BACKGROUND OF THE INVENTION
The invention relates to a method for linearizing, over a wide frequency band, a transmission chain comprising a power amplifier.
The constant increase of traffic in the field of telecommunications raises new problems which are not easy to solve for the equipments. More particularly the transmission equipments must operate over an increasing frequency bandwidth in order to be able to transmit simultaneously a great number of channels which operate at different frequencies and/or with different codes. In fact in telecommunication systems each transmitter may operate either on a frequency division multiple access (FDMA) mode or on a coded division multiple access (CDMA) mode or on a combination of both. In the case of CDMA the modulation is spread over a frequency band of several MHz. It is recalled here that in CDMA to each symbol, such as a bit, is superimposed a code which is a sequence at a higher frequency.
The consequence of the increase of the number of channels to be transmitted is also an increase of the transmission power of the transmission equipments.
The most difficult problem to solve with such increase of power together with the increase of the frequency bandwidth is the linearization of the transmission chains, more particularly the linearization of power amplifiers.
Generally power amplifiers are used near saturation in order to obtain the most efficient use of such amplifiers. However near saturation each amplifier has a non linear behavior, i.e. the gain decreases sharply for high input signals compared to small input signals; moreover the output signal is phase distorted due to the well known AM-PM conversion, i.e. the conversion of amplitude modulation into phase modulation.
When several channels are transmitted simultaneously the difficulty of linearization is increased by the fact that the sum of signal corresponding to these different channels, i.e. different frequencies, presents generally a non-constant envelope, i.e. the sum of such signals varies with time. The non-constant envelope must be transmitted correctly, i.e. with an amplifier working linearly in order, on one hand, to transmit signals with a good quality and, on the other hand, to prevent the transmission of interferences that would pollute the spectrum of same or other operators.
The difficulty of the problem is also increased by the fact that the higher the frequency, the lower is the yield of the amplifiers.
Several technologies are used for linearizing power amplifiers. However none of the conventional technologies is able to provide a transmission chain which may be manufactured at reasonable costs and/or with a reasonable volume, and which is able to minimize the transmission of interferences.
For a wide band amplifier the ability to minimize the transmission of interferences is characterized by the SFDR parameter, i.e. the “spurious free dynamic range”. This parameter is expressed in dB. It is the ratio between the highest of the useful transmitted powers and the maximum stray signal in the useful frequency band. In the GSM standard this SFDR parameter must be at least equal to 75 dB and in the UMTS future standard this SFDR gain is set to be at least 70 dB.
The main solution which is known, up to now, for the linearization over a wide frequency band of a non-constant envelope signal is the feedforward technology wherein no feedback loop is provided. This technology corrects, in real time, the non-linearities, even if they are frequency dependent. The non-linearities are measured by a comparison with a linear signal before amplification and they are substracted in phase opposition from the output signal of the amplifier to linearize.
A conventional feedforward linearization circuit is represented schematically on FIG. 1:
The digital signals to transmit are first processed by a digital input processor 10 and the processed digital signals are transmitted, through an analog chain 12, to the input of a power amplifier 14. For the linearization of such amplifier the output signal is transmitted to the first input 16 1 of a first substractor 16 through an attenuator 18. The second input 16 2 of substractor 16 receives the input signal of amplifier 14. The output of the first substractor 16 is amplified by a residual error amplifier 20, the output of which is connected to the input 22 1 of a second substractor 22, the other input 22 2 of which receives the output of amplifier 14 through a delay device 24.
The purpose of the feedforward circuit, with substractors 16 and 22 and amplifier 20, is to linearize the amplifier 14 itself.
However in the transmission equipment non-linearities are also introduced by the analog chain 12 and error amplifier 20, limiting the overall linearizing effect.
In order to correct the non-linearities introduced by analog chain 12 it is known to provide the circuit 10 with digital correction means which realize a digital predistortion. These correction means comprise, for instance, look-up tables which, in view of the output of amplifier 14 provide predistortion values. As shown on FIG. 1 the output of amplifier 14 is connected to a corresponding input 10 1 of circuit 10.
As the goal of the predistortion means in circuit 10 is to correct the nonlinearities of the combination of analog chain 12 with amplifier 14 but not to correct independently the distortions introduced by the analog chain 12, the amplifier 20 receives non linearized signals because the signals at the input of the amplifier 14 are distorted by analog chain 12. Therefore the input signal of residual error amplifier 20 is at a relatively high level, about 10% of the level of signals at the input of amplifier 14. Moreover, as already mentioned, the amplifier 20 itself, which is nonlinearized, may introduce non-linearities.
In brief the circuit represented on FIG. 1 is power consuming and is unable to provide a quality which is sufficient for wide band transmissions.
The invention overcomes these drawbacks.
The invention provides also a transmission equipment for a base station of a cellular telecommunication system which can transmit, and receive, signals according to different standards, for instance signals with a multiplicity of carriers and with a multiplicity of codes.
SUMMARY OF THE INVENTION
The transmission circuit according to the invention comprises:
input digital processing means with predistortion means receiving input signals from the output of a power amplifier,
these digital processing means being connected to the input of the power amplifier through an analog chain, and
the power amplifier is associated with feedforward linearization means including a first substractor which generates the difference between a signal representing the input signal of the amplifier and a signal representing the output signal of this amplifier, this difference being amplified by a residual error amplifier in order to be substracted from the output signal of the power amplifier.
This transmission circuit is characterized by the fact that it comprises a reference branch for generating a signal representing the input signal of the amplifier which is provided to the first substractor of the feedforward means, this reference branch providing a signal which is independent from the output signal of the power amplifier.
In other words the reference which is used for the feedforward circuit is independent from the predistortion, or precorrection, of the signal, or power, branch comprising the digital processing means and the analog chain. Therefore the reference signal provided to the first substractor is not distorted by the predistortion in the signal branch and the output of the first substractor may be at a lower level (for instance about 20 dB) than the output of the corresponding substractor 16 in the conventional circuit represented on FIG. 1. Moreover the error represents the residual difference between the reference signal and the predistorted amplified signal; therefore the better is predistortion, the lower is this residual error. For these reasons the power of the residual error to be amplified may be drastically reduced and the quality of the output signal of this residual error amplifier is improved with a lower power class error amplifier.
The reference branch may be similar to the signal branch, i.e. it may comprise digital processing means and an analog chain.
In an embodiment, the digital signal processing means of the reference branch comprise predistortion means which receive their input signal from the output of the analog chain of this reference branch. It is to be pointed out here that these predistortion means of the reference branch keep the independence between the feedforward circuit and the predistortion means of the signal branch. These predistortion means of the reference branch may further decrease of 20 dB (from −60 to −80 dBc) the level of spurrii at the reference output.
Although the signal, or power, branch and the reference branch are independent they preferably have common elements such as, for instance, sampling oscillators and/or local oscillators to guaranty phase matching of signals at the output of the first substractor and, therefore, minimize the residual signal representation (for example 50 dB below input level) on top of the error. Compared to the conventional feedforward, this feature contributes to the reduction of dimensioning of the error amplifier.
In a preferred embodiment the power branch comprises equalization means for the amplifier and/or attenuator(s) of the transmission channels. These equalization means are advantageously included in the digital processing means of the power branch. They provide a correction for each frequency, or each frequency subband, in order to minimize the contribution of the useful signal to the residual error signal. These equalization means may also be used to adjust the level of each carrier or subband, in line with the reference signal.
These equalization means may be preadjusted in factory. However, preferably, they are adjusted from time to time, i.e. updated through a comparison between, on the one hand, the input signal of the digital processing means and, on the other hand, the signal at the output of the power amplifier.
This updating may be improved by a selection of the data which are used for the adjustment of the equalization means; this selection is performed by an analysis of the residual error, which consists in an analysis of the spectrum of this residual error or in an analysis based on correlation. For instance the data which will be selected for updating are those for which the intermodulation interference signals are present at frequencies which are distinct from the useful frequencies.
The predistortion means may also be periodically updated in a similar manner.
Although the feedforward circuit is independent from the linearization of the transmission channel, this circuit may be used in order to improve the linearization provided by the equalization or predistortion means. For instance the operational range of the feedforward circuit may be selected in order to compensate the effects of drifts in the signal branch. In this case the period separating two successive updatings of the predistortion and/or equalization means may be increased.
The operational range of the feedforward circuit may be also selected to compensate for other defects of the power branch such as DC offsets, defects of insulation of local oscillators and frequency images, as well as quadrature defects.
In a further embodiment, the contribution of the useful signal to the residual error may be decreased by adjusting the phase of the reference signal provided at the input of the first substractor of the feedforward circuit. This phase adjustment is obtained for instance by the provision of a programmable delay circuit which may be installed at the output of the digital signal processing means of the reference branch.