EP1125411A1

EP1125411A1 - Method and system for adjusting spurious lines of an output frequency spectrum in a sideband frequency translating device

Info

Publication number: EP1125411A1
Application number: EP99949091A
Authority: EP
Inventors: Maria Luisa Gambina; Vincent Fournier
Original assignee: STMicroelectronics SA
Current assignee: STMicroelectronics SA
Priority date: 1998-10-23
Filing date: 1999-10-22
Publication date: 2001-08-22
Also published as: WO2000025494A1; FR2785111B1; US6393258B1; FR2785111A1

Abstract

The invention concerns a method which consists in: delivering in input on two channels of the frequency translating device two parameterable and mutually phase-shifted test signals TXI and TXQ; measuring the level of each of the spurious lines for different test values of different parameters; and determining the reference values for the different parameters for minimising the levels of spurious lines, by digital computation carried out from a predetermined number of different test values and corresponding measured level values, taking into account two parabolic relationships relating the levels of two spurious lines to said parameters. Thus, the adjusting time interval is reduced to a minimum by using a reduced number of test values. The invention is applicable to mobile telephone system.

Description

METHOD AND SYSTEM FOR ADJUSTING THE LEVEL OF INTERFERING RAYS OF THE OUTPUT FREQUENTIAL SPECTRUM OF A SIDEBAND FREQUENCY TRANSPOSITION DEVICE

The invention relates to frequency transposition devices with a single sideband (also designated by "IQ mixers", according to a name usually used by a person skilled in the art) and in particular the adjustment of the radio frequency emission spectrum, in 5 particularly the image rejection and the rejection of the local oscillator line of these IQ mixers.

The invention finds a particularly advantageous and non-limiting application in the field of mobile telephones incorporating such mixers.

10 In a mobile telephone, the useful signal transmitted, in this case the voice, which is a low frequency signal, is transposed into a high frequency signal in the IQ mixer using a high frequency signal originating from a local oscillator. The output spectrum of the signal from the mixer therefore contains a so-called useful line, corresponding to the useful signal transposed to

15 high frequency, but also parasitic lines, in this case the line of the local oscillator signal as well as the image line of the useful signal. However, when a mobile phone is locked for a predetermined period (for example 12.5 ms) on a frequency transmission channel, it is necessary that the spectrum of the transmitted signal occupies the

20 less possible bandwidth in order to disturb as little as possible the other frequency transmission channels which can be allocated to other radio frequency transmitters / receivers. And, the local oscillator line and the image line are likely to disturb the other channels. The purpose of adjusting the mixer is therefore to cancel, or at the very least

25 reduce as much as possible, the levels of the parasitic lines, that is to say the level of the image line and the local oscillator line.

The adjustment consists in adjusting the symmetry of the low frequency input signals from the mixer to compensate for the internal asymmetries of this component which are responsible for the increase in the image and local oscillator lines of the ^' output spectrum.

In this regard, two configurable test signals are used which are respectively delivered at the input of the two channels I and Q of the mixer. Each test signal typically has a continuous component and a periodic component (in practice sinusoidal). The periodic components of the two test signals are mutually out of phase. The adjustment consists, for a fixed value of the frequency of the local oscillator signal, in adjusting the value of the continuous components, the ratio of the amplitudes of the two sinusoidal components as well as the value of the phase shift, in order to minimize the levels of the parasitic lines. A conventional method for making this adjustment consists, for each possible local oscillator frequency (each associated with a transmission channel), of carrying out a scan on the input channel I, then on the channel Q, to measure, for the different values of the different parameters of the test signals, the levels obtained for the image and local oscillator lines, and to repeat this operation by refining the scanning steps until obtaining the desired rejection. Parameter values are then obtained, making it possible to obtain very low levels for the image and local oscillator lines.

However, this method has the drawback of requiring a very large number of measurements, typically of the order of a few tens to a few hundreds for a local oscillator frequency value, and therefore a very long adjustment time.

Furthermore, currently, the adjustment is carried out with a nominal supply voltage. It is further verified that at the end of the telephone battery life, the quality of the output spectrum of the mixer always conforms to the desired specifications. However, during the operation of a telephone, variations in the supply voltage and / or the temperature in particular can modify, for example, certain characteristics of the transistors forming certain elements of the mixer, and consequently modify its behavior, in particular at level of internal asymmetries. This then results in a mismatch between the adjustment parameters established during the test with the nominal voltage. Optimal adjustment of the mixer would then require a new adjustment of the parameters of the test signals in real time. However, such an adjustment in real time, that is to say during a period of locking the telephone on a transmission channel, is completely incompatible with the duration of adjustment currently necessary.

In addition, the efficiency of the current method is all the worse the closer one is to the solution, that is to say the reference values of the parameters making it possible to obtain the desired rejection. Indeed, even very low measurement noise tends to "trap" the iterative search algorithm in local minima, thus leading to obtaining reference values (optimal values) for these parameters, different from the values theoretical benchmarks. The invention aims to provide a solution to these problems.

An object of the invention is to provide an adjustment of the output spectrum of such IQ mixers, which requires a very small number of measurements to determine the parameters of the test signals.

The object of the invention is also to propose a setting which can be implemented in real time in a mobile telephone because of its very short duration.

The invention also aims to increase the efficiency of the adjustment.

The invention results in particular from the fact that it has been found that it is possible to correlate the power (that is to say the level) of the image and local oscillator lines of the output spectrum, to the different parameters of the test signals by two mathematical relations of the paraboloid type.

From there, the invention is remarkable in that the determination of the reference values (optimal values) for the parameters considered, reference values corresponding to theoretically zero levels of these parasitic lines, can be carried out by a numerical calculation. from a reduced number of measurements, typically a few measurement points.

The invention therefore differs from the prior art, in particular by the fact that the determination of these reference values is carried out here by a numerical computation whereas in the conventional method previously used, this determination is carried out by the only observation of the measured levels of a very large number of points, so as to select the values of the parameters having led to a minimum level. The invention thus makes it possible to reduce the duration of the adjustment in a ratio greater than ten, which reduces the cost of the products incorporating such a mixer, since the cost of the test represents a significant percentage of the cost of these products. In addition, the duration of such an adjustment according to the invention makes it easy to be able, if necessary, to implement such an adjustment in real time during the operation of a mobile telephone.

In other words, the invention therefore proposes a method for adjusting the level of the parasitic lines of the output frequency spectrum of a frequency transposition device with a single sideband, this method comprising: - the delivery at the input of the two channels the frequency transposition device, two configurable and mutually phase-shifted test signals,

measuring the level of each of the parasitic lines for different test values of the different parameters of the two test signals, and

- the determination of reference values for the various parameters making it possible to minimize the levels of the parasitic lines.

According to a general characteristic of the invention, said reference values are determined by a digital calculation carried out from a predetermined number of test values different from said parameters and from the corresponding measured level values, taking account of two parabolic relationships connecting the levels of the two parasitic lines to said parameters, so as to minimize the adjustment time by using a reduced number of test values. According to an embodiment of the method, each test signal comprises a continuous component and a periodic time component (for example a sinusoidal component). The continuous components of the two test signals constitute first and second parameters respectively, while the ratio of the amplitudes and the relative phase shift of the two periodic time components of the two test signals constitute third and fourth parameters respectively. The parasitic lines include an image line and a local oscillator line. The first and second parameters are connected at the level of the local oscillator line by a first parabolic relation, while the third and fourth parameters are linked at the level of the image line by a second parabolic relation. The reference values of the first and second parameters are then determined by performing only measurements of the level of the local oscillator line, while the reference values of the third and fourth parameters are determined by performing only measurements of the level of the stripe image.

In general, several variants are possible for determining these reference values, taking into account the paraboloid type correlation connecting the levels of the parasitic lines to the values of the parameters of the test signals.

A first variant of the invention consists in calculating the position of the optimum of the parabolic relation considered by determining the point where the derivative is zero.

More precisely, according to this variant, the reference value of each parameter is determined by using at least four different test values of said parameter, so as to obtain at least four corresponding measured values of level. The value of the parameter corresponding to a zero derivative of the corresponding parabolic relationship is then calculated from said test values and corresponding level values, this value constituting said reference value for the parameter considered.

In practice, the number of test values used for this variant is equal to four.

From the four test values and the four measured level values, two triplets of consecutive test values can be formed, for example the first, the second and the third on the one hand, and the second, the third and the fourth. on the other hand. The difference between the two measured level values corresponding to the extreme test values of each triplet is then advantageously calculated (for example, on the one hand, the difference between the levels corresponding to the first and third test values, and, on the other hand, the difference between the levels corresponding to the fourth and second test values). We then make the relationship between this difference and the difference of said two extreme values of the triplet considered (in this case, on the one hand, the difference between the third and the first test values and, on the other hand, the difference between the fourth and second test values). We thus obtain, for each triplet, a derivative value associated with the median test value of said triplet (in this case, a derivative value associated with the second test value and a derivative value associated with the third test value ). This reference value is then calculated from the equation of the line passing through two derivative values associated with two different median test values (in this case, the equation of the line passing through the two derivative values previously calculated).

Other variants of the invention consist in taking a certain number of measurement points and in solving a linear system of several equations with several unknowns, for example according to the known method known as of Kramer.

More precisely, according to a second variant of the invention, the reference value of each parameter is determined by using only three different test values of said parameter so as to obtain three corresponding measured values of level, and to obtain a linear system of three equations with three unknowns from the corresponding parabolic relation, the three different test values and the three corresponding measured level values. Then said linear system is solved.

Another variant of the invention, even more advantageous since it requires an even smaller number of measurement points, makes it possible to determine the respective reference values of the first and second parameters by using only four different pairs of test values for said values. first and second parameters, so as to obtain four corresponding measured level values and to obtain a first linear system with four equations with four unknowns from the corresponding parabolic relation, from the four pairs of test values and from the four measured level values corresponding. We then solve this first linear system. Similarly, we. determines the respective reference values of the third and fourth parameters using only four different pairs of test values for said third and fourth parameters, so as to obtain four corresponding measured values of level and to obtain a second linear system of four equations with four unknown from the corresponding parabolic relationship, the four pairs of test values and the four measured level values. We then solve this second linear system.

The invention also relates to a system for adjusting the level of the parasitic lines of the output frequency spectrum of a frequency transposition device with a single sideband, comprising delivery means for delivering at the input of the two channels of the transposition device. frequency, two configurable and mutually out of phase test signals, means for measuring the level of each of the parasitic lines for test values different from the different parameters of the two test signals, and means for determining reference values for the different parameters to minimize the level of stray lines.

According to a general characteristic of the invention, the means for determining the reference values comprise processing means, for example carried out in software within a microprocessor, capable of performing a numerical calculation from a predetermined number of different test values of said parameters and corresponding measured level values, taking into account two parabolic relationships connecting the level of the two parasitic lines to said parameters.

The invention also relates to a mobile telephone containing a frequency transposition device and such an adjustment system, so as to allow the implementation of the method according to the invention within the telephone, whether in the factory or else in real time during phone operation.

Other advantages and characteristics of the invention will appear on examining the detailed description of modes of implementation and embodiments, in no way limiting, and the appended drawings, in which: - Figure 1 schematically illustrates an IQ mixer and the associated adjustment means;

- Figure 2 illustrates the output spectrum of an IQ mixer;

- Figure 3 illustrates a correlation according to the invention between the level of the parasitic lines and some of the adjustment parameters;

- Figure 4 illustrates a section in the plane Q of the paraboloid of Figure 3;

- Figure 5 partially illustrates a first variant implementation of the method according to the invention; - Figure 6 illustrates very schematically an embodiment of an adjustment system according to the invention; and

- Figure 7 illustrates another embodiment of an adjustment system according to the invention, particularly compatible with incorporation within a mobile phone. In FIG. 1, the reference DTF designates a frequency transposition device with a single side band, or mixer IQ, comprising two mixing channels VF1 and VF2 joined by a terminal adder AD connected to the output terminal BS of the mixer. The channel VFl comprises a mixer M 1 receiving, on the one hand, the signal TXI present at the input terminal BI of the mixer, and, on the other hand, a signal OLl equal to

LOcosΩt and from a local OL oscillator after 90 ° phase shift in a DP phase shifter.

In an analogous manner, the channel VF2 comprises a mixer M2 receiving, on the one hand, the signal TXQ present at the input BQ of the mixer, and, on the other hand, a signal OL2 equal to LOsinΩt and coming directly from the 'local oscillator OL.

To adjust the output spectrum of such a DTF mixer, two configurable test signals TXI and TXQ are given to the two inputs BI and BQ of the mixer, given respectively by formulas (1) and (2).

TXI = DCI + sin ωt (1)

TXQ = DCQ + r cos (ωt + 0) (2)

In formula (1), DCI represents the continuous component of the TXI signal, and sin ωt the temporal sinusoidal component of pulsation ω. In formula (2), DCQ represents the continuous component of TXQ signal, while r cos (ωt + 0) represents the cosine component of this TXQ signal. r represents the amplitude of this cosine component. It should be noted in this regard that since the amplitude of the sinusoidal component of the signal TXI has been chosen equal to 1, r here in fact represents the ratio of the amplitudes of the two periodic components of the two test signals. Likewise, 0 represents the relative phase shift of the two periodic components of the two signals TXI and TXQ. If the mixer Q was perfect, the output signal SS would only comprise a line called useful line RU centered on the frequency FL-tf, where FL is the frequency of the local oscillator (corresponding to the pulsation

Ω) and where f is the frequency of the input signal (in this case corresponding to the pulsation ω).

This being so, in practice, a mixer has production imperfections, in particular with regard to the production of the phase shifters and / or of the transistors constituting this mixer. It therefore follows that the output spectrum of the signal SS comprises, in addition to the useful line RU, two parasitic lines (FIG. 2), namely the line RL of local oscillator centered on the frequency FL and an image line RM centered on the frequency FL-f.

Adjusting the output spectrum of the DTF mixer consists in varying the different parameters of the two test signals so as to minimize the level or power P of the parasitic lines RM and RL.

The prior art adjustment method consisted in varying the parameters of the TXI signal according to a given scanning step, keeping the parameters of the TXQ signal constant and then repeating this scanning on the Q channel while keeping the parameters of the TXI signal constant. . These two iterative scans were then repeated with smaller scanning steps until the values of the parameters having led to the obtaining of a minimum level for the line RM and the line RL were selected.

The invention proposes a method radically different from this, which requires only a few measurement points.

In this regard, it was found that the power P of the local oscillator line could be correlated to the parameters DCI and DCQ by a first relation of the paraboloid type as defined by formula (3). P (RL) = Kl (DCI - DCIO) ² + Kl (DCQ - DCQO) ² (3)

Similarly, it was found that the power of the image line could be correlated to the other two parameters, r and 0, by a second mathematical relation of the paraboloid type as defined by formula (4).

P (IM) = K2 (r - rO) ² + K2r (0 - 0O) ² (4)

In these formulas, DCI, DCQ, r and 0 represent the current test values of the different parameters, while DCIO, DCQO, rO and 00, represent the reference values of these parameters for which the power of the line RL and the power of the IM line are zero. Kl and K2 are constants.

The purpose of the adjustment therefore consists in determining by calculation these reference values. This determination is made in a BTR processing block (FIG. 1) comprising conventional means of spectrum analysis MS and processing means MT such as a microprocessor.

We will now describe in more detail with particular reference to FIGS. 3, 4 and 5, modes of implementing the method making it possible to determine the reference values DCIO and DCQO of the parameters DCI and DCQ.

In FIG. 3, the reference PBL1 designates the correlation paraboloid between the power of the line RL and the different values of the parameters DCI and DCQ.

For a fixed DCQf value, the section of the paraboloid PBL1 in the DCQf plane is a PBI parabola as illustrated in FIG. 4.

A first variant of the invention for determining the reference value DCIO and the reference value DCQO, will consist in using the fact that the derivative of a parabola is a straight line and will consist in finding the point POR where this derivative is zero . More precisely, one fixes for the other parameters DCQ, r and 0, predetermined values DCQf, rf, and 0f.

Four different test values are also chosen for the DCI parameter, namely the DCU, DCI2, DCI3 and DCI4 test values, and the corresponding test signals TXI and TXQ are successively applied. The measurement means MS of the processing block BTR determine then for each of the points PO1, P02, P03 and P04 of the parabola PBI the corresponding levels PI, P2, P3 and P4 of the local oscillator line RL. From there, for the triplet of measuring point PO1, P02 and P03, the value of the derivative P'2 at point P02 is determined by formula (5).

P3 - PI

P'2 = (5) (DCI3 - DCI1)

Likewise, for the triplet P02, P03 and P04, the value P'3 of the derivative at point P03 is determined by formula (6).

P4 - P2

P'3 = (6)

(DCI4 - DCI2)

The two measurement points P02 and P03 having coordinates (P'2, DCI2) and (P'3, DCI3) respectively, allow to determine the two parameters α and β of the line D of equation:

F = α DCI - β (7)

We can therefore easily deduce the DCIO reference value which is equal to β /.

Obtaining the reference value DCIO was therefore carried out by a simple numerical calculation, carried out in software within the processing means MT of the processing block BTR and required only four separate measurements of the level of the corresponding line RL to four different DCI1-DCI4 test values for the DCI parameter.

Of course, it would have been possible to use a greater number of measurement points to determine in particular with more precision the parameters β and α of the line D. Once the DCIO reference value has been obtained, the same operations are repeated for the calculation of the DCQO reference value of the DCQ parameter. In other words, this time a constant value is chosen for the DCI parameter, for example the DCIO value which has just been calculated. Four different test values DCQ1-DCQ4 are chosen for the parameter DCQ and one calculates in a similar manner to that which has just been described for the reference value DCIO, the reference value DCQO.

We could possibly perform an additional iteration to verify in particular the DCIO value obtained by setting this time for the DCQ parameter, the DCQO reference value which has just been calculated, and by recalculating in a manner analogous to that which has just d 'be described, the reference value DCIO.

Those skilled in the art will therefore have noticed that this variant of the invention required, for a given frequency FL of local oscillator, eight measurements of the level of the line RL to determine the reference values DCIO and DCQO of the parameters DCI and DCQ (four measurements corresponding to the four different DCI1-DCI4 test values and four measurements corresponding to the four different DCQ1-DCQ4 test values).

What has just been described also applies to the calculation of the reference values rO and 00 of the other two parameters r and 0 by measuring this time the level of the image line LM and using the properties of the type relation paraboloid correlating this level of the image line with the values of the parameters r and 0.

Another implementation variant of the invention will now be described, which makes it possible to use only three different test values for the calculation of each parameter.

More precisely, if we refer again to formula (3) giving the correlation of the power of the local oscillator line with respect to the DCI and DCQ parameters, we notice that this formula (3) can, once developed, take the form of formula (8) below.

P (RL) = a DCI ² + b DCI + c (8) In this formula (8), the variables a, b and c are related to the constants K1, DCIO, and DCQ.

Under these conditions, the DCIO reference value is calculated as follows:

One fixes a value DCQf for the parameter DCQ and one chooses this time three different test values DCI1-DCI3 for the parameter DCI, for example those corresponding to the points POl, P02 and P03 of figure 4. One then measures the levels PI, P2 and P3 of the local oscillator line, which then provides the linear system (9) of three equations with three unknowns, in which the three unknowns are the variables a, b and c respectively.

PI = a DCU ² + b DCU + c P2 = a DCI2 ² + b DCI2 + c (9)

P3 = a DCI3 ² + b DCI3 + c

This system can be easily solved and the DCIO value is then equal to the ratio -b / 2a. The same operations can be carried out for the calculation of the reference values DCQO, rO and 00.

Those skilled in the art therefore note that this variant of the invention requires only three different test values for the calculation of each parameter, for a given value of the frequency FL. Another variant of the invention, even more advantageous, makes it possible to calculate pairs of parameters, this time using only four values of different couples for these parameters. More precisely, relation (3) can take the form of relation (10).

P (RL) = a (DCI ² + DCQ ² ) + b DCI + c DCQ + d (10) with a = Kl b = - 2 Kl. DCIO c = - 2 Kl.DCQ0 d = DCIO ² + DCQO ² In this case we choose respectively four different couples (DCI1, DCQ1), (DCI2, DCQ2), (DCI3, DCQ3), and (DCI4, DCQ4), and we measure for each of these couples the level of the line RL . We then deduce a system of four equations with four unknowns in which the unknowns are the variables a, b, c and d. Once this system has been solved by the processing means, the DCIO value is obtained by calculating the ratio -b / 2a, and the DCQO value by calculating the ratio -c / 2a.

An analogous calculation can be carried out to determine the reference values rO and 00 of the pair of parameters r and 0 using this time the level of the image line IM and the relation of the paraboloid type given by the formula (4) below. before.

This variant, which can be described as "global", therefore makes it possible to obtain the four reference values of the four parameters sought by using only eight level measurements (eight acquisitions)

(four level measurements of the RL line and four level measurements of the IM line), for each value of FL set.

Another variant, which can be described as "compact", makes it possible to obtain the four reference values of the four parameters sought by using only four acquisitions for each value of FL fixed. Indeed, using the relation (10) and the homologous relation of the image line, we choose four different quadruplets (DCIi, DCQi, ri, 0i) i = 1 to 4, and we measure for each of these quadruplets, the level of the image line and the level of the local oscillator line. Two independent systems of 4 equations with 4 unknowns are then deduced therefrom which are solved in a manner analogous to what has been described for the global variant, so as to obtain the reference values of the parameters.

Of course, the reference values which have just been obtained have been calculated for a given value of the local oscillator frequency. This calculation can be performed for all the local oscillator frequencies corresponding to the different frequencies of the transmission channels subdividing the radio frequency transmission band of a mobile telephone (for example sixty channels extending in the frequency band 850 MHz- 950 MHz). In practice, for example in a TMC mobile telephone equipped with an AT antenna (FIG. 6), the DTF mixer is preceded by an auxiliary circuit CX formed by two channels VX1 and VX2 respectively connecting the two input terminals BI and BQ from the mixer to a BX terminal receiving the useful signal, in this case the voice. The VX1 channel includes an amplifier

AMI followed by an adder AD1 also receiving an auxiliary voltage delivered by a first auxiliary voltage source ST1. The channel VX2 comprises time delay means MR, for example a delay line, followed by a variable gain amplifier AM2, followed by an adder AD2 also also receiving an auxiliary voltage from a second auxiliary voltage source ST2.

The value of the auxiliary voltage ST1 corresponds to the parameter DCI, while the value of the auxiliary voltage ST2 corresponds to the parameter DCQ. The variable gain of the amplifier AM2 corresponds to the parameter r and the time delay induced by the delay means MR corresponds to the phase shift 0.

Thus, for a given frequency of the local oscillator, the value of the time delay, the gain value of the amplifier AM2 and the values of the auxiliary voltages ST1 and ST2 can be adjusted, so that these values correspond to the reference values. 00, rO, DCIO and DCQO obtained when adjusting the mixer.

While this adjustment can be carried out in the factory and stored in a memory of the mobile telephone, the adjustment method according to the invention advantageously makes it possible to be able to carry out such adjustment in real time during the operation of the mobile telephone TMC. In this regard, the adjustment system illustrated in FIG. 7 can be provided. The BTR processing block comprising a spectrum analyzer and a processor, generally present in any mobile phone, permanently receives the output signal from the DTF mixer.

Furthermore, those skilled in the art know that the mobile telephone is locked for predefined elementary durations (for example 12.5 ms) on a selected frequency channel.

The invention therefore provides, for example at the start of the elementary locking duration, to switch the input of the auxiliary circuit CX to a generator GEN generating a periodic auxiliary signal, by example a sinusoidal signal. The processor included in the processing block BTR then varies the values of the auxiliary voltage sources, that of the gain of the amplifier AM2 and the value of the delay of the delay means MR, and determines by a digital calculation analogous to that which comes to be described, the different reference values taking into account the current frequency of the local oscillator and the actual operating conditions of the telephone.

Once these parameters are fixed, the processor switches the input of the auxiliary circuit CX to the input receiving the speech SU.

Claims

1. Method for adjusting the level of the parasitic lines of the output frequency spectrum of a frequency transposition device with a single sideband, comprising the input as input of the two channels of the frequency transposition device of two test signals which can be parameterized and mutually phase shifted (TXI, TXQ), measuring the level of each of the parasitic lines (RL, IM) for different test values of the different parameters of the two test signals, and determining reference values for the different parameters making it possible to minimize the levels of the parasitic lines, characterized in that the said reference values are determined (DCIO, DCQO, rO, 00) by a numerical calculation carried out from a predetermined number of test values different from the said parameters and the values of corresponding measured levels, taking into account two parabolic type relationships connecting the levels of the two parasitic auxd lines its parameters, so as to minimize the adjustment time by using a reduced number of test values.

2. Method according to claim 1, characterized in that each test signal (TXI, TXQ) comprises a continuous component (DCI, DCQ) and a periodic time component, the continuous components of the two test signals constituting first and first respectively second parameters while the ratio (r) of the amplitudes and the relative phase shift (0) of the two periodic time components of the two test signals constitute third and fourth parameters respectively, by the fact that the parasitic lines comprise an image line (IM) and a local oscillator line (RL), by the fact that the first and second parameters are connected at the level of the local oscillator line by a first parabolic relation while the third and fourth parameters are linked at the line image by a second parabolic relation, by the fact that the reference values (DCIO, DCQO) of the first and second parameters are determined es by performing only local oscillator line (RL) level measurements, while determining the reference values (rO, 00) of the third and fourth parameters by performing only measurements level of the image line (IM).

3. Method according to claim 1 or 2, characterized in that the reference value of each parameter is determined by using at least four different test values (DCI1-DCI4) of said parameter so as to obtain at least four measured values corresponding level, then calculating from said test values and corresponding level values, the value of said parameter corresponding to a zero derivative of the corresponding parabolic relationship, this value constituting said reference value (DCIO) for the parameter considered.

4. Method according to claim 3, characterized in that the calculation of the reference value of said parameter comprises for at least two triplets of consecutive test values, the calculation of the difference between the two measured level values corresponding to the values of extreme test of each triplet, and the ratio between this difference and the difference of said two extreme values of the triplet considered, so as to obtain for each triplet a derivative value (P ') associated with the median test value of said triplet, then the calculation of said reference value from the equation of the straight line (D) passing through two derivative values associated with two different median test values.

5. Method according to claim 1 or 2, characterized in that the reference value of each parameter is determined by using only three different test values (DCI1-DCI3) of said parameter so as to obtain three corresponding measured values of level , and to obtain a linear system of three equations with three unknowns from the corresponding parabolic relation, the three different test values and the three corresponding measured level values, then by solving said linear system.

6. Method according to claim 2, characterized in that the respective reference values (DCIO, DCQO) of the first and second parameters are determined by using only four different pairs of test values for said first and second parameters so as to obtain four corresponding measured values of level and to obtain a first linear system of four equations with four unknowns starting from the corresponding parabolic relation, the four couples test values and the four corresponding measured values of level, then in solving said first linear system, and by the fact that the respective reference values (r0, 00) of the third and fourth parameters are determined by using only four different pairs of test values for said third and fourth parameters so as to obtain four corresponding measured values of level and obtaining a second linear system of four equations with four unknowns from the corresponding parabolic relation, the four test value pairs and the four measured level values, then by solving said second linear system.

7. Method according to one of the preceding claims, for adjusting the level of parasitic lines of the output frequency spectrum of a frequency transposition device incorporated in a mobile telephone locked on a transmission channel for a predetermined elementary duration, characterized by the fact that said reference values are determined during said elementary duration.

8. System for adjusting the level of the parasitic lines of the output frequency spectrum of a frequency transposition device with a single side band, comprising delivery means (CX) for delivering at the input of the two channels of the frequency transposition device two configurable and mutually out of phase test signals (TXI, TXQ), means of measurement (MS) of the level of each of the parasitic lines for test values different from the different parameters of the two test signals, and means of determination (MT) of reference values for the various parameters making it possible to minimize the levels of the parasitic lines, characterized by the fact that the determination means comprise processing means (MT) capable of performing a numerical calculation from a predetermined number of values of different test of said parameters and corresponding measured level values, taking into account two type relationships parabolic connecting the levels of the two parasitic lines to said parameters, so as to reduce the adjustment time by using a reduced number of test values.

9. System according to claim 8, characterized in that each test signal comprises a continuous component and a periodic time component, the continuous components (DCI, DCQ) of the two test signals constituting first and second parameters respectively while the ratio of the amplitudes (r) and the relative phase shift (0) of the two periodic time components of the two test signals constitute respectively third and fourth parameters, by the fact that the parasitic lines comprise an image line and a local oscillator line, by the fact that the first and second parameters are connected at the level of the local oscillator line by a first parabolic relation while the third and fourth parameters are linked to the level of the image line by a second parabolic relation, by the fact that the processing means (MT) determine the reference values of the first and second parameters only from the level measurements of the local oscillator line, and the reference values of the third and fourth parameters only from the level measurements of the line picture.

10. The system as claimed in claim 9, for adjusting the level of the parasitic lines of the output frequency spectrum of a frequency transposition device incorporated in a mobile telephone, characterized in that the means for delivering the test signals are also incorporated in the mobile telephone and comprise a generator (GEN) of a periodic auxiliary signal, a circuit (CX) comprising an input terminal (BX) capable of being connected to the output of the generator, a first channel connecting said terminal d input to a first input (BI) of the frequency transposition device and including a first amplifier (AMI) and a first adder (AD1) receiving on the one hand the output of the first amplifier and the output of a first voltage source variable auxiliary (ST1), and a second channel connecting said input terminal (BX) to a second input (BQ) of the frequency transposition device and including means variable delay (MR), a second variable gain amplifier (AM2) and a second adder (AD2) receiving on the one hand the output of the second amplifier and the output of a second variable auxiliary voltage source (ST2), and by the fact that the processing means are capable of selectively varying the values of the delay, the gain and those of the auxiliary voltages so as to generate said test signals.

11. Mobile phone, characterized in that it contains a frequency transposition device and an adjustment system as defined in one of claims 8 to 10.