GB2484336A

GB2484336A - Receiver architecture with high rejection of out-of-band signals

Info

Publication number: GB2484336A
Application number: GB201016962A
Authority: GB
Inventors: Jan Crols
Original assignee: ANSEM NV
Current assignee: ANSEM NV
Priority date: 2010-10-08
Filing date: 2010-10-08
Publication date: 2012-04-11
Also published as: GB201016962D0

Abstract

A quadrature receiver is disclosed, having a large rejection of signals out of the frequency band targeted for reception. The receiver architecture consists of a first quadrature down-conversion stage 17, a second down-conversion stage 22 using a double quadrature mixer, an optional filtering block 29, an analogue to digital converting block 32 and frequency independent digital IQ imbalance estimation 33 and correction 37. The digital IQ imbalance estimation block 33 may use the mutual stochastic independence of signals situated in different parts of the spectrum. Wide bandwidth signal processing is used at every point in the receiver where complex signal representation is used. This allows frequency independent correction by the digital IQ imbalance correction block 37. All analogue signal processing operations have bandwidths sufficiently wide to cause substantially no frequency dependency in the IQ imbalance observed at the I and Q outputs of the ADC block 32. This, along with the use of two stages of quadrature down conversion, is stated to provide the high rejection of interfering signals (e.g. mirror or image frequencies) by the receiver.

Description

Receiver architecture with high rejection of out-of-band signals

Field of the invention

The invention relates in general to an architecture for receivers in a communication system. More particularly, the presented invention relates to an architecture for receivers in a communication system which converts a signal contained in a frequency band used for transmission into a digital rcpresentation of this signal at a lower frequency than the original frequency band.

Background of the invention

in RE data communication, a signal which is in the frequency domain centered around a certain RE carrier frequency, is received by translating the signal in the frequency domain to lower frequencies.

This allows for technically easier and less power consuming digitalization compared to the situation where the digitalization is done directly on the RE data signal.

As the RE data signal is a real world signal, meaning that it is a continuous function of the time having only real values, its frequency domain representation will have the following property: s(r)= s(-ff (1) in which Sar(f) is the frequency domain representation of the RF signal and the "i" symbol is used for complex conjugate. The band-pass RE signal thcrcfore contains signal power both in a frequency band around f5 and -t5.

In a quadrature receiver either the frequency band around far or -fRF of the RE signal is selected and translated in the frequency domain to lower frequency. This is mathematically done by multiplying the RE signal with a single complex tone at a frequency near -far or far depending on the selected band.

To physically implement this purely mathematical operation of a multiplication with a complex valued signal, quadrature receivers use the concept of "complex signal representation". Two physical distinct signals arc in this representation viewed as one complex signal. One physical signal is viewed as the real part of the complex signal, the other as the imaginary part. Often the name "in-phase" is used in art for the real part of the complex signal, and "in-quadrature" for the imaginary part. Using this complex representation it is possible to perform complex valued linear operations on signals as explained in detail in t'ornplex signal processing is not complex ", Kenneth W Martin. iEEE Transactions on Circuits and Systems, Vol 51, No 9, December 2004. Possible examples of such complex valued linear operations are: band-pass filtering which only passes the negative frequencies of the signal; muhiplying the signal with a single complex tone.

To perform complex valued linear operations on complex signals, multiple physical operations are performed on the pairs of physical signals representing the complex signal and the resulting signals are combined in a specific way. As an example the multiplication of a complex signal x(t) with a complex signal y(t) can he expressed as: x(t)y(t)= (XR(t)+ixf(t)).(yR(e)+A'J(e)) = (XR(e) y5(t)-x1(t)y(t))+ i(x1(t)y5(t)+ x5Q)y1fr)) (2) in which "i" represents the imaginary unit, xR(t) and x(t) represent the real and imaginary part of x(t) respectively, y(t) and y(t) represent the real and imaginary part of y(t) respectively.

in practice the behavionr of a receiver, using the described concept of complex linear signal operations, will never perfectly comply with the mathematical constraints to have a pure linear complex operation.

There will always be small differences between the signal processing blocks used in the real and imaginary signal paths causing deviation from the expected behaviour, In the art the term "IQ imbalance" is often used to identify these imperfections. Possible effects causing TQ imbalance for a receiver integrated in a chip are random variation intrinsic to all semiconductor devices, small asymmetries in the layout, and differences in temperature between the signal processing blocks.

The invention provides a solution to two problems originating from the described imperfections which are always present in an analog physical implementation of complex signal processing stages.

The first problem is known in art as the "mirror frequency problem in down-conversion". In a quadrature down-conversion a certain frequency band is translated to a lower frequency, giving rise to a complex signal. This is done by multiplying the signal with a single complex tone of frequency fLO: = cos(2Ot)+ isin(20t) (3) This multiplication is implemented by niuhiplying the RE signal once with a cosine wave of frequency f0, giving rise to the real part of the output complex signal and once with a sine wave having the same amplitude and frequency, giving risc to the imaginary part of the output complex signal. In a physical implementation small imperfections causes both terms of equation (3) to have slightly different amplitude and a phase difference slightly deviating form 90°. The RE signal will in fact he multiplied by: E ( q I l+-JCOS,2fffLot +i)+i1 = etzouIcosM + i sin1 + If cosM i sinI ç 2) 2 2)) ç2 2) 2)) (4) in which c represents the relative amplitude error and tp the phase. error. An unwanted complex tone, the second term in the right hand side of the equation (4), appears at the negative frequency of the intended tone. The ratio between the magnitude of the intended tone and the unwanted tone is called "image rejection ratio" in the art. The image rejection ratio can be expressed as: 1R cosE9fl+ifsinCfl

-

2 K2) 2) (5) in which the,. . notation was used to denote the complex magnitude. A typical value in art for the achieved image rejection ratio of a quadrature down-conversion is 40 dO, Figure 1 schematically illustrates the effect of this unwanted tone. A RE band-pass signal 1 centered around frequency -fLo -f1 is multiplied both with the wanted complex tone 2 and the unwanted complex tone 3. The image rejection ratio, labelled "lR" and expressed in dO, is the difference in height between the two arrows representing the complex tones. Signal 4, located around -fEF. is the result of the multiplication of the RF signal 1 with the wanted complex tone 2. Signal 5, located at fLo -fjp is called in art the mirror signal. Signal 6 is the result of the multiplication of this mirror signal with the unwanted tone 3. Signal 6, which is in fact a copy of the mirror signal 5 attenuated by the image rejection ratio, will be added to the wanted signal 4. This indicates that, depending on the power level of the mirror signal, the mirror signal can cause a severe degradation of the received signal quality.

The second problem for which the invention provides a solution is the JQ imbalance in non-frequency-shifting complex linear operations like e.g. filtering. Figure 2.A shows a possible implementation of a real valued filtering of a complex signal. Filter 7 filters the in-phase component of the complex input signal and filterS filters the in-quadrature part of the complex input signal. To have the overall effect of a linear filtering on a complex signal, filter 7 and filter 8 should be identical. Due to imperfections in the physical realization of both filters, they will not be perfectly identical.

Figure 2.B presents a signal flow graph model of the in reality expected behaviour of the circuit of Figure LA. The two real signals 9 forming the complex input signal are each processed in a different way. The signals pass throngh the same nominal filter function 10, but an asymmetry caused by the error filter function 11 is added with a different sign to each channeL The equivalent complex signal representation of the performed operation by this signal flow graph is: OUT(f) = H([)1N(f)+ H(f)JN(-j our()= H1(f)+H(f) IN(f)÷ HJ(f)-HQ(f) IN(-fY 2 2 (6) in which IN(fJ and OUT(f) are the spectra of the complex input and output signal respectively. The magnitude of the ratio between 1-1(f) and H(f) is known in art as the "image rejection ratio". The implication of this operation is schematically illustrated in Figure 3. Figure 3.A shows the spectrum of the input signal 12, the nominal filter function 13 and the error filter function 14. The image rejection ratio is marked as "1W' and expressed in dB. The spectrum of the output signal, given in Figure 3.8, is the combination of the result of the nominal filtering 15 and of the result of the error filter function 14 applied on the input signal 12 mirrored around the vertical axis 16. This illustration clearly shows that, depending on the value of the image rejection ratio, this kind of JQ imbalance can cause a severe degradation of the quality of the output signal of the filter.

in practical communication systems out-of-band signals can be up to 100 dil larger than the targeted to-be-received signal. The finite image rejection ratio of the different down-conversion blocks can cause a severe degradation of the quality of the output signal of the receiver. A common approach in art to improve the quality of reception is the use of(digitai) calibration.

in a first category of calibration methods, some of the analog down-conversion blocks with too large IQ imbalance are made tuneable. in US673 1917 Method and apparatus for minimizing image power in the output of a receiver circuit" a digital feedback loop adjusts the phases of the clock signal used to do the second down-conversion, in US2006/0281432 "Radio frequency tuner" a method is presented that allows to increase the image rejection of the filters of a receiver chain by electronically adjusting them. This category has the drawback of increasing the complexity of the analog down-conversion blocks. Making these blocks tuneable will increase their size and possibly their current consumption.

in a second category of calibration methods, the digital output signals of a receiver are post processed in the digital domain to decrease the overall IQ imbalance of the receiver. Two conceptnal different groups can be identified: off-line methods using test tones and on-line blind methods.

in an off-line method, one or more test tones are injected into the receiver chain during a dedicated calibration phase. in this calibration phase the normal operation of the receiver is interrupted (the system is taken off-line). flased on the rcsnlting in-phase and in-quadrature digital output signal of the receiver, an estimate of the total IQ imbalance of the receiver is calculated. This estimate is used in the normal operatiou mode of the receiver to correct the digital output signals for the effect of the 1Q imbalance.

The total JQ imbalance of the receiver will in general have a frequency depending natui-e. in practical receivers reported in art aiming for the high rejecting of out-of-band signals, the used filters arc a main source of this frequency dependent nature, Off-line methods cope with this frequency dependency by doing consecutive measurements in which the frequency of the injected test tone is varied within the frequency range of interest, This leads to complex hardware to implement the measurement and the needed digital correction in the normal operation mode. it should be mentioned that an off-line calibration method with reduced hardware cost is presented in (1820050260949 I/O compensation of frequency dependent response mismatch in a pair of analog low-puss filters A major drawback of off-line calibration methods is that the output signals are con-ected based on an estimate of the iQ imbalance made in the past. A variation of the environment conditions of the receiver, like its temperature or the voltage of its supply, between the calibration phase and the particular moment the correction is done can cause the total 1Q imbalance to change. The correction of the digital output will therefore be done using an estimate that isn't longer accurate. To solve this problem the calibration phase should be repeated periodically. The higher the targeted re.jcction of out-of-band signals the shorter the allowed time between different calibration phases. Not all applications allow for the interruption of its normal operation on the time base as dictated by the needed rejection.

On-line methods calculate the estimate of the total IQ imbalance of the receiver during the normal operation of the receiver itself. No dedicated calibration phase is needed. A blind on-line method nses the statistical independence of signals at different frequencies to calculate an estimate for the total IQ imbalance. The targeted rejection of out-of-band signals after compensation dictates the needed accuracy of the estimate for the total IQ imbalance, This in turn is highly dependent on the statistical properties of the signals nsed to calculate the estimate.

Sununazy of the invention The invention consists of a receiver architecture enabling the reception of a RF band-pass signal with a high rejection of signals in other frequency bands. The block diagram of the receiver architecture is given in Figure 4. The receiver architecture consists of the following subblocks: 1. A first quadrature down-conversion stage 17 converting the incoming band-pass RF signal 48 centered around frequency fur to an in-phase 49 and in-quadrature 20 band-pass signal around the intermediate centre frequency fçr with the aid of a 00 and 90° phased shiflcd vcrsion of a oscillator signal 21 of frequency ur + f0 or f -f.

2. A second down-conversion stage 22 converting the in-phase 19 and in-quadrature 20 output band-pass signals of the first quadrature down-conversion stage 17 from the intermediatc centre frequency fF to an in-phase 23 and an in-quadrature 24 signal centered around the low-IF frequency ftauqp. This down-conversion is done by using a double quadrature mixer structure as described in 1 single-chip 900MHz CMOS receiver front-end with a high performance low-iF topology J. Crols, ill. £ J Steyaert, iEEE Journal qf Solid-State Circuits, Vol 30, No 12, December 1995, The double quadrature mixer implements a complex multiplication of the complex output signal of the first quadrature down-conversion stage 17, which has the in-phase 49 and the in-quadrature 20 output signal of the first quadrature down-conversion stage 17 as real and imaginary part respectively, with a single complex tone at frequency f1 -fLowir, -f + fow, -fn' -fLown or fu - The double quadrature mixer consists of four mixers 25, an addition 26 and a subtraction 27. The complex single tone is implemented using a 0° and 90° phased shifled version of an oscillator signal 28 with a frequency of f1 -fLown or 1n + fLown.

3. An optional filter block 29, filtering out the high frequency components in the complex signal formed by the in-phase 23 and in-quadrature 24 output signals of the second down-conversion stage 22. The filter block 29 can implement a low-pass or band-pass filtering. For the embodiments of this invention in which the filter block 29 isn't included, signal 23 will be directly connected to signal 30 and signal 24 will be directly connected to signal 31.

4. An analog to digital converter (ADC) block 32 sampling and quantizing both the in-phase 30 and in-quadrature 31 analog output signals of the filter block 29 to result into an in-phase 33 and an in-quadrature 34 digital output signal.

5. A digital IQ imbalance estimation block 35 implementing an on-line blind IQ imbalance estimation algorithm. This estimation algorithm will calculate based on the in-phase 33 and in-quadrature 34 digital output signals the ADC block 32, a frequency independent estimate 36 of the IQ imbalance present at the output of the ADC block 32.

6. A digital IQ imbalance correction block 37 correcting the in-phase 33 and iu-qnadrature 34 digital output signals the ADC block 32 in the frequency range, containing the to-bc-received signal, based on the frequency independent estimate of the IQ imbalance 36. The corrected digital in-phase 38 and in-quadrature 38 signals form the output of the whole receiver.

The overall high rejection of interfering signals in the reception of a RE band-pass signal in the invention is realized by the combined effect of the following elements: 1. The use of a quadrature down-converter in the first down-conversion stage 17 in combination with the use of a double quadrature mixer in the second down-conversion stage 22.

2. The use of large bandwidth signal processing at every point throughout the receiver where the complex signal representation is used. This will allow for a frequency independent correction by the digital IQ imbalance correction block 37 of the IQ imbalance observed at the output of the ADC block 32, to the level needed by the overall receiver.

The following describes the essential effect both elements to allow for the high rejection of interfering signals in the reception of a RE band-pass signal: Element 1 Dy using a quadrature down-converter in the first down-conversion stage 17, the output of this down-conversion will be a complex signal having in its spectrum an attenuated copy of the wanted signal at the negative of the frequency of the wanted signals itself The attenuation is the image rejection ratio of the quadrature-down converter, Figure 5 illustrates the described down-conversion. The real input RE signal, consisting parted centered around fp 42 and part centered around -ftp 43, is multiplied by two complex tones 44 and 45 with a magnitude difference determined by the image rejection ratio JR1. The result of this operation consists of the part of the input RE signal centered around 42 in frequency shifted to the frequency f 46, and of an attenuated version the part of the input RE signal centered around -f 43 in frequency shifted to frequency -fa 47.

The output signal of the second down-conversion stage 22 will be a combination of the wanted signal and an interfering signals consisting of an in the frequency domain mirrored and attenuated version of the wanted signal. The magnitude of this attenuation is the image rejection ratio the second down-conversion stage 22. The combination of a quadrature down-conversion in the first down-conversion stage 17 and the double quadrature mixer in the second down-conversion stage 22 makes that the interfering signals are attenuated by the product of the image rejection ratio of the first and the second down-conversion stage.

Figure 6 illustrates the described down-conversion. The complex output signa.l of the first down-conversion stage 17, consisting of a wanted signal 46 and an unwanted signal 47 due to the finite image rejection ratio JR1 of the first down-conversion stage 17, is muhiplied by 2 complex tones 4$ and 49 with a magnitude difference determined by the image rejection ratio JR7 of the second down-conversion stage 22. The result is the signal, in a frequency band around f of the complex output signal of the first down-conversion stage 17, marked 46, placed at frequency fLnwfF 50 and an attenuated version of the signal around -t 47 placed at frequency -fLowip 51. When the frequency fcnwn is lower than the bandwidth of both signal 50 and signal 51, these signals will overlap.

Element 2 The effect of the bandwidth on the rejection of interfering signals for signal processing stages acting on signals using the complex representation can be explained vith the aid of Figure 7. Imperfection in the physical implementation of a real valued low-pass filtering on a complex signal could results in the transfer functions as schematically illustrated in this figure. The in-phase transfer function 52 has a DC gain of A[ and bandwidth RW1, the in-quadrature transfer function 53 has a DC gain of A0 and bandwidth BW0. The difference between the 2 transfer functions is frequency dependent, especially for frequencies close to the edge of the pass-band (i.e. BW1 and BWQ). This will result in frequency dependent image rejection ratio 54 close to the edge of the pass-band.

The observation that the image rejection ratio becomes frequency dependent for frequencies close to the edge of the pass-band of the filter, as described above for the specific case of a low-pass filtering, can be generalized to other ldnd of filtering: "In the frequency region close to the edges of the pass-band of any filter, the image rejection ratio becomes frequency dependent".

In the frequency independent IQ imbalance calibration used in this invention, the digital output signals 33 and 34 of the analog to digital conversion block 32 are corrected using a mathematical correction which is independent of the frequency. The frequency dependent nature of the image rejection ratio will reduce the accuracy of the IQ imbalance correction. To allow for the high rejection of interfering signals in the reception of a RE band-pass signal as targeted by this invention, the frequency dependency of the image rejection ratio should he made small.

It is lruown in art that at each point in the receiver system a certain amount of filtering is present, This filtering can be deliberately added, e.g. like the optional filter block 29, or unintentionally present due to capacitances present at that point or other physical effects.

This invention uses large bandwidths for the filtering present at every point thronghont the receiver where the complex signal representation is used. The spectrum of the to-bc-received signal is at these points placed in the frequency domain within the pass-band of the present filtering at a substantial distance from the edges of the pass-band. By using larger bandwidths than common used in art, the effect of the frequency dependency of the IQ imbalance is minimised.

For filtering implemented by the optional filter block 29, this clement will result in a bandwidth for this filtering substantial larger than the frequency range of the to-be-received, the latter being the common value used in art for this filtering.

Brief description of the drawings

Figure 1 illustrates the corruption of the received signal by the signal on the mirror frequency in a quadrature down-conversion.

Figure 2 shows the signal flow graphs representing the decomposition of IQ imbalance in a filter into a wanted nominal filter function and an error filter fimction.

Figure 3 illustrates the corruption of a signal due to 1Q imbalance in the filter processing this signal.

Figure 4 is a block diagram of the receiver presented in the invention.

Figure 5 is a schematic illnstration of the mirror problem for the first down-conversion stage 17 of the invention, Figure 6 is a schematic illustration of the mirror problem of the second down-conversion stage 22 of the invention.

Figure 7 illustrates the frequency dependency of the image rejection ratio for a low-pass filter acting on a complex signal. having a gain and a bandwidth offset between the in-phase and in-quadrature channel of its implementation.

Figure 8 shows a possible embodiment of the first down-conversion stage 17.

Figure 9 shows a possible embodiment of the second down-conversion stage 22.

Figure 10 shows a possible embodiment of the filter block 29.

Figure 11 shows a possible embodiment of the analog to digital conversion block 32.

Figure 12 shows a possible embodiment of the digital IQ imbalance correction block 37.

Description of the preferred embodiments

Figure 8 shows a possible embodiment of the first quadrature down-conversion stage 1. The RF input signal 18 is filtered with the band-pass filter 55 to remove all signals out of the frequency band of interest which is centered around The filtered signal 56 is split into two channels, One channel, called the in-phase channel or I-channel in Figure 8, is down-converted using mixer 57 and a 0 degrees phase shifted version 58 of a local clock signal 21 of frequency fLo. The other channel, called the in-quadrature channel or Q-channel in Figure 8, is down-converted using mixer 59 and a 90 degrees phase shifted version 60 of the local clock signal 21. The output signals of mixers 57 and 59 are each band-pass filtered with a band-pass filter 61 having its pass-band centered around -f01. The output signals 19 and 20 form a. complex signal, in which 19 is the real part and 20 the imaginary part of the complex signal.

A possible embodiment of the second down-conversion stage 22 is given in Figure 9. The in-phase 19 and in-quadrature 20 output baud-pass signals of the first quadrature down-conversion stage 17 are implemented as differential signals. The four mixers 25 are in this embodiment realised as commutating switches (four-way switches). These switches are controlled by the differential clock signals 62 and 63 both of the same frequency fLu2 but with a 90 degrees phase difference between them, The addition 26 and subtraction 27 are realised by summing the differential output currents of the mixers 25 at the input of the feedback amplifiers 64. The inversion needed in the subtraction 27 is realised by swapping the position of the two signals forming a differential signal.

A possible embodiment of the optional fiher block 29 is given in Figure 10. it consists of two identical low-pass filters 65, one for the in-phase or I-channel and the other for the in-quadrature or Q-channel.

Both the input signals 23 and 24 as the output signals 30 and 31 form a complex signal. This embodiment of the filter block 29 performs a low-pass filtering on the complex signal formed by the output signal with the impulse response of the filter having only real values.

A possible embodiment of the analog to digital conversion block 32 is given in Figure 11. In this embodiment two identical analog to digital converters 66, one for each channel, are used. Possible embodiments of these analog to digital converters 66 could be the common known in art delta-sigma analog to digital converters.

A possible embodiment of the digital IQ imbalance correction block 37 is given in Figure 12. A linear transformation 67 is performed on the vector consisting of the in-phase 33 and in-quadrature 34 digital output signals the ADC block 32. The output of this transformation is the vector consisting of the corrected digital in-phase 38 and in-quadrature 39 output signals of the whole receiver. in this embodiment the 2 by 2 matrix determining this linear transformation is the frequency independent estimate of the IQ imbalance 36 as calculated by the digital IQ imbalance estimation block 35.

Claims

Claims What is claimed is: 1. A communication apparatus for the reception of signals comprising: a first down-convcrsion stage, implementing a quadrature down-conversion from a first frequency to a lower second frequency, having a single signal as input and an in-phase and an in-quadrature signal as output; a second down-conversion stage, converting the in-phase and in-quadrature output signal of the first down-conversion stage from the second frequency to a lower third frequency by the use of a double quadrature mixer structure which consists of four mixers, an addition and a suhiraction arranged in such manner that it implements a multiplication of the complex signal formed by the in-phase and in-qnadrature output signal of the first down-conversion stage with the complex signal formed by a first applied clock signal of a fourth frequency and a second clock signal also of the fourth frequency with a 90 degrees phase difference as compared to the first applied clock signal; an analog to digital conversion block, having the output signals of die second down-conversion stage as its input, sampling and quantizing its input signals to generate a digital in-phase and a digital in-quadrature output signal; a digital IQ imbalance estiniation block, calculating based on the digital in-phase and in-quadrature output signal of the analog to digital conversion block, a frequency independent estimate of the IQ imbalance observed in the digital in-phase and in-quadrature output signal of the analog to digital conversion block; and a digital IQ imbalance correction block, correcting the digital in-phase and in-quadrature output signals of the analog to digital conversion block in the frequency range, containing the to-be-received signal, based on the frequency independent estimate of the IQ imbalance made by the digital IQ imbalance estimation block; wherein all analog signal processing operations of the presented communication apparatus using two distinct physical signals to represents a single complex signal, have bandwidths such that the TQ imbalance observed at the in-phase and in-quadrature output of the analog to digital conversion block is frequency independent to the level needed by the digitallQ imbalance estimation block.
2. The apparatus of claim 1, wherein the analog to digital converter block is implemented as two identical analog to digital converters, each taking one of the two output signals of the second down-conversion stage as input.
3. The apparatus of claim 2, in which the analog to digital converters are implemented as delta-sigma analog to digital converters.
4. The apparatus of claim 1, in which the digital LQ imbalance estimation block calculates the estimate of the IQ imbalance at the same time as the presented communication apparatus does its reception of signals.
5, The apparatus of claim 4, in which the digital LQ imbalance estimation block calculates the estimate of the IQ imbalance using the stochastic properties of the digital in-phase and in-quadrature output signal of the analog to digital conversion block.
6. The apparatus of claim 4, in which the digital 1Q imbalance estimation block calculates the estimate of the IQ imbalance using the mutual stochastic independence of signals situated in different parts of the spectrum of the input signal of the first down-conversion stage.
7. A communication apparatus for the reception of signals comprising: a first down-conversion stage, implementing a quadrature down-conversion from a first frequency to a lower second frequency, having a single signal as input and an in-phase and an in-quadrature signal as output; a second down-conversion stage, converting the in-phase and in-quadrature output signal of the first down-conversion stage from the second frequency to a lower third frequency by the use of a double quadrature mixer structure which consists of four mixers, an addition and a subU-action arranged in such manner that it implements a multiplication of the complex signal formed by the in-phase and in-qnadratnre output signal of the first down-conversion stage with the complex signal formed by a first applied clock signal of a fourth frequency and a second clock signal also of the fourth frequency with a 90 degrees phase difference as compared to the first applied clock signal; a filter block, removing the high frequency components of output signals of the second down-conversion stage which are applied to its input, having an in-phase and an in-quadrature signal as output; an analog to digital conversion block, having the in-phase and in-quadrature output signal of filter block as its input, sampling and quantizing its input signals to generate a digital in-phase and a digital in-quadrature output signal; a digital IQ imbalance estimation block, calculating based on the digital in-phase and in-quadrature output signal of the analog to digital conversion block, a frequency independent estimate of the IQ imbalance observed in the digital in-phase and in-quadrature output signal of the analog to digital conversion block; and a digital TQ imbalance correction block, correcting the digital in-phase and in-quadrature output signals of the analog to digital conversion block in the frequency range, containing the to-be-received signal, based on the frequency independent estimate of the IQ imbalance made by the digital IQ imbalance estimation block; wherein all analog signal processing operations of the presented communication apparatus using two distinct physical signals to represents a single complex signal, have bandwidths such that the IQ imbalance observed at the in-phase and in-quadrature output of the analog to digital conversion block is frequency independent to the level needed by the digital IQ imbalance estimation block.
8. The apparatus of claim 7, wherein the filter block performs a low-pass filtering.
9. The apparatus of claim 7, wherein the filter block performs a band-pass filtering.
10. The apparatus of claim 7, wherein the analog to digital converter block is implemented as two identical analog to digital converters, the first taking the in-phase output signal of the filter block as its input, the other taking the in-quadrature output signal of the filter block as its input.
11. The apparatus of claim 10, in which the analog to digital converters are implemented as delta-sigma analog to digital converters.
12. The apparatus of claim 7, in which the digital IQ imbalance estimation block calculates the estimate of the IQ imbalance at the same time as the presented communication apparatus does its reception of signals.
13. The apparatus of claim 12, in which the digital IQ imbalance estimation block calculates the estimate of the IQ imbalance using the stochastic properties of the digital in-phase and in-quadrature output signal of the analog to digital conversion block.
14. The apparatus of claim 12, in which the digital IQ imbalance estimation block calculates the estimate of the IQ imbalance using the mutual stochastic independence of signals situated in different parts of the spectrum of the inpnt signal of the first down-conversion stage.