ITMI20100927A1 - METHOD, DEVICE AND ELECTRONIC SYSTEM FOR WIDE STRIP DISPLACEMENT OF THE UNBALANCED BETWEEN A CHANNEL IN PHASE AND A CHANNEL IN TRANSCEIVER SQUARE - Google Patents

Description

METHOD, DEVICE AND ELECTRONIC SYSTEM FOR THE BROADBAND COMPENSATION OF THE UNBALANCE BETWEEN A

CHANNEL IN PHASE AND A CHANNEL IN SQUARE OF

TRANSCEIVERS

------

The present invention relates to a method, a device and an electronic system for broadband compensation of the imbalance between a phase channel and a quadrature channel in transceivers, particularly but not exclusively useful in the field of radio frequency communications.

Nowadays radio frequency transceivers, which operate on signals modulated by means of two carriers in phase and quadrature, are increasingly used in telecommunications, with the aim of reducing implementation costs, improving energy savings and bandwidth occupancy.

The adoption of this type of transceiver allows, in fact, to transmit through the same transmission medium and in the same band two different analog signals, which are then separated during the demodulation phase.

The simplest architectural version, therefore less expensive and more used, of radio frequency transceivers, for phase and quadrature modulated signals, is certainly the one that operates with direct frequency conversion, as illustrated in principle in figure 1 and in figure 2.

Such transceivers comprise a transmitter device 10 and a receiver device 30.

The transmitter device 10, which receives in input a pair of signals I and Q to be transmitted numerically, includes a pair of digital-to-analog converters (Digital to Analog Converter or DAC) 11 and 12, which convert time-discrete binary sequences, corresponding to the two signals I and Q, in time-continuous analog signals, followed by two low pass filters (LPF) 13 and 14, which limit the aforementioned signals in the operating passband provided for the transmitter device 10.

The low-pass filters 13 and 14 have respective outputs on which signals are fed which are then sent in input to a complex modulator 15 (I / Q mixer or I / Q mixer), whose oscillation frequency fc, in the radio frequency domain , Is generated by a local oscillator (Local Oscillator or LO) 16.

This complex modulator 15 comprises in detail the local oscillator 16, a phase shifter 17, two multipliers 18, 19 and an adder 20.

The carrier signal coming from the local oscillator 16, therefore, drives the multiplier 19 and, passing through the phase shifter 17, with an ideal phase shift equal to 90 °, consequently drives the multiplier 18.

The output signals from the low-pass filters 13 and 14 are then applied to these multipliers 18, 19, which produce two radiofrequency modulated signals on their respective outputs, which are then added together by means of an adder 20 and the resulting signal is shaped from a band pass filter (Band Pass Filter or BPF) 21.

The signal at the output of the band-pass filter 21 is, therefore, amplified in power by an amplifier device (Power Amplifier or PA) 22, to be finally sent to a transmitting antenna 23, or to another user not shown in the figure.

Similarly, the receiver device 30 comprises, as illustrated in Figure 2, a receiving antenna 31, suitable for receiving the signals modulated and transmitted by a transmitter device 10 of the type described above.

The signal received by the receiving antenna 31 is then filtered by a band pass filter 32 (Band Pass Filter or BPF) and amplified by a generally low noise amplifier (Low Noise Amplifier or LNA) 33.

Subsequently the signal is applied to an I / Q mixer 34, able to reconstruct, in analogue, the received I / Q signal, composed of a local oscillator (Local Oscillator or LO) 35, operating at the carrier frequency fc, a phase shifter 36 with an ideal phase shift of 90 ° and two multipliers 37 and 38.

The resulting signal, in its phase and quadrature components, is filtered by a pair of Low Pass Filters (LPF) 39 and 40 followed by a pair of Variable Gain Amplifiers (VGA) 41 and 42 , which are used to adapt the signal strength level to the conditions of the received radio signal.

The resulting analog signals I and Q are then sampled and quantized in discrete levels by two analog-to-digital converters (ADC) 43, 44 and are subsequently made available for numerical demodulation.

The transceivers described above suffer from some drawbacks, due to the non-ideality of their components both in transmission and in reception, such as the imbalance in amplitude and phase of the two signals I and Q (I / Q unbalance).

The problem of this I / Q imbalance derives precisely from the non-ideal operation of the analog electronic components which are included in the transmitter device 10 and in the receiver device 30, respectively downstream of the DACs 11 and 12 and upstream of the ADCs 43 and 44.

In practice, essentially two contributions to the I / Q imbalance are observed: one that depends only on the carrier frequency fc, but not on the band of the modulating signal, and the other that depends on both the frequency fcs and the band of the modulating signal.

In particular, there is an I / Q imbalance term dependent only on the carrier frequency fc, mainly due to the imperfect quadrature of the I / Q mixer 15 or 34: this imbalance can be of amplitude, if the signals applied to multipliers 18, 19, 37 and 38 are not of equal amplitude, or if the multipliers themselves are not identical, and / or phase, if the phase shifter 17 or 36 does not introduce a phase shift exactly equal to 90 °.

Another I / Q imbalance term is also observed, depending both on the frequency fcs and on the signal band, due to the low-pass filters 13, 14, 39 and 40, by the analog multipliers 18, 19, 37 and 38, and to the stages 22, 33, 41 and 42 amplifiers used. In fact, since these devices are made with electronic components with predefined tolerances, their frequency responses are different, causing a frequency-dependent imbalance between the two channels in phase and quadrature.

The band-pass filters 21 and 32 can also have a non-symmetrical frequency response around the carrier frequency fc, thus causing mutual interference between the in-phase and quadrature components of the transmitted or received signal.

The macroscopic effect of the I / Q imbalance is, in the case in which a signal is applied on a lateral band (right or left), as can be seen in figure 3, the appearance of an attenuated and overturned version of the signal itself on the opposite side band, which is called the image signal.

In this regard, we can also define the concept of Image Rejection Ratio (IRR): it is a function of the frequency and represents the ratio (usually in dB) between the power of the transmitted or received signal. and that of the image signal.

Normally it has an asymmetrical trend with respect to the central frequency fc (that is, according to whether the right or left side band is excited), therefore the right IRR + and left IRR- rejection ratios are defined:

IRR (f) <P> U (<f>)

<+ =>

P <L> (f) f> fc

<P>)

IRR ∠’(f) L (<f>

<=> P <U> (f) f <fc

where PU (f) and PL (f) respectively represent the signal power as a function of the frequency f in the right side band (Upper Side Band i.e. for f> fc) and left (Lower Side Band i.e. for f <fc) with respect to the frequency of carrier fc.

Typical values for image rejection are between 30 and 40 dB in consumer devices and between 45 and 60 dB in professional devices.

Furthermore, as can be seen in figure 3, in practice the effect of the carrier residue fc is also found.

The effect of this residue is to introduce in the signal transmitted by the transmitter device 10 a spectral line at fcanche frequency in the case of carrier suppressed transmission.

This drawback is typically due, in the case of the transmitter device, to the losses due to the carrier generator and to the continuous component present at the inputs of the I / Q 15 mixer.

Similarly, in the case of the receiver device, the imperfect isolation of the I / Q mixer 34 is manifested by a continuous component superimposed on the demodulated I / Q signal.

Various techniques and apparatuses are known today to compensate for the I / Q imbalance, both upstream of the transmitter device 10 and downstream of the receiver device 30, which are capable of bringing the IRR to values close to 70 - 80 dB , that is to bring the spectrum of the image signal to the power level, or below, of the noise of the device.

These I / Q imbalance compensation techniques generally operate on numerical and non-analog signals and are based on the measurement and estimation of the I / Q imbalance parameters and on the subsequent calibration of the transmitter device 10 or of the receiver device 30.

The techniques for estimating the I / Q imbalance can be distinguished in off-line and on-line according to the fact that they refer respectively to measurement and calculation procedures of the calibration coefficients applied when the device is not operational rather than to estimation and calculation procedures applied during the normal operation of the transceiver.

In addition, the techniques for estimating the I / Q imbalance and calculating the calibration parameters can be combined both with open-loop architectures, in which new measurements are made, new estimates made and the new parameters updated when the device is initialized. calibration, and closed loop, in which the measurements, estimates and updating of the calibration parameters take place in real time, for example using an adaptive algorithm.

In both architectures, once the new estimate of the calibration parameters has been made, the compensation usually takes place in real time.

Referring to off-line techniques for estimating and calculating the calibration parameters, the transceiving of signals is generally suspended and suitable test signals are applied; this phase is followed by a measurement of the I / Q imbalance and then by that of estimating and calculating the calibration parameters. Once the calibration parameters have been estimated, they are used to compensate for the device during its normal operation.

The type of measurement can be of a consistent or inconsistent type depending on whether or not there is synchronization between the electronic measurement system and the local oscillator, or clock generator, of the transceiver system under test.

The following step of estimating the calibration parameters in known compensation techniques produces a matrix of correction coefficients, which are valid only for a certain frequency fc.

These compensation techniques therefore have the drawback that the compensation that follows is not valid, effective and efficient in the whole band of the modulated signal that the transmitter device 10 is able to support and therefore it is not possible to eliminate effectively. this I / Q imbalance over the whole aforementioned band.

The purpose of the present invention is to describe a method for broadband compensation of the imbalance between a channel in phase and a quadrature channel in transceivers capable of solving the above mentioned drawbacks.

Another object of the present invention is to describe a method for broadband compensation of the imbalance between a phase channel and a quadrature channel in transceivers capable of providing a wide band calibration.

Another object of the present invention is that of realizing an electronic system for compensating the I / Q imbalance capable of estimating the calibration parameters and compensating the input and / or output signals to the transceiver.

These and other purposes according to the present invention are achieved by realizing a device, an electronic system and a method for broadband compensation of the imbalance between a phase channel and a quadrature channel in transceivers as claimed in claims 1, 9 and 15 respectively. .

Further characteristics of the method, device and electronic system for broadband compensation of the imbalance between a phase channel and a quadrature channel in transceivers are the subject of the dependent claims.

The characteristics and advantages of a method and an electronic system for broadband compensation of the imbalance between a phase channel and a quadrature channel in the transceivers according to the present invention will become more evident from the following description, which is exemplary and not limiting, referring to the drawings. schematic annexes in which:

- figure 1 is a block circuit diagram of a known type of transmitter device;

- figure 2 is a block circuit diagram of a known type of receiver device;

Figure 3 shows diagrams of the power spectra of radiofrequency signals;

Figure 4 is a block circuit diagram of a first embodiment of an electronic system for broadband compensation of the unbalance between a phase channel and a quadrature channel in the transmitter device of Figure 1 according to the present invention;

Figure 5 illustrates a block diagram of a first compensation method implemented by the electronic system of Figure 4;

figure 6 illustrates a reference table of test signals applied to the electronic system of figure 4;

Figure 7 illustrates a block diagram of a second method of broadband compensation of the imbalance between a channel in phase and one in quadrature implemented by the electronic system of Figure 4;

Figure 8 is a block circuit diagram of a second embodiment of an electronic system for broadband compensation of the unbalance between a phase channel and a quadrature channel in the transmitter device of Figure 1 according to the present invention;

Figure 9 illustrates a block diagram of a broadband compensation method of the imbalance between one channel in phase and one in quadrature implemented by the electronic system of Figure 8;

Figure 10 is a block circuit diagram of a first embodiment of an electronic system for broadband compensation of the unbalance between a phase channel and a quadrature channel in the receiver device of Figure 2 according to the present invention;

Figure 11 is a block circuit diagram of a second embodiment of an electronic system for broadband compensation of the unbalance between a channel in phase and a quadrature channel in the receiver device of Figure 2 according to the present invention.

With reference to the figures, an electronic system is shown for the compensation of the I / Q imbalance in a transmitter device in phase and quadrature, indicated as a whole with 50.

This electronic system 50 for the compensation of the I / Q imbalance comprises a means 52 for the generation of two numerical test signals zI (k) and zQ (k), a unit for measuring, estimating and controlling the compensation parameters 80 and an antenna coupler 56; this system 50 is applied to a transmitter device 51 for analog phase and quadrature modulation.

In particular, the transmitter device 51 is of the type shown in Figure 1 and is connected to a means for generating a TS period clock 54, suitable for driving the DACs 11 and 12, and to a local oscillator 55 at frequency fc, for the analog modulation of the Ipe Qpi compensated transmission signals which respectively feed a baseband phase channel and a baseband quadrature channel of the transmitter device 51.

According to the present invention, the means 52 for generating two numerical test signals zI (k) and zQ (k), driven by a means for generating a clock 57 of period Tsâ € ™ â € ™, is advantageously connected to a compensation device 60 for broadband numerical compensation of the I / Q imbalance and of the carrier residue.

This compensation device 60 comprises a first pair of selector means 61 and 62 having a first position and a second switching position thanks to which it is possible to pass from a calibration configuration, in which the compensation device 60 is powered by two test signals zI (k) and zQ (k) generated by the means for generating two digital signals 52 (position 1), to an operating configuration (position 2), in which the compensation device 60 is powered by the two signals baseband transmission in phase I and quadrature Q and in which the transmitter device 10 sends such signals into the ether through its own now balanced phase and quadrature channels. The operational configuration is activated following the end of the calibration procedure which will be described in more detail below.

These selector means 61 and 62, which in particular are diverters capable of operating in the base band, are therefore connected to a second pair of selector means 63, 64 which are selectively connected to a third pair of selector means 65, 66 (position 2) or advantageously to a system of four filters 67, 68, 69 and 70, of the numerical type and preferably of the FIR (Finite Impulse Response) type of length L samples (position 1).

In particular, this system of four numerical filters 67, 68, 69 and 70 implements a matrix convolution product operation between a column vector of inputs, the components of which are the signals which respectively feed the inputs of the second pair of selector means 63, 64 , and a square matrix compensation means (140) of numerical filters, the components of which are the impulse responses of the four aforementioned numerical filters 67, 68, 69 and 70.

In fact, as it is possible to observe in figure 4, the first and third numerical filters 67, 69 are fed at the input by the signal coming from the first selector means 63, while the second and fourth numerical filters 68, 70 are fed by the signal coming from from the second sorting half 64.

Furthermore, the outputs of these filters 67, 68, 69 and 70 are associated two by two and feed the respective inputs of two first summing means 71 and 72 which are also part of the compensation device 60.

In particular, the first filter 67 and the second filter 68 have a respective outlet placed at the inlet of the first adder means 71, while the third and fourth filters 69, 70 have a respective outlet placed in the inlet of the second adder means 72.

According to the present invention, the outputs of the first adding means 71 and 72 are applied to the input of two second adding means 73 and 74 thanks to which they are added to compensation constants for the carrier residue cI and cQ.

The second and third pair of selector means 63, 64, 65 and 66 are advantageously controlled by a measurement, estimation and control unit 80, which calculates the compensation parameters of the in-phase and quadrature channels through a procedure iterative. If these selection means 63, 64, 65 and 66 are in position 1, as indicated in Figure 4, the compensating device 60 is activated; in this configuration the signals first compensated by the filter system 67-70 and by the summing means 71, 72, 73, 74 feed the inputs of the transmitter device 51.

Otherwise, or when the selection means 63, 64, 65 and 66 are in position 2, the compensating device 60 is deactivated; in this configuration the digital signals generated by the means 52 or the digital transmission signals I and Q reach the transmitter device 51 without undergoing any alteration.

The output of the transmitter device 51 is also preferably connected to an antenna coupler 56, such as a directional coupler, which takes part of the signal directed to the over-the-air transmission medium 53 and sends it to the unit of measurement, estimation and control 80, which carries out the measurements and calculates the I / Q imbalance parameters necessary to estimate the calibration coefficients.

In particular, the radiofrequency modulated signal, coming from a first output of the antenna coupler 56, reaches a demodulation means, preferably an envelope detector 81 (AMDET), which calculates the envelope of this signal.

The antenna coupler 56 also has a second output feeding the over-the-air transmission medium 53, which can be an antenna or, in the case of particularly high frequency transmission, a wave guide or other equivalent device.

The output of the demodulation medium is connected to a Low Pass Filter (LPF) 82, which is necessary to remove unwanted spectral components.

The signal at the output of filter 82 is converted by the analog-to-digital converter (ADC) 83, which operates at the sampling frequency fsâ € ™ = 1 / Tsâ € ™ generated by a means for generating clock 84.

The output of the analog-to-digital converter (ADC) 83 is therefore connected to a data processing unit 85, which estimates the I / Q imbalance parameters and calculates a plurality of calibration coefficients, which they will be better described hereinafter in the present description, finally providing for their advantageous application to the filter system 67-70 and to the summing means 73 and 74.

The I / Q imbalance compensation method implemented by the electronic compensation system 50 just described, like the carrier residual compensation method, is of the open-loop offline type and includes an inconsistent measurement phase of the signal power. test 502, 504, 506, 510, 703, 712, a phase of estimating the imbalance parameters 507, 706, a phase of calculating the compensation parameters 508, 709 and applying 512, 513, 719 of these parameters to the compensation 60 which acts on the digital signals to be transmitted.

Advantageously, the compensation method according to the present invention comprises both a carrier residue compensation procedure, illustrated in the flow diagram presented in Figure 5, and an I / Q imbalance compensation procedure, illustrated in Figure 7.

Referring in the first instance to the carrier residue compensation procedure, initially (block 500) the selection means 61, 62 are in position 1 and the selection means 63, 64, 65 and 66 are in position 2, thus deactivating the compensator device 60.

In this way, the test signals zI (k) and zQ (k) coming from the means for the generation of numerical signals 52, driven by the measurement, estimation and control unit 80, are applied directly to the inputs of the transmitter device 51 .

Initially, the means for generating digital signals 52 provides two null signals zI (k) and zQ (k) (block 501), ie zI (k) = 0 and zQ (k) = 0; then subsequently (block 502) a measurement step of the spectrum S (f) of the signal s (t) is carried out downstream of the analog-digital converter 83, having only the direct component and that is S (f = 0).

In ideal conditions and therefore in the absence of carrier residue, we have that the power p1 of the signal s (t) is zero, that is p1 = 0; on the other hand, in the non-ideal case, the carrier residue appears and the power p1 of the signal s (t) can be considered as due to a complex signal d constant, of real part dI and imaginary dQ, that is of module | d | and phase Ï † d to be determined:

d = dI + jdQ = | d | cosÏ † d + j | d | sinÏ † d.

Considering the unknown gain µ due to the measurement chain, we have that p <2 2 2 2>

1 = µ <2> (dI + dQ) = µ d.

Subsequently (block 503), always considering the initial position of the selector means 61, 62, 63, 64, 65 and 66, the means for the generation of I / Q test numerical signals 52, provides a complex and arbitrary signal having only the real positive component aI (block 503): a = aI + j0 or zI (k) = aI and zQ (k) = 0 with aI> 0. At this point (block 504) again a measurement of the direct component of the signal power s (t), indicated by p2, is carried out.

Therefore (block 505) another arbitrary complex signal b of modulo | b | â ‰ | a | having only the real positive component bI, that is zI (k) = bi and zQ (k) = 0, with bI> 0 and the corresponding continuous component of the signal power s (t) is measured, indicated with p3 (block 506).

Starting from the three collected measures p1, p2 and p3 it is possible to solve the following system of three equations in three unknowns:

obtaining (block 507) in closed algebraic form the estimate of the measurement gain µ, of the module | d | and of the phase Ï † d of the carrier residue:

Thanks to this estimate, the processing unit calculates (block 508) an estimate of the carrier residue, in its components in the phase of quadrature dQ:

Therefore the unit of measurement, estimation and control of the compensation parameters 80 controls the means for the generation of numerical signals 52 (block 509) with a constant complex signal ∠'d <Ë †> equal and opposite to the estimated value for d , that is zI (k) = âˆ'd <Ë †> I and zQ (k) = âˆ'd <Ë †> Q, and (block 510) a further measurement p4 of the DC component of the signal power s ( t).

Then it is verified whether this new measure p4 is greater or less than p1 (block 511).

If p4 <p1, then the compensation is correct and the estimated value of d is valid and the compensation parameters are calculated (block 512):

cI = −d <Ë †> I

cQ = −d <Ë †> Q

otherwise it is necessary to change (block 513) the sign to the value obtained for the estimate of dQ.

This is due to the ambiguity of the sign of the arc cosine function (that is cos <-1>) used for the estimate of Ï † d which therefore becomes:

from which the new values of cI and cQ are obtained (block 513):

cI = −d <Ë †> I

cQ = d <Ë †> Q

Remember that the arccosine function returns values in the range [0 pi] while the arcsine function returns values in the range [-pi / 2 pi / 2]. In a possible implementation of the method just described, the estimated values for cI and cQ can be advantageously tabulated in the processing unit 80 as a function of the output power of the transmitter device 51, of the carrier frequency, of the temperature and of the power supply voltage of the device; in this way, the estimated values can be used to compensate for the carrier residue during the operating phase of the transmitter device 51, especially if the transmitter device 51 can operate on several frequency ranges and if it can be adjusted in the output power.

It should in fact be pointed out that the carrier residue depends both on the carrier frequency and on the transmission power, as well as on the temperature and on the supply voltage.

According to the present invention, after the carrier residual compensation procedure, an I / Q imbalance compensation procedure is provided, described in the block diagram of figure 7.

This procedure comprises a generation phase (block 702) of four test I / Q signals, whose phase and quadrature components correspond to simple sinusoids at a given frequency fx, included within the operating band of the transmitter device 51 under test.

This generation step takes place with the selection means 61 and 62 in position 1 and therefore in the calibration configuration and with the compensator device 60 deactivated, that is, with the selection means 63, 64, 65, 66 in position 2 (block 700).

Therefore, for each test signal, the corresponding demodulated signal s (t) downstream of the low-pass filter 82 is acquired in numerical form via the ADC 83.

Subsequently, the data processing unit 85 calculates the modulus of the Fourier Transform of s (n) at the frequency 2fx (block 703).

In a possible implementation of the method, the sampling frequency of the analog-to-digital converter 83 is fixed at 4fx, Ncc samples (Ncpari) are acquired, grouped in the vector s, whose index element k is s (k).

The modulus of the spectrum S (f) at frequency 2fx is then calculated (block 703) by applying the Fourier transform:

In another possible implementation of the invention, if the analog to digital converter operates at a fixed sampling frequency fsâ € ™, independent of the current value of fx, the value of the Fourier Transform at frequency 2fx is calculated as:

The 4 test I / Q signals to be applied in input to the compensator device 60 are indicated in figure 6. The figure shows the number N of samples for the I and Q components of the test signals to be generated.

Since four consecutive measurements Sm (f = 2fx) are carried out for m = 1, â € ¦, 4 in correspondence to the four test signals, it is assumed that in each of them the measurement network introduces a constant µ gain. Furthermore, since the measurements are inconsistent, i.e. there is no synchronization between the measurement, estimation and control unit 80 and the transmitter device 51, in each of them it is assumed that the acquisition start instant by of the DAC is random.

The compensation procedure described below is independent of both the measurement gain (assumed to be substantially constant) and the acquisition start instant, as it operates on the power of the signals.

The amplitude and phase unbalance of oscillator 16 and 35 are defined, indicated with g and Ï †: they represent the ratio of the modules and the difference of the phases of the modulating sinusoids applied to multipliers 18 and 19, in the transmitter device 10 , or 37 and 38, in the receiver device 30.

Furthermore, let "be the difference in the signal propagation delay between the in-phase and quadrature channels inside the transmitter device 10 or in the receiver 30, which is the cause of the band-dependent I / Q imbalance.

It is important to underline that this is an equivalent delay due to the entire transmission chain, both to the components operating in the baseband and to those operating at radio frequency.

It is possible to give the following formulation of the direct problem that links the parameters just defined with the amplitudes of the power spectral lines of s (t) in the case of test signals 1, 2, 3 and 4:

The direct problem is solved by calculating the solutions of the system:

ï £ ³S1 (f 2fx)

The analytical solution can be calculated immediately for the parameter g as:

we obtain the following second degree equation: Î ± cos <2> (4pfi xÏ „) + βcos (4pifxÏ„) + γ = 0

where

g

We obtain (block 706) the estimate of the delay Ï „and of the phase imbalance Ï † which are respectively:

The delay estimate Ï „is ambiguous, as it admits 4 possible solutions Ï„ icon i = 1, â € ¦, 4 depending on the signs / - of the terms appearing in the second degree solving equation and due to the fact that the arccosine function is even.

This means that there are four possible solutions <Ë †> Ï „i, <Ë †> Ï † i (with i = 1, â € ¦, 4) to be verified individually, and then choose the one that provides the best performance, therefore the one which produces a higher IRR. It should also be emphasized that any carrier residue does not alter the values of the spectrum S (f = 2fx) at all, being independent of the I / Q imbalance. After having activated (block 707) the compensator device 60, by placing the second and third pair of selector means 63, 64 and 65, 66 in position 1, the first of the possible solutions for estimating the delay Ï "and phase Ï is selected †, that is <Ë †> Ï „1, <Ë †> Ï † 1 (block 708).

Then one continues generating (block 709) a compensation filter hD (k) and from this the compensation filters 67, 68, 69 and 70.

This hD (k) compensation filter has as impulse response the convolution of the impulse responses of two filters hD1 (k) and hD2 (k):

hD (k) = hD1 (k) ⊠— hD2 (k)

in which hD1 (k) is an amplitude equalizer filter, preferably FIR (Finite Impulse Response), with linear phase, of length L1 samples; hD2 (k) is instead a phase equalizer filter while ⊗ indicates the convolution operation.

In a possible implementation of the present description, it is assumed that the filter hD1 (k) is impulsive with unitary amplitude while the filter hD2 (k) is a delay filter realized as an interpolator.

As regards the generation of the interpolator filter hD2 (k), it is preferably of the FIR (Finite Impulse Response) type with fixed length L2dispari; typical values range from L2 = 5 to L2 = 15 but other values are also possible, both odd and even. The overall filter hD (k), preferably of the FIR (Finite Impulse Response) type, has length L = L1 + L2â € “1.

Of the system of four filters 67-70, a first and a second filter 67 and 68 are realized as means that introduce a delay equal to R = (L-1) / 2 samples (R integer), with equal gain, respectively, a 1 and tan <Ë †> Ï †.

It is important to underline that the delay R is necessary for the feasibility of the filter system, ie for the compensation in real time, as it makes the response to the impulse of all the filters causal.

These filters 67 and 68 therefore have an impulse response equal to:

h11 (k) = Î ́ (k ∠’R)

h12 (k) = tan <Ë †> Ï † Î ́ (k ∠’R)

For symmetry reasons, a third filter 69 of the four-filter system 67-70 has been assumed, preferably null:

h21 (k) = 0

According to the invention, an interpolating FIR filter is introduced, with the following impulse response: hD2 (k, R2, Ï „) Ë † = hs (k, R2, Ï„) Ë † w (k , R2, Ï „) Ë †

where hs (k, R2, Ï „) Ë † is an ideal interpolator (cardinal sine), with R2 integer and fsÏ„ Ë † fractional, operating at sampling frequency fs:

sin [pi (k âˆ'R2 f

= s <Ë †> Ï „)]

h (<s> k, R2, Ï „) Ë †

pi (k −R2 fsË † Ï ")

The integer delay introduced by this filter is equal to R2 = (L2â € “1) / 2 samples.

The filter w (k, R2, Ï „) Ë † is preferably but not necessarily a raised cosine window:

I <ï £ «> 2 (m −1) <pi> (k −R2 fs <Ï„ Ë †>) ï £ ¶

w (k, R 2, Ï ") Ë † = ∑C <m> so £ ¬ï £ ¬m = 1 Lï £ · ï £ ·

ï £ 2ï £ ̧

or any other window that has the purpose of limiting the temporal duration of the interpolator, whose response to the impulse is very long being that of a cardinal sine.

Optimal values for window design are known from the literature and correspond to: C1 = 0.375, C2 = 0.5, C3 = 0.125 with I = 3.

The fourth filter 70 of the system of four filters 67-70 introduces a group delay equal to R −fsÏ ", therefore, compared to the first and second filters 67, 68, it anticipates by fsÏ".

It is clear that, depending on the estimated sign for Ï ", the concepts of advance and delay can be interchanged.

It is possible to use, in another implementation of the method proposed by the present invention, any interpolator with an integer delay equal to R and a fractional delay equal to âˆ'fsÏ „= ∠'Ï„ Ts or with an overall group delay equal to R âˆ'fsÏ ".

For example, in the case of a raised cosine interpolator, the following impulse response occurs:

k âˆ'R T h (k, R, Ï „) Ë † = 1sin (2pi (2 fs <Ë †> Ï„) Tsfn0) cos (2pi (k âˆ'R2 fs <Ë †> Ï „) sfn0) < D> 2 2

F <2> s2pi (k âˆ'R2 fsË † Ï „) Tsfn0 1∠'(4Î ± (k âˆ'R2 + fs <Ë †> Ï„) Tsfn0) where k is the index of the sample, Ts = 1 / fs the sampling time, Î ± the roll-off factor and fn0 the filter cutoff frequency at -6 dB. The latter must be chosen so as not to cut the spectrum of the signal, therefore it is necessary that the sampling frequency is higher (with adequate margin) than double the maximum frequency of the signal.

The fourth filter 70 has response to the impulse hD (k) and is 1

<scaled by factor>. We get gË † cosË † Ï †

therefore:

h (k <,> R <, Ë †> Ï „)

h22 (k) = <D>

g <Ë †> cos <Ë †> Ï †

As indicated in figure 4, therefore, the compensation is carried out by a compensation means with a square matrix H (k) of size 2 x 2 composed of filters h11 (k), h12 (k), h21 (k) and h22 ( k):

ï £ «h11 (k) h 12 (k) ï £ ¶

H (k) <=> ï £ ¬ï £ ¬ ï £

ï £ h21 (k) h 22 (k) ï £

ï £ ̧

Once the four filters 67, 68, 69, 70 have been generated, the 4 test I / Q signals are applied again (block 711), with the compensator device 60 activated, i.e. keeping the selection means 61, 62, in position 1 and placing the selection means 63, 64, 65, 66 in position 1.

The demodulated numerical signals are acquired (block 712) and the 4 updated values of | Sm (f = 2fx) | for m = 1, â € ¦, 4.

The following ratios are then evaluated (block 715):

2

S3 (f = 2fx)

B =

S1 (f = 2fx)

And

2

S4 (f = 2fx)

C =

S1 (f = 2fx)

They should ideally be brought to zero (or to a very small value, typically <10 <-6> for an IRR> 60 dB).

If this is not the case, it is necessary to analyze (block 709) one by one all the other possible solutions for the delay Ï „Ë †, because of the 4 possible ones, only one removes the I / Q imbalance. This uncertainty in the value of Ï „Ë † is due to the use of only inconsistent power measurements, which simplify the realization of the unit of measurement, estimate and control of the compensation parameters 80, but do not guarantee a univocal solution.

If the values of B and C are both lower than the preset threshold, the coefficients of the compensation filters, calculated in correspondence with the determined value of the index i for the parameters <Ë †> Ï „i, <Ë †> Ï † i , can be stored (block 719) in the device 60 or in the estimation and control unit 85.

If, after trying all the values of Ï „Ë † it is not possible to adequately reduce the unbalance (block 717), it is necessary to modify (block 718) the fx frequency or the required IRR threshold and repeat the procedure from the start until a new estimate of the I / Q imbalance parameters indicated above is reached, such as to respect the threshold required on the B and C ratios.

If, after further attempts, it is not possible to reduce the I / Q unbalance to the desired value, it may be necessary to perform a set of several measurements at various fx frequencies and then solve a non-linear system.

Typically, this happens if the imbalance parameters are very small, especially if | 1 -g | <10 <-3> and | Ï † | <10 <-3> radians and, therefore, the measured power values are highly affected by noise.

In another possible implementation, in fact, more values of the test frequency fx are used, in order to have a wider set of measurements.

In this case the test signals in figure 6 are applied for several values of the fx frequency, chosen, for example, with a preferably constant step within the operating band of the transmitter device to be calibrated or considering a non-uniform sampling of the same band, therefore denser where the image rejection of the transmitter device is higher.

Said single frequency fxpla used, if Ntfrequencies belonging to the operating band of the device under test are used, then p = 1, 2, â € ¦, Nt.

We then perform a set of 4Ntmeasures | Sm (f = 2fxp) | with m = 1, â € ¦, 4 and fxp belonging to the operating band of the device.

These measurements can be used to perform a Nonlinear Least Squares (NLS) inversion of the following system of equations:

Thus we have a non-linear system of 3Ntequations in 3 unknowns, which can be solved for example with the well-known Levenberg-Marquardt or Gauss-Newton algorithms.

It is possible to choose the number of equations in order to reach the desired parameter estimation error.

To ensure correct convergence, it is also advisable to take the logarithms of the equations in the non-linear system as the values of the ratios:

they are very small (typically <10 <-4> for an IRR> 40 dB). In fact, if the rejection of the image tends to infinity, both terms B and C tend to zero. The non-linear system, having expressed the quantities in logarithmic scale, becomes:

According to a second preferred embodiment, the system for the compensation of the I / Q imbalance in phase and quadrature transceivers can be realized with bench instrumentation, particularly used in the industrial production / testing phase.

As illustrated in Figure 8, in this case an electronic system 90 for the compensation of the I / Q imbalance differs from the electronic system 50 previously described in that it comprises a unit for measuring, estimating and controlling the compensation parameters 91 preferably composed from a spectrum analyzer 92 and a processing unit 93.

The presence of the spectrum analyzer 92 avoids the use of the envelope detector 81, the low-pass filter 82 and the analog-digital converter 83, as the power measurements are performed by the spectrum analyzer 92.

It is possible to advantageously remove both the first pair of selector means 61 and 62, by applying the I / Q signal generator directly to the selector means 63 and 64, and the antenna coupler 56 and the over-the-air transmission medium 53, by connecting the spectrum analyzer 92 directly to the output of the transmitter device under test 51, in order to perform a bench calibration in the production / industrial test phase.

In an alternative embodiment it is also possible to remove in addition the other pairs of selector means 63, 64, 65 and 66, by directly connecting block 140 to the I / Q input and the adders 73, 74 to the IPe QP outputs.

In this case, there is a preventive phase in which the compensating device is deactivated by setting h11 (k) = h22 (k) = 1, h12 (k) = h21 (k) = 0, cI = 0 and cQ = 0; then the coefficients for the compensation of the I / Q unbalance and the carrier residue are estimated. In the case of estimation and compensation of the I / Q imbalance, we then proceed with an iterative phase in which the four possible solutions for the compensation coefficients are applied sequentially to the system of numerical filters 67-70; similarly, for the procedure for estimating the carrier residue, the two possible solutions for the compensation coefficient cQ are tested.

The coefficients of the compensation filters, in correspondence with which the I / Q imbalance and the carrier residue are rejection, can then preferably be stored directly in the compensator device 60 to be subsequently read from the memory of this device during the operating phase. .

As illustrated in Figure 9, the bench calibration procedure, similarly to before, comprises a step in which the test signals of Figure 6 are generated and applied (block 902) to the transmitter device 51 at a given frequency fx.

This generation step provides that the compensator device 60 is deactivated (block 900).

The numerical test signals, therefore, are directly applied to the inputs of the transmitter device 51, while at the output the radiofrequency spectrum is measured which contains, among others, the three spectral lines at frequency fc + fx, fc-fx and fc. Using the spectrum analyzer 92, the powers of the right side band (Upper Side Band) PU, m (fx) = Pm (fc + fx) and left (Lower Side Band) PL, m (fx) are measured (block 903) ) = Pm (fc- fx) corresponding to the four test I / Q signals with m = 1, â € ¦, 4.

Considering the test I / Q signals 3 and 4 (i.e. respectively for m = 3 and m = 4) in figure 6, it is possible to directly measure the left and right Image Rejection Ratio:

At this point the analytical solutions for the I / Q imbalance parameters are calculated (block 906).

In fact, the measured powers can be expressed as a function of the imbalance parameters, obtaining a system of second degree equations:

Also in this case, thanks to the use of power measurements, no device synchronization procedure is necessary, guaranteeing a considerable simplification of the implementation of the measurement, estimation and control unit 91. For a generic frequency fx, the analytical solutions are available:

where the values of x and y are calculated from the power measurements as indicated below:

As it can be noticed, there are some arccosine functions that imply an ambiguity in the signs of the value of Ï „Ë †; this means that there are four possible solutions <Ë †> Ï „i, <Ë †> Ï † i (with i = 1, â € ¦, 4) to be verified individually, and then choose the one that provides the best performance, i.e. which minimizes both ratios:

This procedure allows you to maximize both the left and right IRR. For the single solutions <Ë †> Ï „i, <Ë †> Ï † ii = 1, â € ¦, 4 see also block 906.

The compensator device 60 (block 907) is activated by operating on the selection means 63-66, and the first (block 908) of the possible solutions is therefore considered, generating (block 909) the calibration filters 67, 68, 69 and 70.

The two test I / Q signals 3 and 4 listed in figure 6 are then applied (block 911), measuring the powers in the right and left side bands (block 912) using the spectrum analyzer 92 and, from these, calculating (block 915) the corresponding rejections of the right and left IRR image.

Then, then, it is evaluated (block 916) if the values of the IRR + and IRR-, calculated in correspondence with the parameters <Ë †> Ï „i, <Ë †> Ï † i, exceed the predetermined threshold; if so, the procedure is completed and the compensation coefficients, calculated in correspondence with the determined value of the index i for the parameters <Ë †> Ï „i, <Ë †> Ï † i, can be stored (block 919) in the compensator device 60 or in the estimation and control unit 93, otherwise, one proceeds (block 917) by iterating all the remaining combinations for the estimated values of delay and phase imbalance <Ë †> Ï „i, <Ë † > Ï † i.

If none of the solutions <Ë †> Ï „i, <Ë †> Ï † is satisfied the image spectrum rejection conditions, the frequency in the baseband fxo is changed (block 920), possibly the threshold required for the IRR values + and IRR-, repeating the calibration procedure from the beginning (block 900).

In the bench calibration configuration it is also possible to calculate the correction factors for the compensation of the carrier residue as already indicated in figure 5.

In a possible implementation of the invention, the value of the carrier frequency fc can be varied and the estimated imbalance parameters tabulated inside a memory of the estimation and control unit 93, or directly in the compensator device 60, in function of fc, to be used during the operational phase.

In a further implementation of the invention, similarly to what was previously described, it is possible to repeat the measurement procedure for more values of fx, obtaining a frequency characterization of the I / Q imbalance of the device, and estimate the parameters g, Ï † and Ï ”with an NLS algorithm.

In a possible implementation of the methods just described, the estimated values for the I / Q imbalance parameters can be advantageously tabulated in the processing unit 80 (91) as a function of the output power of the transmitter device 51, of the carrier frequency, the temperature and the power supply voltage of the device; in this way, the estimated values can be used to compensate for the I / Q imbalance during the operating phase of the transmitter device 51, especially if the transmitter device 51 can operate on multiple frequency bands and if it can be adjusted in output power. Alternatively, the estimated values of the unbalancing parameters can be stored directly in the compensating device 60.

As previously said, the present invention aims to compensate for the I / Q imbalance in a transceiver both in transmission and reception. As illustrated in Figure 10, a first embodiment of an electronic system for the compensation of a broadband I / Q imbalance for the receiver device 30 according to the invention indicated as a whole with 100 can be considered.

This electronic system includes a means for generating I / Q signals 52, a calibrated transmitter device 101 driven by a means for generating a local oscillation fc55 (local oscillator) and a unit for measuring, estimating and controlling the parameters compensation 105, which, like the corresponding unit described above with regard to the transmitter device 51, performs the procedure for estimating the compensation parameters of the in-phase and quadrature channel of the receiver device in an iterative manner.

The electronic system for the compensation of a broadband I / Q imbalance is applied to a receiver device 103 under test, electrically connected with the means for generating a local oscillation fc58 for analog demodulation and a means 59 for generating of the period clock Ts which times the analog-to-digital conversion in ADCs 43 and 44.

The means for generating local oscillations 55 and 58 need not be synchronized with each other. The same consideration also applies to the means for generating the clock 57 and 59. Preferably there is a switching means 106 between the receiver device 103 and the calibrated transmitter device 101; this switching means 106 is equipped with a pair of positions (position 1 and position 2) and allows to select the input coming from this calibrated transmitter device 101 (position 2) and from an over-the-air receiving means 102 (position 1 ), passing respectively from a calibration configuration (position 2) to an operational configuration (position 1).

This calibrated transmitter device 101 must not generate I / Q imbalance, so it can be compensated as described previously.

The electronic system for the compensation of the broadband I / Q imbalance comprises a numerical compensator device 110 comprising two selector means 111 and 112 powered respectively by the output of the in-phase baseband channel and by the output of the baseband channel quadrature of the receiver 103.

Said selection means 111, 112 are selectively connected to two summing means 113 and 114 (position 1) for the compensation of the carrier residue, or to a pair of selection means 115 and 116 (position 2); the compensator device 110, therefore, is activated or deactivated according to whether the selection means 111, 112, 115 and 116 are respectively in position 1 or in position 2.

These summing means 113 and 114 are in turn connected to a system of four filters 117, 118, 119, 120 of the numerical type, preferably of the FIR type.

In particular, the output signal to the first half adder 113 feeds the inputs of the first and third numerical filters 117, 119 respectively, while the output signal to the second half adder 114 feeds the inputs of the second and fourth numerical filters 118, 120 respectively. .

The outputs of the four numerical filters 117, 118, 119, 120 are then associated two by two and added together at the input of two adding means 121 and 122.

In detail, the first filter 117 and the second filter 118 have a respective outlet placed at the inlet of the first adder means 121, while the third and fourth filters 119, 120 have a respective outlet placed in the inlet of the second adder means 122.

It is important to underline that, as in the case of the transmission I / Q imbalance compensation system, in reception the four filters 117, 118, 119, 120 implement a square matrix compensation means 150.

However, the components of such square matrix compensation means 150 are transposed with respect to the components of the square matrix compensation means 140 described above in the transmission system.

The output of the compensator device 110 is advantageously connected to the estimation and control unit 105, which performs the estimation of the parameters for the compensator device 110, drives the means for the generation of digital signals 52 so as to produce suitable signals test I / Q and controls the selection means 106, 111, 112, 115, 116.

In a possible implementation of the method, the means 52 and 101 can be realized by means of a single device for generating radio frequency test signals piloted with signals I and Q through the unit 105.

The estimation of the imbalance parameters can be performed directly starting from the power spectrum of the complex signal v (k) = vI (k) jvQ (k), with the compensation device 110 deactivated, considering the selection means 111, 112, 115 , 116 in position 2.

Similarly to what previously described, the test signals listed in Figure 6 are applied, noting that the complex signal in the baseband contains three spectral components at frequency - fx, 0 and fx.

The zero frequency component depends only on the carrier residue, therefore it is possible to directly apply the compensation technique already described and illustrated in figure 5.

As regards the I / Q imbalance, follow the procedure already described and illustrated in figure 9.

In this case the powers PU, m (fx) and PL, m (fx) are obtained in the estimation and control unit 105, calculating the square modulus of the Fourier Transform of the complex signal v (k) = vI (k) jvQ (k), and evaluating the frequency components - fx (Lower Side Band, PL) and fx (Upper Side Band, PU) in correspondence to the test signals.

In a possible implementation of the present receiving compensation technique, if the sampling frequency fs = 1 / Ts of the receiver device 30 (103) cannot be adjusted, the powers can be calculated with the following expressions:

f 2

Nc −1 −j2 pi <x> k

f

PU, m <(> fx <)> = ∑v <(> k <)> and s

k = 0

2

Nc âˆ'1j2 pi f <x> k

PL, m <(> fx <)> = f

v <(> k and s

k ∠‘<)>

= 0

where N is the number of samples acquired in the baseband. In another possible implementation of the present invention, if the sampling frequency of the demodulator is adjustable, it is advisable to choose fs = 4fx, in order to simplify the calculation of the Transforms of Fourier, as in the expressions:

2

N 2

c −1−j pi k <N> c âˆ'1

PU, m (fx) =

k∑v <(> k <)> and 2 =

= 0 k ∠‘k

v <(> k <)> (∠’j)

= 0

N 2

c âˆ'1 + j pi 2

k <N> c −1

PL, m (fx) = ∑v <(> k <)> and 2 =

= 0 ∠‘k

v <(> k <)> (j)

k k = 0

The calibration procedure follows exactly the description in figure 9.

The only difference in the procedure lies in the expressions of the terms x and y:

There are four possible solutions <Ë †> Ï „i, <Ë †> Ï † i (i = 1, â € ¦, 4) to estimate the delay Ï„ and phase imbalance Ï †; therefore the estimation procedure must be repeated until an adequate rejection of the I / Q imbalance is obtained by identifying the correct combination of the solutions, exactly as previously described.

In a possible implementation of the invention, the carrier frequency in the transmitter device 10 and in the receiver device 30 is varied and the estimation procedures are repeated for several values of fc, tabulating the results in the unit of measurement, estimation and control. 105, or in the compensating device 110, so as to be used in the subsequent operating step.

A second embodiment of a system for the compensation of the I / Q imbalance in reception is illustrated in figure 11 and is an electronic system for the compensation of the I / Q countertop broadband unbalance indicated as a whole with 160.

Unlike what has been seen previously, the compensation parameter estimation and control unit 130 advantageously comprises a spectrum analyzer 132 and a processing and control unit 131. The characteristics of the method are clear from the description. and of the electronic system object of the present invention, as well as the relative advantages are clear.

In fact, thanks to the implementation of the filters it is possible to obtain an imbalance compensation no longer at a single frequency but in a wide range of frequencies.

Furthermore, the opportunity to carry out inconsistent power measurements of the modulated signal by means of a spectrum analyzer allows the electronic compensation system according to the present invention to be used even in the industrial production phase.

Finally, it is clear that the method and the electronic system thus conceived are susceptible to numerous modifications and variations, which are obvious to a person skilled in the art, without thereby departing from the scope of protection provided by the attached claims; for example the FIR filters, described in the present description, can for example be replaced by equivalent IIR filters (filters with infinite impulse response).

Claims (14)

CLAIMS 1) Broadband compensation device (60, 110) of an imbalance between a channel in phase and a quadrature channel of a transmitter device (51) and / or a receiver device (103) having a respective passband, said device compensation (60, 110) comprising: - at least one input fed by at least one transmission signal in phase baseband (I) and at least one transmission signal in quadrature baseband (Q); - a square matrix compensation means (140, 150) capable of processing said at least one transmission signal in phase baseband (I) and said at least one transmission signal in quadrature baseband (Q); characterized in that the components of said square matrix compensation means (140, 150) are numerical filters (67-70; 117-120) configured to compensate for said imbalance in a frequency range included in said passband of said transmitter device (51) and said receiver device (103). 2) Broadband compensation device (60, 110) of an unbalance between a phase channel and a quadrature channel of a transmitter device (51) and / or a receiver device (103) according to claim 1, characterized by the fact that said compensation means (140, 150) with a square matrix of numerical filters (67-70; 117-120) comprises at least one numerical filter (70, 120) resulting from a combination of an amplitude equalizer and a phase equalizer . 3) Broadband compensation device (60, 110) of an unbalance between a phase channel and a quadrature channel of a transmitter device (51) and / or a receiver device (103) according to claim 2, characterized by the fact that said compensation means (140, 150) with a square matrix of numerical filters (67-70; 117-120) further comprises a first numerical filter (67, 117) retarder, a second numerical filter (68, 119) of the retarder type and a third numerical filter (69, 118) null. 4) Compensation device (60, 110) according to claim 1 comprising a first calibration configuration for estimating said unbalance of said channels in phase and in quadrature and a second operating configuration for transmitting and / or receiving transmission signals respectively by said transmitter device (51) and / or receiver device (103). 5) Broadband compensation device (60, 110) of an imbalance between a channel in phase and a quadrature channel of a transmitter device and / or a receiver device (51, 103) according to one of the preceding claims, characterized by the fact to comprise a first pair of selecting means (63, 64, 111, 112) and a second pair of selecting means (65, 66, 115, 116), said first pair of selecting means (63, 64, 111, 112) being selectively connected to said second pair of selection means (65, 66, 115, 116) or to said square matrix compensation means (140, 150) of numerical filters (67-70; 117-120). 6) Broadband compensation device (60, 110) of an imbalance between a channel in phase and a quadrature channel of a transmitter device (51) and / or a receiver device (103) according to claim 1, characterized by the fact that said numerical filters (67-70; 117-120) comprise respective two-by-two outputs feeding a first half-adder (71, 121) and a second half-adder (72, 122), the outputs of a first filter (67, 117) and a second filter (68, 118) being placed in the input of said first half adder (71, 121), the outputs of a third filter (69, 119) and fourth filter (70, 120) being placed in the input of said second half adder (72, 122). 7) Broadband compensation device (60) for an unbalance between a channel in phase and a quadrature channel of a transmitter device (51) and / or a receiver device (103) according to claim 6, characterized by the fact that it comprises a first pair of adding means (73, 74), having respective inputs fed by said first adding means (71, 72) and by constants for compensation of a carrier residue. 8) Broadband compensation device (110) of an unbalance between a channel in phase and a quadrature channel of a transmitter device (51) and / or of a receiver device (103) according to claim 6, characterized by the fact that additionally comprising a second pair of summing means (113, 114), adapted to add two compensation constants of a carrier residue to two output signals from said receiver device (103). 9) Electronic system for broadband compensation (50, 90, 100, 160) of an imbalance between a channel in phase and a quadrature channel of a transmitter device and / or a receiver device (51, 103), comprising a means for the generation of digital signals (52) and a unit of measurement, estimation and control of the compensation parameters (80, 91, 105, 130) of said channel in phase and of said quadrature channel, characterized by the fact of comprising a compensating device (60, 110) according to any one of claims 1-9. 10) Electronic system for broadband compensation (50, 90, 100, 160) of an unbalance between a channel in phase and a quadrature channel of a transmitter device and / or a receiver device (51, 103) according to claim 9 characterized in that said compensating device (60, 110) comprises selection means (61, 62, 106), which switch said compensating device (60) between said first calibration configuration, in which said compensating device (60) is powered by two test signals (zI (k), zQ (k)) generated by said means for the generation of two digital signals (52), and said second operating configuration, in which said compensating device (60) is powered by two numerical signals (I, Q) in the baseband. 11) Electronic system for broadband compensation (50, 90, 100, 160) of an unbalance between a channel in phase and a quadrature channel of a transmitter device and / or a receiver device (51, 103) according to claim 9 characterized in that said compensating device (60, 110) comprises selection means (61, 62, 106), which switch said compensating device (110) between said first calibration configuration and said second operating configuration, in said first calibration configuration said receiver device (103) being powered by a signal coming from a calibrated transmitter (101), in said second operating configuration, said compensating device (110) being powered by a signal coming from an over-the-air receiving means (102). 12) Electronic system for broadband compensation (50, 90, 100, 160) of an imbalance between a channel in phase and a quadrature channel of a transmitter device and / or a receiver device (51, 103) according to one of the claims 9-11 characterized in that said unit of measurement, estimation and control of the compensation parameters (80, 91, 105, 130) comprises a spectrum analyzer (92, 132). 13) Electronic system for broadband compensation (50, 90, 100, 160) of an imbalance between a channel in phase and a quadrature channel of a transmitter device and / or a receiver device (51, 103) according to one of the claims 9-12 characterized in that it comprises a coupler means (56), connected at the output to a means for over-the-air transmission (53) and to said unit for measuring, estimating and controlling the compensation parameters (80, 91, 105, 130). 14) Electronic system for broadband compensation (50, 90, 100, 160) of an imbalance between a channel in phase and a quadrature channel of a transmitter device and / or a receiver device (51, 103) according to one of the claims 9-13 characterized in that said unit for measuring, estimating and controlling the compensation parameters (80, 91, 105, 130) drives and controls said square matrix compensation means (140, 150), said summing means (73 , 74; 113, 114) and said selection means (61-66; 106, 111, 112, 115, 116) 15) Method for the broadband compensation of an unbalance between a channel in phase and a channel in quadrature of a transmitter device (51) and / or of a receiver device (103) characterized by the fact of including the phases which consist in: - generating a plurality of test signals (zI (k), zQ (k)) and applying said test signals (zI (k), zQ (k)) to said transmitter device (51) and / or to said receiver device ( 103); - taking an output signal from said transmitter device (51) and / or said receiver device (103) and applying said signal to units of measurement, estimation and control of the compensation parameters (80, 91, 105, 130) of said channel in phase and of said channel in quadrature; - measure and estimate a plurality of parameters (g, Ï „, Ï †) of said unbalance between a channel in phase and a quadrature channel through said unit of measurement, estimation and control of the compensation parameters (80, 91, 105, 130); - compensating for said unbalance between a channel in phase and a channel in quadrature by means of a plurality of numerical filters (67-70; 117-120), operating as a function of said parameters. 16) Method for broadband compensation of an unbalance between an in-phase channel and a quadrature channel of a transmitter device (51) and / or of a receiver device (103) according to claim 15 characterized by the fact that it additionally comprises the phases that consist of: - measure and estimate a carrier residue; - compensating for said carrier residue by means of a plurality of adding means (73, 74, 113, 114). 17) Method for broadband compensation of an unbalance between an in-phase channel and a quadrature channel of a transmitter device (51) and / or of a receiver device (103) according to claim 15 characterized by the fact that said phase measurement and estimation is recursive. 18) Method for broadband compensation of an unbalance between a channel in phase and a channel in quadrature of a transmitter device (51) and / or of a receiver device (103) according to claims 15 to 17 characterized by the fact that said measurement and estimation step is applied to a plurality of test signals at different frequencies included in an operating band of said transmitter device (51) and / or of a receiver device (103).