Low noise I-Q mixer

[0002] The present invention relates to a low noise I-Q phase-quadrature modulator. More particularly, the invention relates to a low noise I-Q phase-quadrature for applications in radio frequency signal receivers, of the type comprising a plurality of Gilbert cell input stages driven by a voltage feed line receiving respective square wave command signals in input from a local oscillator.

[0003] As is well known, in recent years there has been notable progress and changes in the world of telecommunications. In particular, the development of new integrated electronic devices on semiconductors has contributed to a large number of these changes.

[0004] Amongst these improvements is the reduction of power consumption in portable electronic equipment. This reduction of energy consumption has permitted a reduction in weight and dimensions of this portable equipment, lengthening at the same time their useful life.

[0005] All this has resulted in a large decrease in costs, something which has made this type of portable electronic equipment accessible to a large part of the population as for example mobile cellular telephones or portable computers.

[0006] The effort to reduce the size of electronic circuitry is continuing. For example, various studies are under way at present for a new type of a radio frequency receiver RF that could be very compact.

[0007] At the moment, many designers are working on receivers based on an architecture that includes a rejection-type image mixer to permit the reduction of the stringent specifications of a band-pass filter RF, downstream, therefore making it possible to have a significant decrease in the relative costs.

[0008] An architecture of rejection-type mixer of the image band of the known type has been proposed by Hatley and is represented in FIG. 1. This architecture allows the conversion to the frequency RF of a desired band, Sw(f) of the RF signal, suppressing the image band, Su(f).

[0009] A signal 20 coming from a source LO of a local oscillator is applied to a network 21 of quadrature producing two signals 22, 23 out of phase shifted the one to the other by 90°. Such signals are applied to an I-Q modulator 24. The modulator 24 produces two output signals 25, 26 which are further out of phase in order to be applied to the input of a combiner node 27 obtaining useful signals in phase and image signals in counter-phase. In the absence of errors between the two distinct signal pathways, the image signal of the band IS at intermediate frequency is totally eliminated.

[0010] In a real mixer, nevertheless, the presence of module and phase errors in gains of the two signal pathways determines an incomplete suppression of the image at intermediate frequency. For example, the rejection requisite of image of 30 dBc requires at the same time a phase error which is less than 3° and an amplitude error which is less than 0.3 dB (3.5%).

[0011] An embodiment of the present invention includes a current I-Q modulator having structural and functional characteristics that would allow the use of an image rejection-type mixer with a very low level of introduction of noise.

[0012] The modulator can operate in reception as if there was a “fusion” of the input stages formed by the Gilbert cells, therefore reducing at the same time the polarization noise of each input stage.

[0013] The characteristics and the advantages of a modulator according to the invention will become clear from the description hereinbelow, of an example of embodiment reported as indication and not limitation thereto of the invention, which makes reference to the attached drawings.

[0014] FIG. 1 shows a schematic view of an I-Q modulator realized according to the prior art.

[0015] FIG. 2 shows a schematic view in circuit detail of the modulator of FIG. 1.

[0016] FIG. 3 shows a schematic view of an I-Q modulator according to the present invention.

[0017] With reference to such figure, and in particular to the example in FIG. 3, a low-noise modulator 1 is wholly and schematically indicated, which is realized according to the present invention.

[0018] The phase-quadrature modulator 1 according to the invention structurally presents a mirrored symmetry which is highlighted by the letters A and B, which respectively indicate circuit stages relative to signals in phase, indexed I, and in quadrature, indexed Q.

[0019] The structure of the modulator 1 can, though, be considered also from another point of view, and in particular as a circuit device essentially comprising two functional blocks 2, 3. The first block 2 is destined to carry out a voltage-current conversion, while the other block 3 is destined to carry out a frequency conversion.

[0020] The voltage-current conversion block 2 comprises a double couple of transistor Q1a, Q2aand Q1b, Q2b; a resistance RE1 of degeneration and a pair of current generators I_B1.

[0021] The transistor Q1a, Q2aand Q1b, Q2bare all of the NMOS type, even if there is nothing preventing the realization of a dual circuit incorporating PMOS transistors. The transistor Q1aand Q1bhave respective source terminals. Amongst such source terminals and a reference voltage Vee there is one of the current generators I_B1.

[0022] The transistors Q2aand Q2balso have respective gate terminals in common, as they also have respective source terminals. Between such source terminals and the Vee voltage reference is the other current generator I_B1.

[0023] The resistance RE1 of degeneration is connected to the source terminals of transistor Q1a, Q1band Q2a, Q2b.

[0024] Block 3, which carries out the frequency conversion, comprises a pair of Gilbert cells 4 and 5.

[0025] The Gilbert cells are essentially differential cells with a particular structure, but are of known type. For example, Gilbert cells similar to those of the modulator to according to the invention are also incorporated in the known type of modulator illustrated in FIG. 2.

[0026] The first cell 4 comprises the transistors from Q3aand Q6ato Q6b. Each cell 4, 5 receives in input a respective square wave signal V_LO,I; V_LO,Qproduced by a local oscillator LO, as the one shown in FIG. 1.

[0027] Cells 4 and 5 are fed by a line 6 of Vcc feed voltage. Furthermore, charge resistances Rc are foreseen between the line 6 and each branch of the cells 4, 5 in order to correctly polarize the transistors incorporated in the cells themselves. The first cell 4 comprises two circuit nodes A1, A2 connected to the voltage-current conversion block 2. The first node A1 is connected to the drain of the transistor Q1a, while the second node A2 is connected to the terminal of drain of the transistor Q2a. The second cell 5 comprises two circuit nodes B1, B2 connected to the voltage-current conversion block 2. The first node B1 of the second cell is connected to the drain of the transistor Q1b, while the second node B2 is connected to the drain terminal of the transistor Q2b. Advantageously, according to the invention, the modulator 1 also comprises a plurality of sink resistors. In particular a resistance R_Bis foreseen between each of the nodes A1, A2, B1, B2 and the Vcc feed voltage reference.

[0028] For ease of illustration the operation of a modulator according to the invention will be now described, although only making reference to section A, as sections A and B are identical. A voltage radio frequency signal VRF is converted into current by the voltage-current converter 2 comprising the transistors Q1a, Q2aand of the resistance RE1. The current so generated concerns the Gilbert cell 4 comprising the transistors Q3a-Q6awhich invert or not the phase of the signal according to the command in voltage V_LO,Ireceived from the local oscillator. The signal of out-going current from cell 4 is therefore converted into a voltage signal via the charge resistances Rc. The operation which results is an analogic multiplication of the radio frequency signals V_RFfor the square wave signal V_LO,I. The resulting signal V_IF,1is formed by various harmonic components and is generally filtered in order to extract the intermediate frequency component desired, which is the difference of the fundamental frequencies of the signals V_RFand V_LO,I. The resistances of the sinks R_Ballows, as will be explained later, the reduction of the polarization current of the cell 4 with the objective of decreasing the noise it generates. In order to highlight the advantages of the proposed solution it is useful to compare it to that of a traditional I-Q modulator with separate voltage-current converters, as illustrated in FIG. 3. For this reason it is convenient to carry out this comparison on equal power dissipation, that is requiring the current generators I_B1=2 I_B2. In such conditions, the relations of respective conversion gains for the circuits presented in FIGS. 2 and 3 are respectively: $\begin{array}{cc}{G}_1\equiv \frac{V_{\mathrm{IF}}}{V_{\mathrm{RF}}}=\frac{2R_c}{\pi R_{\mathrm{E1}}}\frac{g_m{R}_{\mathrm{E1}}}{1+g_m{R}_{\mathrm{E1}}}& \left(1\right)\\ G_2\equiv \frac{V_{\mathrm{IF}}}{V_{\mathrm{RF}}}=\frac{2R_c}{\pi R_{\mathrm{E2}}}\frac{g_m{R}_{\mathrm{E2}}}{1+\frac{g_m{R}_{\mathrm{E2}}}{2}}& \left(2\right)\end{array}$

[0029] Each of the relations (1) and (2) shows a term, dependent on transconductance g_m, which takes into account a not ideal transferal of the voltage-current converter. The requirement of linearity of the circuit depends upon such term. In fact, in conditions of ample signal, the transconductance g_mvaries considerably because of the strong variation of the current in the transistors, thus generating a dependence of the conversion gain from the amplitude of the signal itself.

[0030] As can be inferred by the expressions (1) and (2), with the condition R_E1=(½)R_E2the two modulators show the same requirement of linearity and the same conversion gain.

[0031] Therefore, at equal dissipation of power and with a degeneration resistance R_E1=(½)R_E2the modulator proposed by the disclosed embodiment of the present invention has the same linearity and the same gain of voltage as the traditional one.

[0032] The disclosed embodiment is more advantageous with respect to existing designs in terms of less noise. From this point of view, with the idea of comparing the two solutions, the mathematical relations for the equivalent resistance of input noise of the circuits represented in FIGS. 3 and 2 are respectively shown hereinbelow: $\begin{array}{cc}{R}_{\mathrm{EQ1}}=R_{\mathrm{E1}}+2r_b+\frac{1}{g_m}+\frac{R_{\mathrm{E1}}^2{\pi }^2}{2R_C}+2\frac{R_{\mathrm{E1}}^2}{R_B}+\frac{g_m{R}_s^2}{2\beta }+\gamma \left(r_{\mathrm{bq}}+\frac{1}{2g_{\mathrm{mq1}}}+\frac{g_{\mathrm{mq1}}{R}_{\mathrm{sq}}^2}{8}\right){\left(g_{\mathrm{mq1}}{R}_{\mathrm{E1}}\frac{\pi }{2}\right)}^2& \left(3\right)\\ R_{\mathrm{EQ2}}=R_{\mathrm{E2}}+2r_b+\frac{1}{g_m}+\frac{R_{\mathrm{E2}}^2{\pi }^2}{2R_C}+\frac{g_m{R}_s^2}{2\beta }+\gamma \left(r_{\mathrm{bq}}+\frac{1}{2g_{\mathrm{mq2}}}+\frac{g_{\mathrm{mq2}}{R}_{\mathrm{sq}}^2}{8}\right){\left(g_{\mathrm{mq2}}{R}_{\mathrm{E2}}\frac{\pi }{2}\right)}^2& \left(4\right)\end{array}$

[0033] wherein the following symbols are used:

[0034] r_bbase resistance of the transistors of the voltage-current converter

[0035] g_mtransconductance of the transistors of the voltage-current converter

[0036] RS resistance of the command source of the voltage-current converter

[0037] β current gain of the transistors of the voltage-current converter

[0038] γ coefficient dependent on the commutational speed of the phase-quadrature modulator

[0039] g_mqtransconductance of the transistors of the phase-quadrature modulator

[0040] r_bqbase resistance of the transistors of the phase-quadrature modulator

[0041] R_sqcommand source resistance of the phase-quadrature modulator.

[0042] Comparing the two expressions, and having set R_E1=(½)R_E2so as to restore the same conditions of linearity, the circuit in FIG. 3 features, with respect to traditional topology, a contribution of noise due to the degeneration resistance of the voltage-current converter (R_E1) divided by two and a contribution of noise due to the charge resistances (Rc) reduced by a factor ¼. Furthermore, due to the presence of the sink resistance R_B, which subtracts polarization current from the Gilbert cell and reduces the value of the transconductance G_mc, the contribution of noise due to cell 4 is less than that of the circuit in FIG. 2.

[0043] Clearly, the sink resistance R_Bitself generates a noise contribution which, in any case, given the values of RB commonly used (with R_B>>R_E1) is negligible.

[0044] In order to better understand the advantages of the solution according to the invention, it might be useful to find out the numeric values obtained with the expressions (3) and (4) for a typical circuit.

[0045] Having set: I_B1=3 mA; R_B=30Ω; R_E1=50Ω; R_B=4 KΩ; Rc 500Ω; R_S=100Ω; R_sq=30Ω; β=70 and γ=0, from the expressions (3) and (4) is obtained:

[0046] R_EQ1=376Ω and R_EQ1=2,4 KΩ

[0047] Or equivalently in terms of noise figure:

[0048] F₁=6,8 dB and F₂=14 dB.

[0049] Therefore, at equal dissipation and linearity, the modulator disclosed herein introduces a notably lower noise level than existing circuits.

[0050] The disclosed embodiment of the present invention finds a use in applications in which a low noise level is necessary, such as, for example, receivers with a passive antenna and, in general, receivers with particularly stringent noise specifications. The disclosed embodiment has the advantage of introducing, at work conditions being equal, a notably lower noise level with respect to that introduced by commonly used circuit solutions.