JPWO2013011973A1 - IQ mismatch correction method and RF transceiver - Google Patents

Abstract

The present invention corrects IQ mismatches in both transmission and reception that are easy to mount using a resource-saving digital signal processing technology that takes advantage of a miniaturized CMOS process. A feedback path from the output side of the quadrature modulator to the input side of the quadrature demodulator is provided, and a first path of the path from the input side of the quadrature modulator to the output side of the quadrature demodulator via the feedback path is provided. 1 is estimated, a transmitter-side correction function for improving IRR is calculated, and then a transmitter-side correction coefficient is calculated. Next, the transmission I channel signal and the transmission Q channel signal corrected by the transmitter side correction coefficient are input to the quadrature modulator, the second transfer characteristic of the same path as described above is estimated again, and the receiver side for improving the IRR After calculating the correction function, the receiver side correction coefficient is calculated. [Selection] Figure 1

Description

  The present invention relates to an IQ mismatch correction method for correcting an IQ mismatch generated in a transmitter-side quadrature modulator and a receiver-side quadrature demodulator of an RF transceiver and an RF transceiver.

  In wired and wireless digital modulation / demodulation RF band communication systems, in order to achieve high throughput, a wider band per communication channel including secondary modulation such as Orthogonal Frequency Division Multiplexing (OFDM) is progressing. One example is the MoCA (Multimedia over Coax Alliance) standard that uses coaxial cables in the home as the communication medium. The MoCA 2.0 standard, which is the next generation standard of MoCA 1.x, uses OFDM communication using a 100 MHz band per communication channel. System.

  In communication using the RF band, the transceiver includes an up-converter and a down-converter that perform frequency conversion between the RF band and the baseband. On the transmitter side, a digital signal generated from the DSP is converted into a baseband analog signal via a DA converter, then frequency-converted to an RF band, and transmitted to a medium such as a coaxial cable. On the other hand, on the receiver side, the received analog signal in the RF band from the media is converted to a baseband analog signal, and then converted into a digital signal via an AD converter, and digital signal processing is performed by the DSP. .

  Conventionally, various architectures have been adopted for the up-converter and the down-converter. In particular, the direct conversion method (sometimes referred to as homodyne method or Zero-IF method) has been increasingly adopted due to the recent progress toward broadband communication. An advantage of the direct conversion method is that the conversion speed of the DA converter and AD converter can be reduced by half. That is, according to Shannon's sampling theorem, only conversion of signals that can be modulated / demodulated up to half the frequency of the conversion frequency can be performed. Therefore, if the MoCA 2.0 standard is taken as an example, a conversion rate of 200 Msps is the minimum. It becomes necessary. However, in the direct conversion method, a 100 MHz wide RF band signal can be handled as a half 50 MHz baseband signal, so that the performance required for the DA / AD converter can be reduced to at least 100 Msps.

  In this way, the point that a low-spec DA / AD converter can be used is the greatest merit of the direct conversion system, but as a price, the demands for the up-converter and the down-converter are increased. One of them is a problem of IQ mismatch in the quadrature modulation / demodulation mixer. The IQ mismatch problem can be cited on both the transmitter side and the receiver side, and will be described below by taking the receiver side architecture shown in FIG. 12 as an example.

  In FIG. 12, 12Rx is a receiving antenna, 5RxI is an I channel side mixer, 5RxQ is a Q channel side mixer, 6 is a local oscillator, 7Rx is a local signal source, 4RxI is an I channel low pass filter, 4RxQ is a Q channel low pass filter, 3RxI is an I channel side AD converter, 3RxQ is a Q channel side AD converter, and 31Rx is an OFDM demodulator.

ただし、

である。In this receiver, ideally, in quadrature demodulation, two types of oscillation sine waves output from the local signal source 7Rx and having different phases by exactly π / 2 are multiplied by the received signal represented by (Equation 1). Is done by. However, in (Formula 1), j represents an imaginary unit, and () ^* represents a conjugate complex number.

However,

It is.

したがって、ベースバンド信号として取り出せる信号は、（式２）の信号そのものとなり、送信信号が受信機によって理想的に復元できる。As shown in FIG. 12, baseband signals generated by multiplying two types of sine waves are referred to as an I channel signal and a Q channel signal, respectively. If the data path and the local signal path branched into two I and Q shown in FIG. 12 are in an ideal state with no amplitude or phase error, the received signal of (Equation 1) is converted to (Equation 3) in the entire system. Are multiplied by the local signal and passed through the low-pass filters 4RxI and 4RxQ, only the baseband signal is extracted.

Therefore, the signal that can be extracted as the baseband signal is the signal itself of (Equation 2), and the transmission signal can be ideally restored by the receiver.

受信信号（式１）と非理想ローカル信号（式４）を掛け合わせ、ローパスフィルタ４ＲｘＩ，４ＲｘＱを通過させることによって取り出せる信号は、（式５）のようになってしまう。

ただし、

である。However, if there is an amplitude error g _RX or phase error φ _RX in the data path or the path of the local signal source 7Rx, the situation is different. As a non-ideal local signal, the signal of (Equation 4) is multiplied by the received signal.

A signal that can be extracted by multiplying the received signal (Equation 1) and the non-ideal local signal (Equation 4) and passing through the low-pass filters 4RxI and 4RxQ is as shown in (Equation 5).

However,

It is.

When the signal obtained by AD conversion of the signal of (Expression 5) and sampled is y [n], the output of the OFDM demodulator (DFT calculator) 31Rx is applied to the expression of the discrete Fourier transform as follows (Expression 7) )become that way. Here, L is the number of subcarriers in OFDM communication, and m is the subcarrier number.

As can be seen from (Equation 7), in the communication of subcarrier number m, the mirror frequency with respect to DC as the symmetry point—the m-th communication signal component is mixed as an interference wave (referred to as an image signal) by the weight of coefficient K _2. End up. Conversely also vice, in the communication of the sub-carrier number -m, the coefficient K ₂ weight fraction m-th communication signal will be mixed. As is clear from (Equation 6), the greater the amplitude error g _RX and the phase error φ _RX , the greater the weight K ₂ , which hinders the signal originally intended to communicate. In general, an IRR (Image Rejection Ratio) is defined as (Equation 8) as an index of the disturbance. A low IRR means a reduction in communication throughput.

  For example, in a miniaturized CMOS process in recent years, no matter how the layout is drawn symmetrically, the IRR is lower than that required by the standard due to variations in the manufacturing process. For example, the value of IRR 50 dB to 60 dB required in recent communication standards of several hundred Mbps class such as MoCA 2.0 cannot be achieved as it is. In the above example, the receiver side is taken up, but the same can be said for the IQ mismatch on the transmitter side.

  With respect to this, as will be introduced later as conventional technology, various means for correcting IQ mismatch in this direct conversion RF transceiver in analog or digital signal processing have been proposed. However, there are few proposals for means for simultaneously performing both transmission and reception corrections.

  In addition, in mobile terminals and the like, an RF transceiver is driven by a battery, so an architecture that can achieve low current consumption is required. In such an RF transceiver, low current consumption can be achieved without performing IQ quadrature modulation as typified by a polar transmitter (Non-Patent Document 3) or an out-fading transmitter (Non-Patent Document 4). Such an architecture is adopted. Also in this case, it is usual to employ a quadrature demodulator such as a direct conversion system for the receiver. Since the polar transmitter and the out-fading transmitter are outside the scope of the claims of the present invention, description thereof will be omitted.

  In this case, the transmitter does not have an IQ mismatch problem and does not need to be corrected. However, the above-mentioned problem still remains in the receiver, and it is meaningful to propose a means for correcting IQ mismatch in the receiver.

  In recent years, an architecture such as a sampling RF receiver as shown in (Non-patent Document 3) has been proposed as a quadrature demodulator of a receiver. Even in this case, problems similar to those of the quadrature demodulator using the conventional analog mixer remain, and are included in the quadrature demodulator in a broad sense.

<First Conventional Example (Patent Document 1)>
The first conventional example uses the known signal and pilot subcarrier in the preamble in the communication standard using OFDM, as shown in FIG. 13, and the amplitude of the propagation path between the transmitter Tx and the receiver Rx and This is a method of estimating and correcting IQ mismatch including phase distortion. In FIG. 13, the transmitter Tx outputs a pilot signal generating unit 32 that generates a pilot signal as one of the subcarriers, an OFDM modulator 31Tx, an output signal of the OFDM modulator 31Tx, and a known signal generating unit 33. A switch SW3 for switching between known propagation path estimation signals, a transmitter-side analog front end 34Tx that orthogonally modulates a baseband signal to an RF band signal, and a transmission antenna 12Tx are provided. The receiver Rx includes a receiving antenna 12Rx, a receiver-side analog front end 34Rx that orthogonally demodulates a received signal to a baseband signal, an OFDM demodulator 31Rx, a propagation path estimation unit 36, and a propagation path equivalent unit 35. .

Problems with this technology include the following.
(1) A specific known signal is required and has communication protocol dependency.
(2) The IQ mismatch in the transmitter-side quadrature modulator and the IQ mismatch in the receiver-side quadrature demodulator are not necessarily corrected individually. That is, IQ mismatch remains in both sides, and correction is made as a total of the transceiver.
(3) Due to the above two points, re-estimation and re-correction are required when communication is performed from a specific transmitter to another receiver.
(4) Moreover, it is necessary to implement the same algorithm in all the receivers, and the interoperability is poor. In other words, some kind of unified arrangement is required between different system vendors and device vendors participating in the standard.

<Second Conventional Example (Non-Patent Document 1, Non-Patent Document 2)>
In the second conventional example, as shown in FIG. 14, in one transceiver 30, a part of a transmission signal is extracted by a coupler 9, amplified by a loopback amplifier 11, and fed back to be orthogonal to the transmission path 37. Phase mismatch (φ _TX ), amplitude mismatch (g _TX ) of local signal source 7Tx in the modulator, and frequency characteristic mismatch between low-pass filters 4TxI and 4TxQ inserted immediately after I-channel and Q-channel DA converters 3TxI and 3TxQ Is a method of correcting the above. At this time, a transmitter-side mismatch equalizer 38 is inserted before the DA converters 3TxI and 3TxQ in the transmission path 37, and a transmitter-side mismatch estimator 40 is inserted in the output of the feedback path 39. In the feedback path 39, it is necessary to prevent an inadvertent IQ mismatch from occurring in the feedback path. For this reason, the local oscillator 41 and the mixer 42 that oscillate at a frequency different from the local oscillator 6 used in the quadrature modulator of the transmission path 37 once down-converts the signal frequency to the IF frequency and passes through the low-pass filter 43. Then, it is converted into a digital signal by a single AD converter 44. As a result, a technique of separating the I component and the Q component of the baseband using the local signal oscillated by the digital oscillator 45 having no mismatch, the mixers 46 and 47 and the phase shifter 48 is adopted.

  In the case of this technique, the transmitter-side mismatch can be corrected by a single device by devising a mismatch estimation method by utilizing actual signal characteristics using a wide band such as OFDM. In addition, the problem of poor interoperability as in the first conventional example is eliminated.

ただし、○の中に×を表した記号は、畳み込み演算を表す。The mathematical principle of mismatch correction of the second conventional example will be described below. An equivalent mathematical model of the transmitter including the IQ mismatch shown in FIG. 14 can be expressed as shown in FIG. However, h _TXI and h _TXQ in FIG. 15 indicate impulse responses of the low-pass filters 4TxI and 4TxQ, respectively. First, consider the case where the transmission signals s _I and s _Q are transmitted without any correction, that is, the case of s _I = p _I and s _Q = p _Q. At this time, the complex transmission signal u whose frequency is up-converted to the RF band can be expressed as (Equation 9).

However, a symbol representing x in the circle represents a convolution operation.

ただし、Ｕ(ω)、Ｓ_I(ω) 、Ｓ_Q(ω)、Ｈ_TXI(ω) およびＨ_TXQ(ω)はそれぞれ、ｕ、ｓ_I 、ｓ_Q、ｈ_TXI、ｈ_TXQのフーリエ変換である。ここで例として分かりやすくＳ_I(ω)、Ｓ_Q(ω)としてＯＦＤＭ信号を取り上げる。ｍ番目のサブキャリアと−m番目のサブキャリアで送出される複素シンボルをそれぞれｄ_m、d_−mとし、m番目のサブキャリアと−m番目のサブキャリアのみに着目するとｓ_I、ｓ_Qは（式１１）のように表される。

When (Equation 9) is Fourier transformed, (Equation 10) is obtained.

However, U (ω), S _I (ω), S _Q (ω), H _TXI (ω) and H _TXQ (ω) are Fourier transforms of u, s _I , s _Q , h _TXI , and h _TXQ , respectively. is there. Here, as an example, an OFDM signal is taken up as S _I (ω) and S _Q (ω) for easy understanding. Assuming that the complex symbols transmitted by the m-th subcarrier and the -m-th subcarrier are d _m and d _-m , respectively, and paying attention to only the m-th subcarrier and the -m-th subcarrier, s _I and s _Q are It is expressed as (Equation 11).

In (Equation 11), in particular attention only to signals transmitted at a carrier exp (+ j [omega] _m t) is substituted into (Equation 10) is expressed as (Equation 12). However, F is a Fourier transform.

In (Equation _{12), H TXI (ω m} ) = H TXQ (ω m) = H TX (ω m), g TX = 1 and phi _TX = 0, that is IQ when mismatch is assumed completely free situation, the rightmost side The second item disappears, as shown in (Equation 13), indicating that it is in the ideal transmission state.

That is, when there is an IQ mismatch, it can be seen that the first item on the rightmost side of (Equation 12) is the desired signal and the second item is the disturbing image signal. With this in mind, it becomes as in the symbol d _m sent by the m-th sub-carrier given the IRR defined by equation (8) (Equation 14).

In order to increase IRR (m) in a state where there is an IQ mismatch, when considering from (Expression 14), correction coefficients C _I (ω _m ) and C _Q (ω _m ) are expressed as (Expression 15). It is understood that IRR (m) should be taken as shown in (Equation 16).

From (Expression 11), (Expression 12), and (Expression 16), C _I (ω _m ) and C _Q (ω _m ) are transmitted to the transmission signals S _I (ω _m ) and S _Q (ω _m ), respectively. It can be seen that an improvement in IRR (m) can be expected by multiplying and sending. At this time, C _I (ω _m ) and C _Q (ω _m ) may be any complex numbers that satisfy (Equation 15). Furthermore, since IQ mismatch generally has frequency characteristics and the value of IRR varies depending on the value of m, (Equation 15) is generalized to (Equation 17).

As shown in FIG. 16, a transmitter-side mismatch equalizer 38 is obtained by appropriately dividing C _I (ω) and C _Q (ω) so as to satisfy (Equation 17) and applying an inverse Fourier transform. It can be realized with a simple circuit configuration. However, since IQ mismatch has frequency characteristics, c ₁₁ , c ₂₁ , c ₁₂ and c ₂₂ in FIG. 16 are not constant multipliers but real impulse responses. In the digital circuit, these are implemented as four FIR filters. In Non-Patent Document 2, in consideration of reduction of the calculation amount and resource saving of the circuit, an approximate correction coefficient such as (Equation 18) is used instead of (Equation 17), and a transmitter as shown in FIG. The side mismatch equalizer 38 is implemented.

However, the impulse response f _TX and the constant λ _TX in FIG. 17 are expressed by (Equation 19). In (Equation 19), F ⁻¹ represents the inverse Fourier transform. Further, a delay unit 381 in FIG. 17 is inserted to cancel the delay of the impulse response f _TX between the I channel and the Q channel.

In order to determine the above c ₁₁ , c ₂₁ , c ₁₂ and c ₂₂ , it is necessary to estimate an IQ mismatch. In general, the impulse response from the transmission signals s _I and s _Q shown in FIG. 14 to the generation point of the feedback baseband signals y _I and y _Q via the transmission path 37, the coupler 9, the loopback amplifier 11, and the loopback path 39. The model is represented in FIG. However, h ₁₁ , h ₂₁ , h ₁₂ and h ₂₂ in FIG. 18 are real impulse responses. These impulse responses are estimated by the transmitter mismatch estimator 38 in FIG. As an estimation method, there are a method using a statistical average or a least square method as introduced in Non-Patent Document 1, but the description is omitted because it is outside the scope of the present invention.

この（式２０）のＶ_TX（ω）を使用し、送信機側ミスマッチ等価器３８を構成することになる。At this time, H _α (ω), H _β (ω) and the correction function V _TX (ω) are defined as in (Equation 20). Then, the coupler 9 subsequent passes in 14 contribution of these in the denominator molecules V _TX (omega) since it is common that between the I-channel, Q-channel is reduced fraction, V _TX (omega) is (formula 17) Will match. However, in (Equation 20), F represents a Fourier transform, and ^ represents an estimated value.

The transmitter-side mismatch equalizer 38 is configured by using V _TX (ω) of (Equation 20).

Problems with this technology include the following.
(1) As shown in FIG. 14, two types of oscillators, local oscillators 6 and 41, are required as the oscillator. Since an oscillator is usually composed of a PLL, there is a corresponding resource overhead.
(2) Furthermore, it is necessary to provide a feedback path dedicated to calibration as shown in FIG. Although the receiver side can adopt a direct conversion method corresponding to a wide band, resources for a feedback path dedicated for calibration are required, and there is an overhead for this.

<Third Conventional Example (Patent Document 2)>
As shown in FIG. 19, the third conventional example is a technique that employs a direct conversion method for both transmission and reception, and includes a mismatch estimator 51, a transmitter-side mismatch equalizer 52Tx, and a receiver-side mismatch equalizer 52Rx. This is similar to the present invention described later. In this case, there is no lack of interoperability that was a problem of the first conventional example, and all the problems of the second conventional example can be overcome.

The correction mechanism in this technique is described below. First, there is a loopback delay between the baseband signals p _I and p _Q and the signals y _I and y _{Q in} FIG. For this reason, a correlation function is calculated by the delay time estimation unit 511 in the mismatch estimator 51 shown in FIG. 20, and a portion having a high degree of correlation is determined as a delay amount of the signals p _I and p _Q in the delay time compensation unit 512. . Next, the transmission signal is looped back by setting both 0 and π / 2 of the phase shift of the variable phase shifter 701 in the local signal source 7Rx in FIG. And it transmits to the equivalent parameter calculation part 513 in the mismatch estimation part 51 shown in FIG. 20, and makes a "multidimensional non-simultaneous equation" by the method as described in the paragraphs 0068-0106 of patent document 2. FIG. Finally, this equation is solved by some method to determine equivalent parameters to be provided to the transmitter-side mismatch equalizer 52Tx and the receiver-side mismatch equalizer 52Rx.

  In Patent Document 2, the mismatch between the transmitter-side low-pass filters 4TxI and 4TxQ and the mismatch between the receiver-side low-pass filters 4RxI and 4RxQ in FIG. 19 are not considered. That is, frequency-dependent IQ mismatch is not considered. However, this can be taken into account. That is, if the impulse response as shown in FIG. 16 as introduced in the second conventional example is taken into consideration as the transmitter-side mismatch equalizer 52Tx and the receiver-side mismatch equalizer 52Rx, the frequency dependence It can be extended to IQ mismatch.

Problems with this technology include the following.
(1) As a means for solving a multidimensional non-simultaneous equation, there is one as introduced in paragraph 0109 of Patent Document 2. Generally, however, the calculation amount is large, and the circuit scale is large when mounted on hardware. growing.
(2) As described above, it can be extended to frequency-dependent IQ mismatch. In communication using a wide band such as OFDM, when the frequency dependence of the loopback path delay time is strong, that is, when the group delay characteristic is not constant, the delay time estimation unit 511 in the mismatch estimator 51 can obtain a good correlation. Absent.
(3) In order to overcome the above problems, the delay time estimation unit 511 must have a means for obtaining a correlation for each subcarrier, which increases the circuit scale.
(4) Furthermore, if an attempt is made to deal with a frequency-dependent IQ mismatch, an equivalent parameter must be determined as an impulse response. For this reason, it is easily imagined that the dimension of the non-simultaneous equations further increases, resulting in an unrealistically large circuit scale.

JP 2008-252301 A JP 2008-022243

L. Anttila, M. Valkama, and M. Renfors, “Frequency-selective I / Q mismatch calibration of wideband direct-conversion transmitters,” Circuits and Systems II: Express Briefs, IEEE Transactions on, vol. 55, no. 4, pp.359-363, April 2008. J.Luo, A.Kortke, and W.Keusgen, “Joint calibration of frequency selective time variant I / Q-imbalance and modulator DC-offset error in broadband direct-conversion transmitters,” Communications, Circuits and Systems, 2009 International Conference, pp.255-259 RBStaszewski, K. Muhammad, D. Leipold, CMHung, .C. Ho, L. Wallberg, C. Fernando, K. Maggio, R. Staszewski, T. Jung, J. Koh, S. John, IYDeng, V.Sarda, O.Moreira-Tamayo, V.Mayega, R.Katz, O.Friedman, OEEliezer, E.de-Obaldia, and PTBalsara “All-Digital TX Frequency Synthesizer and Discrete-Time Receiver for Bluetooth Radio in 130-nm CMOS, ”IEEE J. Solid-State Circuits, vol.39, no.12, pp.2278-2291, December 2004 (Bluetooth is a registered trademark) M.E Heidari, M.Lee and A.A.Abidi “All-Digital Outphasing Modulator for a Software-Defined Transmitter,” IEEE J. Solid-State Circuits, vol.44, no.4, pp.1260-1271, April 2009

  An object of the present invention is to make it possible to easily correct IQ mismatch by a resource-saving digital signal processing technique that takes advantage of the miniaturized CMOS process. Specifically, in the case of an RF transmitter / receiver employing a quadrature modulation scheme for both the transmitter / receiver, the IQ mismatch between the two is corrected for the RF transmitter / receiver employing a quadrature modulation scheme only for the receiver. In other words, the IQ mismatch can be easily corrected.

In order to achieve the above object, an IQ mismatch correction method according to a first aspect of the present invention includes a quadrature modulator that quadrature modulates a transmission I channel signal and a transmission Q channel signal on a transmitter side, and a reception I on a receiver side. An IQ mismatch correction method in an RF transceiver including an orthogonal demodulator that orthogonally demodulates a channel signal and a received Q channel signal, from an input side of the orthogonal modulator to an output side, and further from an input side of the orthogonal demodulator Estimating a first transfer characteristic of a path to the output side, calculating a transmitter side correction coefficient for correcting a mismatch between the transmission I channel signal and the transmission Q channel signal in advance on the input side of the quadrature modulator, The transmission I channel signal and the transmission Q channel signal corrected by the transmitter side correction coefficient are input to the quadrature modulator, and from the input side of the quadrature modulator The second transfer characteristic of the path from the input side to the output side of the quadrature demodulator is estimated again, and the receiver correction function for improving the IRR is calculated using the result of the estimation again. Then, using the receiver-side correction function, a receiver-side correction coefficient for correcting a mismatch between the received I channel signal and the received Q channel signal is calculated on the output side of the quadrature demodulator.
According to a second aspect of the present invention, in the IQ mismatch correction method according to the first aspect, the calculation of the transmitter-side correction function is performed by using respective local signals input to the I channel mixer and the Q channel mixer of the quadrature modulator. Normal state 1 that does not swap and does not invert the polarity of each local signal input to the I channel mixer and Q channel mixer of the quadrature demodulator, and a local signal input to the I channel mixer and Q channel mixer of the quadrature modulator Normal state 2 in which one of the local signals input to the I channel mixer and the Q channel mixer of the quadrature demodulator is inverted, and the I channel mixer and the Q channel mixer of the quadrature modulator. The local signal to be input is swapped, and the I channel channel of the quadrature demodulator is Swap state 1 in which the polarities of the local signals input to the Q channel mixer and the Q channel mixer are not inverted, the local signals input to the I channel mixer and the Q channel mixer of the quadrature modulator are swapped, and the I channel of the quadrature demodulator A swap state 2 in which one of the local signals input to the mixer and the Q channel mixer is inverted in polarity is set, and calculation is performed using four transfer characteristics obtained in each state. .
According to a third aspect of the present invention, in the IQ mismatch correction method according to the second aspect, the calculation of the receiver-side correction function is obtained in each state by setting the normal state 1 and the normal state 2 again. The calculation is performed using two transfer characteristics.
According to a fourth aspect of the present invention, there is provided an RF transmitting / receiving apparatus comprising a quadrature modulator for orthogonally modulating a transmission I channel signal and a transmission Q channel signal on the transmitter side, and orthogonally receiving the I channel signal and the reception Q channel signal on the receiver side. In an RF transceiver including a quadrature demodulator for demodulation, feedback path forming means for forming a feedback path from the output side of the quadrature modulator to the input side of the quadrature demodulator, an I-channel mixer of the quadrature modulator, A swapping switch that selects one of a normal state in which each local signal is input to the Q channel mixer as it is and a swap state in which the respective local signals are interchanged with each other by switching, the I channel mixer and the Q of the quadrature demodulator Each local signal input to the channel mixer A switching means for selecting one of a non-switching state to be input as it is and a switching state in which one of the respective local signals is inverted and input, and a transmission I channel by a transmitter-side correction coefficient on the input side of the quadrature modulator A transmitter-side mismatch equalizer for correcting in advance a mismatch between a signal and a transmission Q-channel signal, and a receiver for correcting a mismatch between the reception I-channel signal and the reception Q-channel signal by a receiver-side correction coefficient on the output side of the quadrature demodulator A first transfer characteristic is estimated by switching the swapping switch and the switching means in a side mismatch equalizer and a path from the input side of the quadrature modulator to the output side of the quadrature demodulator via the feedback path. The transmitter side correction function for improving the IRR is calculated using the estimation result, and the transmitter side correction function is calculated. And calculating the transmission side correction coefficient for correcting the mismatch between the transmission I channel signal and the transmission Q channel signal in advance on the input side of the quadrature modulator, and then correcting the transmission I channel with the transmitter side correction coefficient A signal and the transmission Q channel signal are input to the quadrature modulator, and the second switching unit is switched in the path from the input side of the quadrature modulator to the output side of the quadrature demodulator via the feedback path. A mismatch estimation unit that calculates a receiver-side correction coefficient after calculating a receiver-side correction function for improving IRR using the result of the estimation again. Features.
According to a fifth aspect of the present invention, in the RF transceiver according to the fourth aspect, the transmitter correction function is calculated in a normal state 1 in which the swapping switch is in a normal state and the switching means is in a non-switching state. Normal state 2 in which the swapping switch is in a normal state and the switching means is in a switching state, swap state 1 in which the swapping switch is in a swap state and the switching means is in a non-switching state, and the swapping switch is in a swap state And the swap state 2 in which the switching means is switched to each other is set and calculated using four transfer characteristics obtained in each state.
According to a sixth aspect of the present invention, in the RF transmitting / receiving apparatus according to the fifth aspect, the calculation of the receiver-side correction function can be obtained by setting the normal state 1 and the normal state 2 again and obtaining each state 2 The calculation is performed using individual transfer characteristics.
The invention according to claim 7 comprises a transmitter having a block for controlling the phase of the output signal and a block for controlling the amplitude of the output signal, and a quadrature demodulator for quadrature demodulating the received I channel signal and the received Q channel signal. And a third transmission of a path from the input side to the output side of the transmitter and further from the input side to the output side of the quadrature demodulator. A receiver side correction function for improving IRR is calculated using the estimation result, and the received I channel signal is output on the output side of the quadrature demodulator using the receiver side correction function. And a receiver side correction coefficient for correcting a mismatch between the received Q channel signals.
According to an eighth aspect of the present invention, in the IQ mismatch correction method according to the seventh aspect, the calculation of the receiver-side correction function is performed by using respective local signals input to the I channel mixer and the Q channel mixer of the quadrature demodulator. The normal state 3 in which the polarity is not inverted and the normal state 4 in which one of the local signals input to the I channel mixer and the Q channel mixer of the quadrature demodulator is inverted in polarity are respectively set, and 2 obtained in each state. The calculation is performed using individual transfer characteristics.
The invention according to claim 9 comprises a transmitter having a block for controlling the phase of the output signal and a block for controlling the amplitude of the output signal, and a quadrature demodulator for quadrature demodulating the received I channel signal and the received Q channel signal. A feedback path forming means for forming a feedback path from an output side of the transmitter to an input side of the receiver, an I channel mixer and a Q channel of the quadrature demodulator On the output side of the quadrature demodulator, switching means for selecting one of a non-switching state in which each local signal input to the mixer is input as it is, and a switching state in which one of the local signals is input with the polarity inverted. Receiver-side mismatch that corrects mismatch between received I-channel signal and received Q-channel signal using receiver-side correction coefficient A third transfer characteristic is estimated by switching the switching means in a path from the transmitter to the output side of the quadrature demodulator of the receiver via the feedback path forming means, and a result of the estimation And a mismatch estimation unit that calculates a receiver-side correction coefficient after calculating a receiver-side correction function for improving IRR.

  According to the present invention, closed IQ mismatch correction is performed within a single RF transceiver, so that overhead due to the presence of a loopback dedicated to calibration and a plurality of local oscillators can be eliminated. Further, since the transfer characteristics are estimated by setting the swapping switch and switching means, it is not necessary to take a correlation between the transmission signal and the feedback signal. Therefore, IQ mismatch correction can be performed for both the transceiver and the receiver that can be easily mounted by taking advantage of the miniaturized CMOS process.

It is a block diagram of the RF transmission / reception apparatus of the Example of this invention. It is a block diagram of an example of the local signal source of the quadrature modulator of the RF transmission / reception apparatus of FIG. FIG. 3 is an operation waveform diagram of the local signal source of FIG. 2. FIG. 2 is an operation waveform diagram of a local signal source in normal states 1 and 2 and swap states 1 and 2 of the RF transceiver of FIG. 1. It is explanatory drawing of the equivalent mathematical model in the normal states 1 and 2 of RF transmitting / receiving apparatus of FIG. It is explanatory drawing of the equivalent mathematical model in the swap states 1 and 2 of RF transmission / reception apparatus of FIG. It is explanatory drawing of the impulse response model in the normal states 1 and 2 of the RF transmission / reception apparatus of FIG. It is explanatory drawing of an example of the transmitter side mismatch equalizer of RF transmission / reception apparatus of FIG. It is explanatory drawing of an example of the receiver side mismatch equalizer of RF transmission / reception apparatus of FIG. It is a frequency characteristic figure of an example of the transmission side IRR of RF transmitting / receiving apparatus of FIG. It is a frequency characteristic figure of an example of the receiving side IRR of RF transmitting / receiving apparatus of FIG. It is a block diagram of the conventional RF receiver. It is a block diagram of the RF transmission / reception apparatus (patent document 1) of a 1st prior art example. It is a block diagram of the RF transmitter (nonpatent literature 1, 2) of the 2nd prior art example. It is explanatory drawing of the equivalent mathematical model of RF transmitter of FIG. It is a block diagram of an example of the transmitter side mismatch equalizer (nonpatent literature 1) of RF transmitter of FIG. It is a block diagram of an example of the transmitter side mismatch equalizer (nonpatent literature 2) of RF transmitter of FIG. It is explanatory drawing of the impulse response model of the loop back path | pass of RF transmitter of FIG. It is a block diagram of RF transmission / reception apparatus (2nd patent document) of the 3rd prior art example. FIG. 20 is a configuration diagram of a mismatch estimator of the RF transmitting / receiving apparatus of FIG. 19.

The IQ mismatch correction method of the present invention corrects an IQ mismatch on the transmitter side, and then corrects a mismatch on the receiver side. First, the first transfer characteristic of the path from the input side to the output side of the quadrature modulator and further from the input side to the output side of the quadrature demodulator is estimated. Using the estimation result, a transmitter-side correction function for improving IRR is calculated, and the transmitter-side correction function is used to transmit a transmission I channel signal and a transmission Q channel signal in advance on the input side of the quadrature modulator. The transmitter correction function for correcting the mismatch is calculated. Next, the transmission I channel signal and Q channel signal corrected by the transmitter side correction coefficient are input to the quadrature modulator, and the path from the input side of the quadrature modulator to the output side and further from the input side to the output side of the quadrature demodulator The second transfer function is estimated. And the receiver side correction function for IRR improvement is calculated using the result. Finally, using the receiver-side correction function, a receiver-side correction coefficient for correcting a mismatch between the received I channel signal and the received Q channel signal on the output side of the quadrature demodulator is calculated.
Specifically, the following method is used.
(1) A known signal as in the first conventional example is required, and a method having poor interoperability is not adopted, and IQ mismatch correction closed by an individual transceiver is performed.
(2) In order to support broadband communication such as the MoCA 2.0 standard, an orthogonal modulation / demodulation method such as a direct conversion method is adopted for both transmission and reception as in the third conventional example. In applications that aim to achieve low current consumption, such as mobile terminals, a method such as polar modulation or out-fading modulation is adopted on the transmission side, and a quadrature modulation / demodulation method such as a direct conversion method is adopted on the reception side. As a result, the overhead due to the presence of a plurality of local oscillators as in the calibration-dedicated loopback path and the second conventional example is eliminated at the same time.
(3) As an estimation method, a method using a statistical average, a least square method, or the like as introduced in the second conventional example is used. This eliminates the drawback of having to take a correlation between the transmission signal and the feedback signal as seen in the third conventional example.
(4) After IQ mismatch estimation on both the transmitter side and the receiver side, the transmitter-side correction function V _TX (ω) of (Equation 20) and a new receiver-side correction function found in the second conventional example V _RX (ω) is calculated. 16 and 17 shown as Tx equalizers are similarly prepared for Rx and inserted immediately before the transmitter and immediately after the receiver. Using this, the IQ mismatch on each of the transmitter side and the receiver side is corrected.
(5) In order to calculate the transmitter / receiver correction functions V _TX (ω) and V _RX (ω), a swapping switch as shown in FIG. 2 to be described later is adopted as the local signal source on the quadrature modulator side. . Furthermore, a mechanism that can invert the polarity of the local signal is provided on at least one of the I channel side and the Q channel side of the quadrature demodulator.
(6) After equalizing the IQ mismatch on the transmitter side, the IQ mismatch on the receiver side is then equivalent by assuming that the transmitter side is performing ideal quadrature modulation.

<Example>
FIG. 1 is a schematic diagram of an embodiment of a wideband RF transceiver that employs an orthogonal modulation / demodulation method such as a direct conversion method on both the transmitter side and the receiver side, which is the subject of the present invention. In FIG. 1, 1 is a mismatch estimator, 2Tx is a transmitter-side mismatch equalizer, 2Rx is a receiver-side mismatch equalizer, 3TxI is a transmitter-side I-channel DA converter, 3TxQ is a transmitter-side Q-channel DA converter, and 3RxI is Receiver side I channel AD converter, 3RxQ is receiver side Q channel AD converter, 4TxI is transmitter side I channel low pass filter, 4TxQ is transmitter side Q channel low pass filter, 4RxI is receiver side I channel low pass filter, 4RxQ is Receiver side Q channel low pass filter, 5TxI is transmitter side I channel mixer, 5TxQ is transmitter side Q channel mixer, 5RxI is receiver side I channel mixer, 5RxQ is receiver side Q channel mixer, 6 is local oscillator, 7Tx Is the local signal source on the quadrature modulator side, 7 x is a quadrature demodulator side local signal source, 8 is a transmitter side adder, 9 is a coupler, 10Tx is a transmitter side power amplifier, 10Rx is a receiver side low noise amplifier, 11 is a loopback amplifier, 12 is an antenna, SW1 is The antenna changeover switch SW2 is a feedback loop forming switch.

The quadrature modulator includes an I channel mixer 5TxI, a Q channel mixer 5TxQ, a local signal source 7Tx, and an adder 8. The quadrature demodulator includes an I channel mixer 5RxI, a Q channel mixer 5RxQ, and a local signal source 7Rx. Also, here, the mismatch components are indicated by transmitter-side amplitude mismatch g _TX , transmitter-side phase mismatch φ _TX , receiver-side amplitude mismatch g _RX , and receiver-side phase mismatch φ _RX . Further, at the time of mismatch estimation described later, the switch SW2 is switched to the loopback amplifier 11 side as shown in the figure.

  FIG. 2 shows a detailed circuit of the local oscillator 6 and the local signal source 7Tx on the quadrature modulator side in the present invention. The local oscillator 6 includes a voltage controlled oscillator 61 and a divide-by-two 62 that divides the sine wave oscillated therein by two. The transmitter-side local signal source 7Tx includes DFF circuits 71 and 72 connected to form a frequency divider, swapping switches 73 and 74, and DFF circuits 75I and 75Q that function as output latches. As described above, in this embodiment, in order to realize IQ mismatch equivalence on both the transmitter side and the receiver side, the swapping switches 73 and 74 are provided in the local signal source 7Tx constituting the quadrature modulator on the transmitter side. Keep it.

  When the swapping switches 73 and 74 are switched to the side indicated by the solid line, the differential outputs AP and AN of the DFF circuit 71 are input to the latch 75I, and the differential outputs BP and BN of the DFF circuit 72 are input to the latch 75Q. It becomes a normal state to do. When switching to the side indicated by the broken line, the differential output AP, AN of the DFF circuit 71 is input to the latch 75Q, and the differential output BP, BN of the DFF circuit 72 is input to the latch 75I. That is, in the swap state, the I channel local signal and the Q channel local signal are switched. In this embodiment, the behavior shown in the timing chart of FIG. 3 is obtained, and a local signal having a frequency that is ¼ of the output clock CKP of the voltage controlled oscillator 61 is finally generated. Further, the orthogonal demodulator side local signal source 7Rx is provided with switching means (not shown) that can invert the local signal (exchange the differential local signal) at least on either the I channel side or the Q channel side. deep. The switching means can be realized by means similar to the swapping switches 73 and 74, for example. In that case, switching of the polarity is realized by switching the signal on the plus side and the minus side of I (or Q) which is a differential signal.

FIG. 4 shows the relationship of the output waveforms of the local signal sources 7Tx and 7Rx when the swapping switches 73 and 74 in FIG. 2 are in the normal state (solid line) and the swap state (broken line). In the figure, the transmitter side phase mismatch φ _TX and the receiver side phase mismatch φ _RX are clearly shown. However, here, in the orthogonal demodulator-side local signal source 7Rx, switching means is provided that can invert the local signal on the I channel side.

5 is an equivalent mathematical model when the swapping switches 73 and 74 in FIG. 2 are in the normal state (solid line), and FIG. 6 is an equivalent mathematical model when the swapping switches 73 and 74 in FIG. 2 are in the swap state (broken line). It is an equivalent mathematical model. However, l = l _RE + jl _{IM in} FIGS. 5 and 6 represents a signal transfer characteristic of the coupler 9 and the loopback amplifier 11, that is, a complex impulse response.

Here, when the transmitter-side mismatch equalizer 2Tx and the receiver-side mismatch equalizer 2Rx have single unit impulse characteristics, that is, s _I = p _I , s _Q = p _Q and y _I = z _I , y when _Q = z _Q, consider the signal transmission in the time domain in the path of the s _I in FIG. 5 showing the normal state, the s _Q to z. First, r _I and r _Q in FIG. 5 are expressed as (Equation 21) using s _I and s _Q and l _RE and l _IM .

Considering this, when considering signal transmission in the time domain in the path from s _I and s _Q to z, the transfer characteristic is as shown in (Equation 22).

When this is Fourier transformed, (Equation 23) is obtained. However, the description of the variable ω is omitted to avoid complexity of the mathematical formula.

Further, in FIG. 5 showing the normal state, for example, signal transmission in the time domain in the path from s _I and s _Q to z when the local input on the I channel side of the orthogonal demodulator side local signal source 7Rx is inverted is performed. Considering (Equation 21), the transfer characteristic is as shown in (Equation 24).

When this is Fourier transformed, (Equation 25) is obtained. However, the description of the variable ω is omitted to avoid complexity of the mathematical formula.

Next, when the transmitter-side mismatch equalizer 2Tx and the receiver-side mismatch equalizer 2Rx have single unit impulse characteristics, that is, s _I = p _I , s _Q = p _Q and y _I = z _I , y Consider the signal transmission in the time domain in the path from s _I and s _Q to z in FIG. 6 showing the swap state when _Q = z _Q. First, r _I and r _Q in FIG. 6 are expressed as (Equation 26) using s _I and s _Q and l _RE and l _IM .

Considering this, when considering signal transmission in the time domain in the path from s _I and s _Q to z, the transfer characteristic is as shown in (Equation 27).

When this is Fourier transformed, (Equation 28) is obtained. However, the description of the variable ω is omitted to avoid complexity of the mathematical formula.

In FIG. 6 showing the swap state, for example, signal transmission in the time domain in the path from s _I , s _Q to z when the local input on the I channel side of the orthogonal demodulator side local signal source 7Rx is inverted is performed. Considering (Equation 26), the transfer characteristic is as shown in (Equation 29).

When this is Fourier transformed, (Equation 30) is obtained. However, the description of the variable ω is omitted to avoid complexity of the mathematical formula.

Here, in (Formula 23), (Formula 25), (Formula 28), and (Formula 30), H _α-N1 , H _β-N1 , H _α-N2 , H _β-N2 , H _α-S1 , H _β-S1 , H _α-S2 and H _β-S2 were defined. Similar to the case of the impulse response model shown in FIG. 18 considered in the second conventional example,
(Normal state 1): The normal input on the I channel side of the quadrature demodulator is normal in the normal state (Normal state 2): The local input on the I channel side of the quadrature demodulator is inverted in the normal state (Swap state 1): Swap The I-channel local input of the quadrature demodulator in the state is normal (swap state 2): The impulse response model in the four states in which the local input on the quadrature demodulator I-channel is inverted in the swap state is as shown in FIG. When expressed, these are given by (Equation 31).

However, F represents a Fourier transform. Further, ^ represents an estimated value, which is a value to be estimated by a method using a statistical average, a least square method, or the like as in the case of the second conventional example.

Here, the following function of (Expression 32) is considered as V _TX (ω), and H _α-N1 , H _β-N1 in (Expression 23), (Expression 25), (Expression 28), and (Expression 30), H _alpha-N2, so the last term of the _{H β-N2, H α-} S1, H β-S1, H α-S2 and H _beta-S2 assigns a definition is calculated (equation 32). However, the description of the variable ω is omitted to avoid complexity of the mathematical formula.

This calculation result is the same as in (Equation 17), and can be regarded as a correction function that improves IRR on the transmitter side. That is, using the calculation result of V _TX (ω), it is appropriately divided into C _I (ω) and C _Q (ω) and inverse Fourier transformed, and the transmitter side mismatch equalizer 2Tx in FIG. 16 or the circuit shown in FIG. 17 can be configured.

  In the above discussion, it is assumed that the physical wiring of the switch connection in the normal state and the swap state of the swapping switches 73 and 74 is equivalent (equal length and equal load). When actually implemented on a semiconductor integrated circuit, this may not be equivalent, but at this time, an error occurs in the above discussion. Therefore, when implemented, it is preferable to mount so that the physical wiring of the switch connection is as equivalent as possible. The same can be said for the physical wiring when the local signal is inverted.

In the above example of IQ mismatch correction on the transmitter side, the polarity of the local input on the I channel side of the quadrature demodulator can be switched between forward and reverse,
(Normal state 1): The normal input on the I channel side of the quadrature demodulator is normal in the normal state (Normal state 2): The local input on the I channel side of the quadrature demodulator is inverted in the normal state (Swap state 1): Swap In the state, the local input on the I channel side of the quadrature demodulator is normal (swap state 2): In the swap state, the local input on the I channel side of the quadrature demodulator is inverted and the impulse response model shown in FIG. The impulse response was estimated and corrected using the estimation results.

However, instead, the polarity of the local input on the Q channel side of the quadrature demodulator can be switched between forward and reverse,
(Normal state 1): The normal input on the quadrature demodulator Q channel side is normal in normal state (Normal state 2): The local input on the quadrature demodulator Q channel side is inverted in normal state (swap state 1): Swap In the state, the local input on the Q channel side of the quadrature demodulator is normal (swap state 2): In the swap state, the local input on the Q channel side of the quadrature demodulator may be inverted.

In addition, the polarity of the local input on the I and Q channel side of the quadrature demodulator can be switched between forward and reverse,
(Normal state 1): The normal input on the I and Q channel sides of the quadrature demodulator is normal in the normal state. (Normal state 2): The local input only on the I channel side of the quadrature demodulator is inverted in the normal state (Swap state 1) ): Local input on the I and Q channel sides of the quadrature demodulator in the swap state is normal (swap state 2): The local input only on the Q channel side of the quadrature demodulator in the swap state may be inverted.

In addition, the polarity of the local input on the I and Q channel side of the quadrature demodulator can be switched between forward and reverse,
(Normal state 1): In normal state, the local input of the quadrature demodulator on the I and Q channel sides is normal (Normal state 2): In the normal state, the local input only on the Q channel side of the quadrature demodulator is inverted (swap state 1) ): The local input on the I and Q channel sides of the quadrature demodulator in the swap state is normal (swap state 2): The local input only on the I channel side of the quadrature demodulator in the swap state may be inverted. Thus, with the same concept, the result of the final term of (Equation 32) can be derived, and the IQ mismatch on the transmitter side can be corrected.

Thus, the IQ mismatch correction on the transmitter side is completed, and the IQ mismatch correction on the receiver side is performed as the next step. Therefore, after the transmitter-side mismatch equalizer 2Tx is appropriately set, the same estimation is performed in the normal state 1 and the normal state 2. In FIG. 5, which shows an equivalent model of the normal state, r _I and r _Q are expressed by (Equation 21) using s _I , s _Q and l _RE and l _IM . Since IQ mismatch correction is completed, it can be equivalently expressed by (Equation 33) (can be regarded as φ _TX = 0, g _TX = 1, h _TXI = h _TXQ = h _TX ) .

Considering this, considering the signal transmission in the time domain in the path from s _I and s _Q to z in the normal state 1, when an expression corresponding to (Expression 23) is derived, (Expression 34) is obtained. .

Similarly, considering signal transmission in the time domain in the path from s _I and s _Q to z in the normal state 2, when an expression corresponding to (Expression 25) is derived, (Expression 35) is obtained.

Here, in (Expression 34) and (Expression 35), H _α-N1-post , H _β-N1-post , H _α-N2-post and H _β-N2-post are defined. As in (), the impulse response model is a value to be expressed by (Equation 36) using the impulse response estimated in FIG.

Here, the following function of (Expression 37) is considered as V _RX (ω), and H _α-N1-post , H _β-N1-post , H _{α-N2-post in} (Expression 34) and (Expression 35) are considered. Substituting and defining the definition of H _β-N2-post gives the final term of (Equation 37). However, the description of the variable ω is omitted to avoid complexity of the mathematical formula.

This calculation result can be regarded as corresponding to (Equation 17) on the transmitter side of the correction function on the receiver side, and can also be used as a function for improving IRR as in the case of the transmitter. In other words, using the calculation result of V _RX (ω), it is appropriately divided into C _I (ω) and C _Q (ω) and inverse Fourier transformed to obtain the transmitter side mismatch equalizer 2Tx in FIG. 16 or the receiver side mismatch equalizer 2Rx can be configured based on the same concept as the circuit shown in FIG.

  Note that it is possible to employ an architecture that generates an output signal from phase and amplitude instead of IQ quadrature modulation as a transmitter surrounded by a block 100 as the RF transceiver shown in FIG. For example, there are techniques such as a polar transmitter and an out-fading transmitter. When such an architecture that does not have a quadrature modulation unit and achieves low current consumption and is changed to an RF transceiver that employs a quadrature demodulator as a receiver, the problem of IQ mismatch on the transmitter side does not occur. Absent. Therefore, the IQ mismatch of the receiver is corrected by using only the means described in the paragraphs 0069 to 0074 in the present specification by regarding the IQ mismatch on the transmitter side as being corrected. Can do.

<Verification>
The correctness of the above-mentioned invention was demonstrated by numerical simulation in a form compliant with the MoCA 2.0 standard. However, the simulation conditions are as follows.
(1) The mismatch parameters g _TX , φ _TX , g _RX , and φ _RX shown in FIG. 1 are 1.05, 5 °, 0.95, and −5 °, respectively.
(2) The low-pass filters 4TxI and 4TxQ shown in FIG. 1 are fourth-order Butterworth type low-pass filters, and mismatches are set so as to have cutoff frequencies of 50 MHz and 52 MHz, respectively.
(3) The low-pass filters 4RxI and 4RxQ shown in FIG. 1 are 7th-order Butterworth low-pass filters, and mismatches are set so as to have cutoff frequencies of 50 MHz and 48 MHz, respectively.
(4) The transfer characteristics of the coupler 9 and the loopback amplifier 11 in the loopback path shown in FIG. 1 are arbitrarily set as (√1 / 2) × (1−j) over all frequencies ( l = (√1 / 2) × (1−j)).
(5) The sampling frequency of the DA converters 3TxI and 3TxQ and the AD converters 3RxI and 3RxQ shown in FIG. However, the data samples are not quantized.
(6) The impulse responses shown in FIG. 7 are all estimated as 64 taps, a random QPSK signal is generated for each subcarrier of the OFDM signal, and estimation is performed by the least squares method. V _TX (ω) was obtained, and then V _RX (ω) represented by (Equation 37) was obtained. However, at this time, V _TX (ω) and V _RX (ω) are obtained as discrete values every 200/64 [MHz].
(7) The transmitter-side mismatch equalizer 2Tx is configured as shown in FIG. 21 is a 32-clock delay device, and 22 is a 64-tap FIR filter. Here, the multiplier λ _TX is (imaginary part) / (real part), that is, sinφ _TX / cosφ _TX in one item of discrete inverse Fourier transform of 1 / V _TX (ω). Further, the coefficient f _TX [0:63] of the 64-tap FIR filter 22 is expressed as f _TX [0:63] = (Re {v) where the discrete inverse Fourier transform of V _TX (ω) is v _TX [1:64]. _TX [33:64]}, Re {v _TX [1:32]}). However, Re represents a real part. The FIR filter 22 and the delay device 21 operate at 200 MHz.
(8) Unlike the non-patent document 2, the receiver-side mismatch equalizer 2Rx is configured as shown in FIG. 9 by changing the order of the FIR filter and multiplication operation. 23 is a 32-clock delay device, and 24 is a 64-tap FIR filter. The multiplier λ _RX is (imaginary part) / (real part), that is, sinφ _RX / cosφ _RX in one item of the discrete inverse Fourier transform of 1 / V _RX (ω). Further, the coefficient f _RX [0:63] of the 64-tap FIR filter 24 is expressed as f _TX [0:63] = (Re {v) when the discrete inverse Fourier transform of V _RX (ω) is v _TX [1:64]. _TX [33:64]}, Re {v _TX [1:32]}). However, Re represents a real part. The FIR filter 24 and the delay device 23 operate at 200 MHz.

  The effects of improving the IRR on the receiver side and the transmitter side in the simulation are shown in FIGS. 10 and 11, respectively. In each figure, legend Δ is before the insertion of the equalizer, and legend □ is after the insertion of the equalizer. It can be seen that both the receiver side and the transmitter side show improved IRR over the entire frequency range.

1: mismatch estimator 2Tx, 2Rx: mismatch equalizer 3TxI, 3TxQ: DA converter 3RxI, 3RxQ: AD converter 4TxI, 4TxQ, 4RxI, 4RxQ: low-pass filter 5TxI, 5TxQ, 5RxI, 5RxQ: mixer 6: local oscillator, 61: Voltage controlled oscillator, 62: 2 frequency divider 7Tx, 7Rx: Local signal source, 71, 72: DFF circuit, 73, 74: Swap switch, 75I, 75Q: DFF circuit 8: Adder 9: Coupler 10Tx, 10Rx: Amplifier 11: Loopback amplifier 12: Antenna 100: Transmitter that does not perform IQ quadrature modulation, such as a polar transmitter or an out-fading transmitter

Claims (9)

In an RF transmission / reception apparatus including a quadrature modulator that orthogonally modulates a transmission I channel signal and a transmission Q channel signal on the transmitter side and an orthogonal demodulator that orthogonally demodulates the reception I channel signal and the reception Q channel signal on the receiver side IQ mismatch correction method,
A first transfer characteristic of a path from the input side to the output side of the quadrature modulator and further from the input side to the output side of the quadrature demodulator is estimated, and the transmission I channel signal is preliminarily set on the input side of the quadrature modulator. And a transmitter-side correction coefficient for correcting a mismatch between the transmission Q channel signals and
Next, the transmission I channel signal and the transmission Q channel signal corrected by the transmitter side correction coefficient are input to the quadrature modulator, and from the input side to the output side of the quadrature modulator, and further input to the quadrature demodulator The second transfer characteristic of the path from the side to the output side is estimated again, and using the result of the estimation again, a receiver-side correction function for improving IRR is calculated, and the receiver-side correction function is used. A receiver-side correction coefficient for correcting a mismatch between the received I channel signal and the received Q channel signal on the output side of the quadrature demodulator is calculated. The IQ mismatch correction method according to claim 1,
The calculation of the transmitter correction function is as follows:
Normal state in which the local signals input to the I-channel mixer and the Q-channel mixer of the quadrature modulator are not swapped, and the polarities of the local signals input to the I-channel mixer and the Q-channel mixer of the quadrature demodulator are not reversed. 1 and
Normal in which local signals input to the I-channel mixer and Q-channel mixer of the quadrature modulator are not swapped, and one of the local signals input to the I-channel mixer and Q-channel mixer of the quadrature demodulator is inverted. State 2 and
Swap state 1 for swapping local signals input to the I-channel mixer and Q-channel mixer of the quadrature modulator and not reversing the polarity of the local signals input to the I-channel mixer and Q-channel mixer of the quadrature demodulator;
Swap state in which local signals input to the I channel mixer and Q channel mixer of the quadrature modulator are swapped, and one of the local signals input to the I channel mixer and Q channel mixer of the quadrature demodulator is inverted in polarity. 2,
Is calculated using four transfer characteristics obtained in each state,
IQ mismatch correction method characterized by the above. The IQ mismatch correction method according to claim 2,
The calculation of the receiver side correction function is as follows:
The normal state 1 and the normal state 2 are set again, and calculation is performed using two transfer characteristics obtained in each state.
IQ mismatch correction method characterized by the above. In an RF transmission / reception apparatus including a quadrature modulator that orthogonally modulates a transmission I channel signal and a transmission Q channel signal on the transmitter side and an orthogonal demodulator that orthogonally demodulates the reception I channel signal and the reception Q channel signal on the receiver side ,
Feedback path forming means for forming a feedback path from the output side of the quadrature modulator to the input side of the quadrature demodulator;
A swapping switch that selects one of a normal state in which the local signals are input to the I-channel mixer and the Q-channel mixer of the quadrature modulator as they are and a swap state in which the local signals are input to each other by switching.
Switching to select one of a non-switching state in which each local signal input to the I channel mixer and Q channel mixer of the quadrature demodulator is input as it is and a switching state in which one of the local signals is input with the polarity inverted. Means,
A transmitter-side mismatch equalizer for correcting in advance a mismatch between a transmission I channel signal and a transmission Q channel signal by a transmitter-side correction coefficient at the input side of the quadrature modulator;
A receiver-side mismatch equalizer that corrects a mismatch between the received I-channel signal and the received Q-channel signal by a receiver-side correction coefficient at the output side of the quadrature demodulator;
A first transfer characteristic is estimated by switching between the swapping switch and the switching means in a path from the input side of the quadrature modulator to the output side of the quadrature demodulator via the feedback path, and the estimation result is obtained. A transmitter-side correction function for improving the IRR, and using the transmitter-side correction function, the mismatch between the transmission I channel signal and the transmission Q channel signal is corrected in advance on the input side of the quadrature modulator. Calculating the transmission side correction coefficient to be input, and then inputting the transmission I channel signal and the transmission Q channel signal corrected with the transmitter side correction coefficient to the quadrature modulator, and from the input side of the quadrature modulator The second transfer characteristic is reestimated by switching the switching means in the path that reaches the output side of the quadrature demodulator via the feedback path, and the estimation is performed again. Results using a mismatch estimation unit for calculating the receiver side correction coefficient from the calculated receiver-side correction function for the IRR improvement,
An RF transmitting / receiving apparatus comprising: The RF transmitting / receiving apparatus according to claim 4.
The calculation of the transmitter correction function is as follows:
Normal state 1 in which the swapping switch is in a normal state and the switching means is in a non-switching state;
A normal state 2 in which the swapping switch is in a normal state and the switching means is in a switching state;
A swap state 1 in which the swapping switch is in a swap state and the switching means is in a non-switching state;
A swap state 2 in which the swapping switch is in a swap state and the switching means is in a switching state;
Is calculated using four transfer characteristics obtained in each state,
An RF transmitting / receiving apparatus. The RF transmitting / receiving apparatus according to claim 5.
The calculation of the receiver side correction function is as follows:
The normal state 1 and the normal state 2 are set again, and calculation is performed using two transfer characteristics obtained in each state.
An RF transmitting / receiving apparatus. A transmitter having a block for controlling the phase of the output signal and a block for controlling the amplitude of the output signal, and a receiver having a quadrature demodulator for quadrature demodulating the received I channel signal and the received Q channel signal. An IQ mismatch correction method in an RF transceiver device,
A third transfer characteristic of a path from the input side to the output side of the transmitter and further from the input side to the output side of the quadrature demodulator is estimated, and a receiver for improving the IRR is used by using the estimation result Calculating a side correction function, and using the receiver side correction function, calculating a receiver side correction coefficient for correcting a mismatch between the received I channel signal and the received Q channel signal on the output side of the quadrature demodulator. IQ mismatch correction method characterized by this. The IQ mismatch correction method according to claim 7, wherein
The calculation of the correction function on the receiver side includes normal state 3 in which polarities of local signals input to the I channel mixer and Q channel mixer of the quadrature demodulator are not inverted, and the I channel mixer and Q channel mixer of the quadrature demodulator. Normal state 4 in which one of the local signals input to each of them is inverted in polarity, and calculated using two transfer characteristics obtained in each state.
IQ mismatch correction method characterized by the above. A transmitter having a block for controlling the phase of the output signal and a block for controlling the amplitude of the output signal, and a receiver having a quadrature demodulator for quadrature demodulating the received I channel signal and the received Q channel signal. In the RF transceiver,
Feedback path forming means for forming a feedback path from the output side of the transmitter to the input side of the receiver;
Switching to select one of a non-switching state in which each local signal input to the I channel mixer and Q channel mixer of the quadrature demodulator is input as it is and a switching state in which one of the local signals is input with the polarity inverted. Means,
A receiver-side mismatch equalizer that corrects a mismatch between the received I-channel signal and the received Q-channel signal by a receiver-side correction coefficient at the output side of the quadrature demodulator;
A third transfer characteristic is estimated by switching the switching unit in a path from the transmitter to the output side of the quadrature demodulator of the receiver via the feedback path forming unit, and the estimation result is used. A mismatch estimation unit that calculates a receiver-side correction coefficient after calculating a receiver-side correction function for improving IRR;
An RF transmitting / receiving apparatus comprising:

Images