the foregoing that clipping can be performed anywhere between the encoder and the output amplifiers; in the event that clipping is performed on the DAB signal while it is in analog, the interpolation circuits have no longer any reason to be; of even if the sampling in the encoder 1 is at a frequency sufficient to avoid that the aliasing of the spectrum after clipping is annoying, the interpolation circuits are no longer necessary.

Claims (3)

Transmitter of digital radio signals, comprising, coupled in a serial circuit, coding means (1) for receiving a signal to be transmitted and supplying a pair of quadrature signals, this pair constituting the Cartesian coordinates of a complex number of input representative of the signal to be transmitted, a modulator (3) with four phase states, amplification means (P0-P5), characterized in that it comprises, inserted in series in the circuit, between the coding means ( 1) and the amplification means (P0-P5), clipping means (T) followed by low-pass filtering means (Fb1, Fb2), the clipping means performing clipping at a predetermined value of the module of the complex entry number. Transmitter according to claim 1, characterized in that the pair of quadrature signals supplied by the coding means (1) being in digital value, the clipping means (T) are connected to the coding means to receive this pair, calculate the module and the argument of the complex input number and deliver a pair of quadrature output signals constituting the Cartesian coordinates of a complex number with the same argument as the complex number of input and module equal to that of the complex number input clipped to a predetermined value. Transmitter according to claim 2, characterized in that the coding means (1) comprise, at the output, interpolation means (M1, M2).

Images