MXPA00009458A - Distortion correction circuit for direct conversion receiver - Google Patents

Abstract

A distortion correction circuit (5) for a direct conversion receiver, the direct conversion receiver comprising a local oscillator (2) for generating a local oscillator signal, a mixer (3) for multiplying a radio frequency signal, and a local oscillator signal together and supplying the resultant mixer output signal to a demodulated signal path (8), the demodulated signal path including a first low-pass filter (4) for selecting a baseband signal at a first filter output terminal, wherein the offset correction circuit (5) includes a squaring circuit (6) for squaring the mixer output signal and supplying the resultant distortion estimate signal to a distortion estimate signal path (9), signal subtraction means (52) for subtracting the distortion estimate signal at a subtraction point (10) in the distortion estimate signal path, and adaptive processing means (28) for equalising the transfer function of the distortion signal estimate path and the transfer function of the demodulated signal path between the mixer output and the subtraction point.

Description

DISTORTION CORRECTION CIRCUIT FOR DIRECT CONVERSION RECEIVER The present invention relates to direct conversion • as, for example, in ho odin receivers, and in particular to circuits for the correction of DC deviation and intermodulation distortion of the second order in said receivers. One of the main problems affecting direct conversion receptors is the presence of deviations of CC in the baseband in lines of • output in phase and quadrature. They are difficult to eliminate because most of the digital modulations have a DC component, which must be conserved. Deviations from CC are caused by leakage from the local oscillator back to the antenna port of the terminal and for circuit imbalances. An AC coupling can be used to block DC deviations, but this coupling reduces the effectiveness of the demodulation due to the delay of additional group that is applied to the signal. The AC coupling can be used in low intermediate frequency (IF) conversion, where the converted channel has a normal low frequency deviation so that no component received from the channel desired one has a DC component, but these schemes _.s ____... _ have difficulty in obtaining selectivity of adjacent channels. The signal leak from the radio frequency input to the local oscillator port, and the imbalance of the circuit combined with non-linearity within the mixer and / or other circuit components (eg, amplifiers and filters), cause a second and most damaging form of DC deviation. Both effects cause the radio frequency signal to mix with itself generating a DC component that varies with the square of the amplitude of the input signal. This form of DC deviation is a result of second-order inter-modulation distortion within the mixer and / or other components. The recovery of the baseband signal demodulated by direct conversion receiver can therefore be interfered by any interference signal outside of a large channel, regardless of the frequency. In TDMA systems this distortion causes the DC deviation to pulse with the transmission pulses of a strong neighboring transmission. For transmissions employing non-constant envelope modulations, additional modulation deviations will occur, these modulation deviations often have spectral components with bandwidth that exceeds a channel bandwidth, and therefore are capable of causing interference to the systems that they operate in the low IF mode. Not as in the effects of inter-modulation, where the interfering signals must have a certain predefined frequency ratio before experiencing significant distortion, the second order intermodulation distortion can be found in current direct conversion receivers provided that there are powerful signals. At present, there is a need to effectively correct the distortion introduced in the direct conversion receivers by the DC deviation and second order intermodulation. With this in mind, the present invention provides a distortion correction circuit for a direct conversion receiver, the direct conversion receiver comprises a local oscillator for generating a local oscillator signal, a mixer for multiplying together a radio frequency signal and a local oscillator signal and supplying the resulting mixer output signal to a demodulated signal track, the demodulated signal track includes a first low pass filter to select a baseband signal at a first filter output terminal, at where he "* - *" - - - ~ deviation correction circuit includes a square circuit to square the output signal of the mixer and supply the resulting estimated distortion signal to an estimated signal distortion track, the signal subtraction means for subtracting the estimated distortion signal at a subtraction point in the estimated distortion signal track, and adaptive processing means for equalizing the transfer function of the estimated signal distortion track and the function of transfer of the signal track demodulated at the output of the mixer and the subtraction point. If the direct conversion receiver uses a quadrature demodulator then the distortion correction circuit can be applied to each of the quadrature arms and correlated phases. In one embodiment of the invention, the signal subtraction point is between the mixer and the first low pass filter. In another embodiment, the estimated distortion signal track further comprises a second low pass filter and the signal subtraction point is in the output terminal of the first filter. Conveniently, the subtraction of the distortion signal estimated at the subtraction point generates an error signal, an adaptive processing means acts to equalize said transfer function by minimizing the error signal. The adaptive processing means may be a linear output mixer n. The adaptive linear output mixer n preferably implements a minimum quadratic mean error algorithm. The adaptive processing means may be implemented by digital signal processing. The distortion correction circuit may also provide delay means to introduce a delay in the demodulated signal track, before the subtraction point. The adaptive linear mixer may include at least a first weight update coefficient for adjusting the weights of the plurality of said outputs. The adaptive linear mixer can include a DC output to minimize a DC deviation in the track of demodulated signal. In addition, the adaptive linear mixer can include a second weight update coefficient for adjusting the weight of the DC output separately. The following description refers in more detail to the different characteristics of the distortion correction circuit of the present invention. To facilitate understanding of the invention, reference is made in the description of the accompanying drawings wherein the distortion correction circuit is illustrated in a preferred embodiment. It should be understood that the distortion correction circuit of the present invention is not limited to the preferred embodiment as illustrated in the drawings. In the drawings: FIGURE 1 is a schematic diagram of a first embodiment of a direct conversion receiver that includes a distortion correction circuit according to the present invention; FIGURE 2 is a schematic diagram of a second embodiment of a direct conversion receiver that includes a distortion correction circuit according to the present invention; FIGURE 3 is a detailed view of a linear mixer of n outputs of the distortion correction circuit shown in FIGURE 2; FIGURES 4 and 5 are graphical representations of the inter-modulation signal intensity referred to the input signal plotted against the maximum interferent signal power, of a direct conversion receiver, respectively with and without inter-modulation distortion correction provided by the distortion correction circuit of this • invention. FIGURES 6 through 9 are graphical representations of changes generated from intermodulation distortion at the DC level in a direct conversion receiver caused by a carrier pulse in the TDMA environment; and ^ k: 10 FIGURE 10 is a schematic diagram of a receiver using a quadrature demodulator that includes a deviation correction circuit according to the present invention. The receiver can be used either for direct conversion to baseband or for conversion at a low intermediate frequency. Now observing FIGURE 1, there is generally shown a direct conversion receiver 1 comprising a local oscillator 2 and a mixer 3 followed by a low pass filter 4. the direct conversion receiver is adapted to receive a radio frequency carrier modulated in a first input terminal of the mixer. The local oscillator signal generated by the local oscillator 2 is applied to the other input terminal of the mixer. The mixer 2 multiplies the radio frequency carrier signal and the local oscillator signal together and provides the resulting multiplied signal v at the input terminal of the mixer. The low pass filter 4 has its input • connected together with the radio frequency carrier signal 5 modulated to a frequency change of the carrier signal, generating a signal component whose frequency is the sum of the signal frequency of the local oscillator and the frequency of the carrier, and a component of signal whose frequency is the difference ßfc 10 between the frequency of the signal of the local oscillator and the frequency of the carrier. In homodyne receivers, the frequency of the local oscillator and the frequency of the carrier are substantially equal, so that the "difference" of the signal component effectively eliminates the modulation of the carrier signal. The "difference" of the signal is the baseband signal recovered from the desired channel. The low pass filter 4 acts to accept the desired channel in the baseband to reject the "sum" of the signal component, and to reject other adjacent adjacent channels that have been transformed in the mixing process. In doing so, the low pass filter 4 satisfies some channel selectivity requirements of the direct conversion receiver. 25 Any change or distortion product at the DC level may result in distortion in the recovered baseband signal. All translated interfering signals are present before the low pass filter 4. These interfering signals include those very powerful signals that cause distortion, and in particular, the intermodulation interference of the second order in this case. The envelope and modulation of these powerful signals is preserved through the conversion carried out by the mixer 3. In order to minimize this distortion, a distortion correction circuit 5 is provided. The direct correction circuit generates a correction signal w or x, the prefiltered signal v passing through a non-linear circuit that substantially produces the dominant distortion characteristics. In order to generate an estimate of the second order intermodulation distortion, a square circuit 6 is used to square the output signal of the mixer. The output of the quadratic circuit is carried through an appropriate gain scaling means 7, before subtraction of the demodulated signal track in order to cancel the effect of the unwanted interference. The output v of the mixer 3 is supplied to the demodulated signal track 8, which includes a low pass filter 4 to eliminate unwanted out-of-band signals from the unwanted output signal v of the mixer. Similarly, the square output signal v 2 • is supplied to a distortion estimation signal track 9. In a first embodiment of the invention, the distortion estimation signal w is subtracted from the demodulated signal track 8 at a point between the mixer 3 and the first filter low pass 4. In this example, the fc 10 subtraction means 10 subtract the estimated distortion signal w from the output signal of the mixer v before the input of the low pass filter 4. this configuration advantageously allows the filter of low step 4 perform both jobs of the channel selectivity and elimination of unwanted out-of-band products of the quadratic process in the estimated distortion signal w. However, many homodine designs place the filtrate directly after the mixer, in which In this case, the distortion estimation signal may be fed forward and subtracted from the demodulated baseband signal after the low pass filter 4. In this case, a second low pass filter 11 may be included in the signal track of estimation of distortion 9 in order to eliminate unwanted and harmonic products from the distortion estimation signal w, and to introduce the transfer function of the first low pass filter 4 in the estimation signal track • Distortion 9. 5 Since the gain, response frequency and distortion characteristics of the direct conversion receiver can change with temperature and from circuit to circuit, it may be preferable for gain scaling media and for the pass filter under 11 a-- 10 to be adjusted adaptively during the operation of the direct conversion receiver. Accordingly, the gain-scaling means 7 and / or the adjustment of the low-pass filter 11 can be carried out by the adaptive process means which act to equalize the transfer function of the distortion estimation signal track 9 and the transfer function of the demodulated signal track 8 between Q the output mixer 3 and the point at which the distortion estimation signal is subtracted from the signal of baseband demodulated. It should be noted that in TDMA systems, the disparity between the two filters can result in short duration trips in the times of pulse rise and fall caused by the interfering signal due to delay time disparity and increase of the deviation correction circuit 5.

The input signal to the direct conversion receiver 1 consists of a radio frequency baseband signal m (t) cos (?. T) and an interferent signal to • (t) eos (cúc +? O) in some frequency deviation,? O- Normally, through the receiver's selectivity circuits. Accordingly, the input signal u is represented by the equation: • 10 After passing through mixer 3 it is multiplied by the signal, r (t), at the local oscillator (LO) port, which consists of a cosine wave plus part of the input signal u (t) that is has escaped within the LO track. The leak track has a gain of << < 1 and is a possible cause of the second order intermodulation distortion (IM2). The signal r is therefore given by: The signal at the output of the mixer 3 v (t) becomes 25, (= 0.5 (Re { / P (.}. + R aÍ?; ^ '.}.) T'0.25í: m ( f) |; + ¡í3 (/) f- + 2Re { r (* 3 (/) and '^'.}.) • where terms with frequencies more than? C are assumed to be filtered. The first term is the desired signal from the mixer, the second term is the interference signal at a carrier frequency? 0 that is normally eliminated by subsequent filtering fl 10 by the low pass filter 4. The third and fourth terms are band signals with a bandwidth of up to twice m (t), they are therefore a potential source of interference. The first term shows the cross modulation between the desired signals and interferers and at the same frequency of term two. Interference occurs when m (t) is smaller and (t) is large so that the second order intermodulation interfering signal, which is called (0.5 k2 | a (t) | 2), is similar amplitude that the signal of baseband demodulated (Re | m (t) I). The interference effect increases with the square of the amplitude of the interfering signal. The cancellation of the interfering signal is possible because the dominant signal in v (t) is the second term. The dominant output of the square circuit is therefore: 2 - . 2 - efaC V '"' j) in embodiments of the present invention wherein the estimated distortion signal is subtracted after filtering by the low-pass filter 4, it is second term at twice the frequency of deviation can ^ fc 10 be eliminated by filtering in the second low-pass filter 11, leaving the first term, which can be passed and scaled by g2 to cancel the interfering signal. Accordingly, x is given by the relation: 15 ^. -025A, .í: The output of the low pass filter 4, after subtracting 20 the signal x, is in accordance: A 0 _ Rc. { m. { ,)) 25 when g2 is adjusted to equal k2 and because k2m (t) 2 is small compared to other terms. The coefficient g2 of the gain scaling means 7 can be carried out by adaptive processing means which can be carried out using digital signal processing. The signal used to change these weights is defined by the error to the mean square between the estimated distortion signal and the demodulated baseband signal. These weights are adjusted to minimize this error. FIGURE 2 shows a modality of said arrangement. In this FIGURE, a direct conversion receiver 21 having a local oscillator 22, a mixer 23 and a low pass filter 24, operatively identical to the circuit shown in FIGURE 1, is presented. A distortion correction circuit 25 is shown in FIG. which includes a square circuit 26 which connects the output of the mixer 23. A low pass filter 27 is connected to the output of the mixer 26. The coefficient g2 of the gain climbing means 7 can be adjusted using correlation techniques. The distortion signal estimated at the output of the low pass filter 27 should not be correlated with the demodulated baseband signal, with the estimated distortion subtracted, when the adjustment is perfect. The residual correlation can be used to adjust the g2 coefficient and equalize the signal transfer functions of the estimated distortion track and the signal track • demodulated between the output of the mixer and the subtraction point by noting that the average square value of the output signal of the direct conversion receiver is minimized when the cancellation is perfect. Numerous tuning procedures are possible for this type of problem, including least squares, and algorithms RLS and LMS. However, if there is a DC deviation in the estimated distortion signal track, and there is DC present in the output signal of the mixer (caused by leakage in the local oscillator to the frequency output port), the correlation may give an exit false. The optimal setting of g2 to minimize the distortion of second order intermodulation is unlikely to correspond to the optimum setting to eliminate the DC component. A minimum quadratic mean error algorithm has been found that produces a A balanced solution that allows some CC to remain and some second order intermodulation distortion in the output signal, and the direct conversion receiver. A separate DC cancellation coefficient is required if both the CC and the distortion of inter-modulation of second order must be canceled.

Other methods of correcting the DC deviation are possible. The adaptive processing means 28 shown in FIGURE 2 are effected by an adaptive linear mixer using an LMS algorithm. The adaptive linear mixer 28 comprises, in this illustrative example, six delay elements 29 to 34 connected in series to the output of the low pass filter. Seven values of output samples 35 to 41 are taken from the adaptive linear mixer, the gain or weight of the signal at each output is determined by the coefficients 42 to 48. the signals obtained are summed by the adder 49. The adaptive linear mixer 28 also includes a DC 50 output to minimize any DC deviation in the modulated signal track, and a coefficient or weight 51 that adjusts the gain of the DC output signal. The signal from the DC output 50 is also summed with the other output signals by the adder 49. Two equalizer update coefficients μx and μdc are provided by applying small adjustments respectively to the weights 42 to 48 and to the weight of CC 51. The step size of the update coefficients determines the range of convergence of the adaptive linear mixer, as well as its stability and precision. The estimated distortion signal produced at the output of the low pass filter 49 is subtracted from the baseband signal demodulated at the output of the low pass filter 24 by the subtraction circuit 52. The delay means 53 can be inputted within of the demodulated signal track, before the subtraction point of the estimated distortion signal, in order to synchronize the estimated distortion signal and the demodulated baseband signal during subtraction. Conveniently, the delay introduced in the demodulated signal track may be approximately equal to half the delay introduced in the distortion signal track estimated by the equalizer 28, or in other words, enter approximately half the number of delay elements in the demodulated signal track of those in the equalizer 28. FIGURE 3 provides a more detailed representation of the equalizer 28 shown in FIGURE 2. The estimated distortion signal x of the output of the low pass filter 27 should ideally have its amplitude and phase aligned with the distortion in the demodulated baseband signal, s, for good cancellation. The n-output equalizer 28 effectively combines the functions of the gain-scaling circuit 7 shown in FIGURE 1 with the required equalization of the transfer function of the estimated distortion signal track and the demodulated signal track between the output of the mixer and the subtraction point. Generally speaking, the disparity between the low pass filter 24 and the low pass filter 27 will occur, and worse disparity more outputs are necessary in the equalizer. On the other hand, if the transfer function of the two signal tracks are identical except for the gain then only one output is required, and this performs the function and operation of the gain scaling circuit 7 of FIGURE 1 only. The LMS algorithm carried out by the adaptive linear mixer 28 proceeds by calculating the value of the output signal of the direct conversion receiver and carrying out the operation: - H- -. Y1 x w! The coefficients of the equalizer are equalized, as represented by: w, + 2.u, > ? t and * wfc = ^ + 2μ "cy The data are changed by the elements of delay 29 to 34 and the new output samples are taken. It should be noted that the estimated analog distortion signal x and the demodulated baseband signal at the output of the low pass filter 24 are digitized by the analog digital converters at a digital analog interface 54, and that the function of the adaptive linear mixer is carried out using digital signal processing. In another embodiment of the present invention, however, the equalization function can be performed by analogous means. In an experimental embodiment of the present invention, the two low pass filters 24 and 27 had cutoff frequencies of 20 kHz, and the DSP sample range was set at 48 kHz for analysis and for processing FIGURE T. A cell device Gilbert carried out the quadratic function. The first test used a two-tone interference signal for u, at 6 MHz of separation from the desired channel. The tones were separated by 2 kHz to give a rhythm frequency of 4 kHz. FIGURE 4 indicates the amplitude of the second order intermodulation of 4 kHz (IM2) of the distortion signal (with reference to the input) against the peak intensity of the two-tone interfering signal. The terms of the fourth order intermodulation (IM4) (8kHz) and the sixth order intermodulation (IM6) (12 kHz) are also plotted and become dominant when the interference is extremely high (-23dBm for this device ). When the canceller is activated, as can be seen in FIGURE 5, the IM signal is canceled close to the sound floor. The increase in the interference margin depends on the bandwidth of the desired signal. A DAMPS signal, for example, requires a bandwidth of 15 kHz (RF bandwidth of 30 kHz) or using this experimental device would generate a noise power of -115.5 dBm (with reference to the 'input). An interfering signal of -35 dBm would generate an interference signal of the same power (-115.5 dB) without the canceller (FIGURE 3 above), however, an interference signal of 23 dBm can be tolerated with a canceller (FIGURE 3 below) ); an improvement of 12 dB (or 8 dB for the bandwidth of 100 kHz GSM). The improvement would be greater for narrower bandwidth transmissions, or for receivers with lower power numbers. The two-tone interfering signal, however, does not test dynamic changes satisfactorily, since it is easy to set the cancellation for a single set of frequencies. A test signal that produces a higher spectral content in the IM2 signal is desirable. The following test imitates the TDMA environment and involves the use of a carrier pulse using the recommended GSM duration, rise and fall times. The waveform graphs show the changes generated in the IM2 at the CC level when the pulse is present. FIGURE 6 shows the waveform in the receiver output signal and (FIGURE 2) without correction; the deviation of CC and the IM2 variations are present. In this system IM2 causes the negative change in the CC level. FIGURE 7 shows only the correction of the CC level; still present in the lowering of the CC level when the pulse is present. FIGURE 8 shows the correction of CC and IM2, using only 1 output. Unequalized tracks cause distortion flashes each time the interfering signal has a high degree of change, such as when rising and falling times occur. When a 7-output equalizer is added, these flashes are reduced as shown in FIGURE 9, but a slight increase in noise level is still apparent when the carrier pulse is present. This seems to be caused by an increase in the amount of noise from the frontal analog circuits. When adding more outputs the operation of the equalization is improved for low convergence since the adaptation coefficient μ must be reduced in line according to the LMS theory.

Referring now to FIGURE 10, there is shown an additional example of a direct conversion receiver using a distortion correction circuit in accordance with • present invention. The direct conversion receiver 60 5 comprises an RF amplifier 61 and a signal separator 62 for generating two copies of the input signal to the receiver. Each divided signal is supplied to mixer 63 and mixer 64. A local oscillator 65 supplies a 90 degree separator, which in turn p 10 supplies each mixer 63 and mixer 64 with the local oscillator signal in quadrature phase. In the illustrated example, the output of the mixer 63 defines the track of the output quadrature signal while the output of the mixer 64 defines the output track of the signal in phase. The low pass filters 67 and 68 are respectively connected to the output terminals of the mixer 63 and the mixer 64. Each arm of the quadrature demodulator w shown in FIGURE 10 operates in the same manner as the direct conversion receiver 1 of the FIGURE 1. The distortion correction circuit 69 is connected between the output of the mixer 63 and the signal subtraction pin 71 on the quadrature arm. Similarly, the direct correction circuit 70 is connected between the output of the mixer 64 and the point of ___________ subtraction of signal 72 in the arm in phase. Both distortion correction circuits 69. and 70 operate as previously described. It can be appreciated from the foregoing that the present invention provides post-distortion cancellation which is effective to improve the distortion operation of direct conversion receiver. The distortion correction circuit of the present invention uses a quadratic function to cancel the second-order intermodulation distortion primarily in a homodyne receiver as long as the baseband signal track and the estimated distortion track have functions of transfer which are equalized for a good operation under dynamic conditions. Finally, it should be understood that various modifications and / or additions can be integrated into the deviation correction circuit without departing from the scope of the present invention as defined in the appended claims. For example, while the previously described embodiments of the present invention act to correct the distortion in a direct conversion receiver, the present invention is also suitable for use in a receiver which uses intermediate low frequencies wherein the signal of distortion of interest. -modulation of second order affects the intermediate frequency signal.

Claims (1)

1. CLAIMS A distortion correction circuit for a direct conversion receiver, the direct conversion receiver comprises a local oscillator to generate a local oscillator signal, a mixer to multiply a radio frequency signal, and a local oscillator signal so that in As a result of the set-up of the output signal of the resulting mixer to a demodulated signal, the demodulated signal track includes a first low-pass filter for selecting a baseband signal in a first filter output terminal, wherein the deviation includes: a square circuit to square the output signal of the mixer and supply the resulting estimated distortion signal to the estimated distortion signal track, a signal subtraction means to subtract the estimated distortion signal at a subtraction point in the estimated distortion signal track, and adaptive processing means to equalize the to the transfer function of the estimated distortion signal track and the transfer function of the demodulated signal track between the output of the - »-» ^ ** - mixer and the subtraction point. The distortion correction circuit according to claim 1, wherein the signal subtraction point is between the mixer and the first low pass filter. The distortion correction circuit according to claim 1, wherein the estimated distortion signal track further comprises a second low pass filter, and the signal subtraction point is at the output terminal of the first filter. A distortion correction circuit according to any of the preceding claims, wherein the subtraction of the distortion signal estimated at the subtraction point generates an error signal, the adaptive processing means acts to equalize said transfer functions by minimizing the error signal. A distortion correction circuit according to any one of the preceding claims, wherein the adaptive processing means is an adaptive linear mixer of n outputs. The distortion correction circuit according to claim 5, wherein the n-output adaptive linear mixer implements a • * > • * - - minimum square root mean square error algorithm. The distortion correction circuit according to any of claims 5 or 6, wherein the adaptive processing means are implemented by the digital signal processing. The distortion correction circuit according to any of claims 5 to 7, further comprising delay means for introducing a delay in the demodulated signal track, before the subtraction point. The distortion correction circuit according to any of claims 5 to 8, wherein the adaptive linear mixer includes at least a first weight update coefficient for adjusting the weights of a plurality of said outputs. The distortion correction circuit according to any of claims 5 to 8, wherein the adaptive linear mixer includes a DC output to minimize a DC offset in the demodulated signal track. The distortion correction circuit according to claims 9 and 10, wherein the adaptive linear mixer further includes a second weight update coefficient for separately adjusting the weight on the DC output __________