WO2009059831A2

WO2009059831A2 - Mixing apparatus

Info

Publication number: WO2009059831A2
Application number: PCT/EP2008/062246
Authority: WO
Inventors: Pete Sivonen; Sami Vilhonen
Original assignee: Nokia Corporation
Priority date: 2007-11-05
Filing date: 2008-09-15
Publication date: 2009-05-14
Also published as: WO2009059831A3; GB0721733D0

Abstract

A mixing arrangement comprising: an input configured to receive an input signal at a first frequency; first and second outputs configured to provide output signals; a first mixer connected between said input and said first output, said first mixer configured to mix said input signal with a first mixing signal; a second mixer connected between said input and said second output, said second mixer configured to mix said input signal with a second mixing signal; wherein said first mixer and said second mixer are configured such that in use, for at least part of the time only one of said first and second mixers is arranged to mix said input signal with a respective mixing signal.

Description

Mixing Apparatus

FIELD OF THE INVENTION

The present invention relates to a mixer apparatus and, in particular but not exclusively, to a mixing apparatus for use in a base station, mobile station or similar entity in a wireless telecommunications network.

BACKGROUND TO THE INVENTION

An area covered by a cellular telecommunications network is divided into a plurality of cells. Each of the cells has a base station arranged to transmit signals to and receive signals from mobile stations in the ceil associated with the respective base station. Mobile stations will be in active communication with the base station associated with the cell in which the mobile station is located.

Both mobile stations and base stations take received signals which are at a radio frequency and downconvert them to a baseband signal. The radio frequency signals are either directly downconverted to baseband or are downconverted via one or more intermediate frequencies. In order to downconvert the radio frequency a mixer is used. The mixer receives one input from the radio frequency signal to be downconverted and a second input which comprises a frequency component. This second input to the mixer is generated by a local oscillator. The mixer mixes the two inputs and the resulting signal output by the mixer will have the signal information contained in the first input and will be at a frequency which is typically the difference of the input frequencies. For example, if a signal which is to be downconverted has a frequency A (first input) and the signal with which it is to be mixed has a frequency B (second input), the mixer will output a frequency of A - B.

Zero-IF or direct conversion (henceforth zero-IF) radio receivers are starting to gain increasing attention. These receivers downconvert a received radio frequency signal directly to baseband. These receivers offer improved performance (for example by removing any problems associated with imaging in the frequency spectrum) and simplify circuit design. A block diagram of the analogue portions of a typical zero-IF receiver is shown in Figure 1. An antenna 2 feeds a received radio frequency (RF) signal to a bandpass filter 4 that selects a desired RF band. One or more low noise amplifiers (LNA) 6, amplify the filtered RF signal in order to reduce the effect of noise contributions in the following stages. The amplified and filtered signal is then split into I and Q paths and provided to mixers 8 and 9 on these two paths respectively. Mixers 8 and 9 downconvert the amplified and filtered RF signals to baseband. Radio frequency signals LOI and LOQ drive mixers 8 and 9.

A local oscillator 10 provides a radio frequency signal LO which is provided to phase shifter 11. This phase shifter receives this local oscillator signal LO and phase shifts the signal to produce signals LOI and LOQ which are phase shifted by 90 degrees from each other. The phase shifter may additionally amplify the signals LOI and LOQ.

Finally, the baseband signals are provided by mixers 8 and 9 to a filter stages 12 and 13 respectively. The resultant downconverted output signals (Vouτ,ι and VOU_T.Q) may subsequently be processed further.

In a typical zero-IF receiver, two mixers driven by separate, phase shifted, signals (LOl and LOQ) are required to enable effective processing of the baseband signals (V_Ouτ,ι and V_Ouτ,α).

Attention is similarly being given to low-IF architectures in which the received radio frequency signal is downconverted to a low, but non-zero frequency. For example, a low-IF mixer may typically downconvert a gigahertz range received radio frequency signal to an output signal of 3 megahertz. Low-IF architectures offer similar advantages to zero-IF architectures, however more complex signal processing may be required in the later stages of the receiver.

Increasing the level of integration in such receivers is desirable since it reduces the printed circuit board (PCB) area and complexity while lowering the component cost. This in turn provides low-power, low-cost, and high- performance circuitry.

Conventionally, bipolar and BiCMOS (integrated bipolar and CMOS) technology has been used for radio transceivers. However, Complementary metal oxide semiconductor (CMOS) technology is beginning to be used. This provides the advantage that it enables a complete systems-on-a-chip (SOC) to be provided.

Low-frequency flicker (1/f) noise and linearity are of particular concern in zero-lF receivers, especially in CMOS implementations and in narrow band applications such as GSM (Global System for Mobile communications). Consequently there is a need for a downconversion mixer with low level of 1/f noise and a high linearity.

CMOS mixers may be active or passive mixers. Active mixers are able to provide additional power gain to a received signal. However this benefit comes with associated drawbacks such as increased 1/f noise. Passive mixers are known in the art, however these are unable to provide the power gain of active mixers. This loss of signal strength requires additional amplification to be provided which often introduces noise and distortion such that the benefits of passive mixers are lost. For the benefits of passive mixers to be realized, the signal loss should be reduced.

Passive CMOS mixers normally commutate voltage, however this may lead to distortion due to the voltage swings within the mixer. Passive CMOS mixers are equally able to commutate current, that is to say the signal is carried as a current rather than a voltage.

SUMMARY OF THE INVENTION

It is an aim of some embodiments of the present invention to address one or more of the problems discussed above. According to an aspect of the invention there is provided a mixing arrangement comprising: an input configured to receive an input signal at a first frequency; first and second outputs configured to provide output signals; a first mixer connected between said input and said first output, said first mixer configured to mix said input signal with a first mixing signal; a second mixer connected between said input and said second output, said second mixer configured to mix said input signal with a second mixing signal; wherein said first mixer and said second mixer are configured such that in use, for at least part of the time only one of said first and second mixers is arranged to mix said input signal with a respective mixing signal.

Preferably said first and second mixers are substantially the same.

Preferably said input signal comprises a current signal.

Preferably said mixing arrangement is configured such that said first and second mixing signals have the same frequency.

Preferably said mixing arrangement is configured such that there is a difference in phase between said first and second mixing signals. More preferably said difference in phase is substantially equal to a quarter of a period of said mixing signals.

Preferably said first and second mixers comprise a plurality of switching elements. More preferably said mixers are configured such that said switching elements are operable for less than 35% of a period of said mixing signal. Yet more preferably said mixers are configured such that said switching elements are operable for substantially 25% of a period of said mixing signal.

Preferably said mixers comprise transistors. More preferably said first and second mixers each comprise four transistors. Yet more preferably said four transistors of each of said first and second mixers are arranged as pairs of transistors such that each pair of transistors operate concurrently. Yet more preferably said mixing arrangement is configured such that in use, for at least part of the time, a first pair of transistors of one of either said first or second mixer is on, and a second pair of transistors of said one of said first or second mixer is off. Yet more preferably said mixing arrangement is configured such that in use, for at least part of the time, the four transistors of one of said first and second mixer are off. Yet more preferably said transistors are driven by said mixing signals.

Preferably the mixing arrangement comprises a transconductance amplifier, the output of which is connected to said input.

Preferably the mixing arrangement comprises a first and a second transimpedance amplifier respectively connected to said first and second outputs.

Preferably said mixers comprise complementary metal oxide semiconductor elements.

In some embodiments of the invention there is provided an integrated circuit comprising the mixing arrangement as described above.

In some embodiments of the invention there is provided a radio receiver comprising the mixing arrangement as described above.

Preferably said radio receiver is a mobile communications device or a base station in a communications network.

According to another aspect of the invention there is provided a method of mixing comprising: receiving at an input an input signal at a first frequency; mixing at a first mixer connected to said input, said input signal with a first mixing signal; mixing at a second mixer connected to said input, said input signal with a first mixing signal; wherein for at least part of the time only one of said first and second mixers is arranged to mix said input signal with a respective mixing signal.

BRIEF DESCRIPTION OF DRAWINGS For a better understanding of the present invention and as to how the same may be put into effect, reference will now be made by way of example only to the accompanying drawings in which:

Figure 1 shows a block diagram of a typical zero-IF receiver; Figure 2 shows a single phase passive current-mode CMOS mixer arrangement;

Figure 3 shows a simulated LO signal with a 50% duty cycle;

Figure 4 shows a quadrature passive current-mode CMOS mixer according to an embodiment of the invention; Figure 5 shows simulated I and Q LO signals with a 50% duty cycle;

Figure 6 shows simulated I and Q LO signals with a 25% duty cycle according to embodiments of the invention;

Figure 7 shows quadrature passive current-mode CMOS mixer according to a further embodiment of the invention; Figure 8 shows an example of a mobile telecommunications network in which embodiments of the invention may be implemented.

DESCRIPTION OF EMBODIMENTS OF THE INVENTION To assist in the understanding of embodiments of the present invention a single phase passive current-mode CMOS mixer arrangement is illustrated in Figure 2.

In this arrangement, a transconductance amplifier 14 is provided. An input RF voltage (V_RF) is applied across input pins 16 and 18 of the transconductance amplifier and converted to an RF current (I_RF). V_RF may be provided by a low noise amplifier (not shown).

IRF is output by the transconductance amplifier 14 on first and second IRF wires 20 and 22, the first and second wires carrying equal and opposite currents. Capacitors 24 and 26 may be provided on these output lines and act to oppose the flow of any DC current.

IRF is provided to a switching quad which comprises four switching field effect transistors (FETs) 28, 30, 32 and 34. The switching quad is driven by a single- phase balanced local oscillator (LO) signal (V_Lo) and serves to downconvert the input RF current signal I_RF to a baseband output current signal I_BB which is provided as an output on first and second I_BB wires 36 and 38.

The mixer comprises input nodes 29 and 33 connected to IRF wires 20 and 22 respectively, and output nodes 31 and 35 connected to I_BB output wires 36 and 38 respectively.

First and second NMOS transistors 28 and 30 are connected at their sources to the first input node 29. These transistors are connected at their drains to the first and second output nodes 31 and 35 respectively.

Third and fourth NMOS transistors 32 and 34 are connected at their sources to the second input node 33. These transistors are connected at their drains to the second and first output nodes 35 and 31 respectively.

The first and third transistors (28 and 32 respectively) are driven at their gates by a first LO signal V_Lo+-

The second and fourth transistors (30 and 34 respectively) are driven at their gates by a second LO signal V|_o-

The two LO signals V_Lo+ and V_Lo- are inverses of one another and may be produced by one of many methods known in the art.

Overall, the switching quad with the V_Lo input forms a passive current-mode mixer 40 which serves to downconvert the I_RF to I_BB. This is done by commutating the input current I_RF between the two output wires 36 and 38 at a frequency determined by the LO signal.

The two I_BB wires 36 and 38 are each connected to respective inputs of a transimpedance amplifier 42. This transimpedance amplifier comprises an operational amplifier (OPA) 44 and first and second resistor-capacitor (RC) feedback circuits 46 and 48. The ]_BB wires are provided as respective inputs to the OPA 44 with I_BB wire 36 being to the negative input of the OPA 37 and IBB wire 38 being to the positive input of the OPA.

The OPA is a fully differential OPA and therefore comprises the two inputs and two outputs, each output corresponding to one of the inputs, such that the negative output will correspond to the positive input and vice versa. Each of the first and second RC circuits 46 and 48 are connected between one input and a corresponding output.

The RC circuits 46 or 48 may comprise a resistor 41 or 45 and a capacitor 43 or 47 connected in parallel across the a respective input and output of the OPA. Alternatively the RC circuits may comprise a more complex arrangement of components. Such arrangements are known in the art and as such will not be described further.

The OPA provides a virtual ground at inputs 39 and 37. Consequently the current output of the mixer, IBB, is driven into the RC circuits. The current signal on wire 36 is driven into the RC circuit 48 and the current signal on wire 38 is driven into the RC circuit 46.

The transimpedance amplifier serves not only to convert the signal current IBB to a signal voltage V_BB, but also serves to low-pass filter the mixer output current. Baseband signal output V_BB is provided at pins 50 and 52. V_BB may subsequently be processed in further circuitry, this processing may be performed in the analogue and/or the digital domains.

Figure 2 further shows dashed resistor 54 marked R_Ouτ and dashed capacitors 56 and 58. These are representations of resistive and capacitative effects which are used in the analysis of the circuit below. They are not necessarily features of the circuit.

Since mixer shown in Figure 2 operates with single-phase differential LO signal, this mixer type is referred to as a single-phase mixer. In operation an input RF signal, represented as voltage V_RF is applied to the transconductance amplifier 14 and converted into current signals I_RF on wires 20 and 22. The current signals on these two wires are substantially equal and opposite.

These current signals are provided to the mixer 40. The transistors of the mixer are driven by mixer signals Vι_o+ and V_L0.. An exemplary waveform of these signals is shown in Figure 3. V_Lo+ and VLO- drive the gates of the transistors such that they alternate between being in a conductive state (i.e. when the mixing signal is high) and being in a substantially non-conductive state (i.e. when the mixing signal is low). The connections of the respective mixing signals to the transistors are such that either the first and third transistors 28 and 32 are on, i.e. in a conductive state, and the second and fourth transistors 30 and 34 are off, i.e. in a substantially non-conductive state, or vice versa.

The switching transistors serve to commutate input RF current I_RF, received at nodes 29 and 33, between output nodes 31 and 35 in dependence on the mixing signals. I_BB is the current signal produced at these output nodes. The mixer therefore mixes the input RF frequency with the mixing frequency. In the above embodiment, the mixing frequency is substantially the same as the RF frequency. Consequently, the output current IBB will be at baseband (the difference between the RF and mixing frequencies being substantially zero).

The output current may also contain a high frequency component at a frequency which is the sum of the RF and mixing frequencies. This high frequency component is removed by the transimpedance amplifier as described later.

The output current is passed along wires 36 and 38 to the OPA of the transimpedance amplifier 44. It will be noted that the currents on these two wires will be substantially equal and opposite.

The current on wire 38 flows into the resistor 41 of RC circuit 46. This creates a potential difference between the positive input and negative output of the OPA. This difference serves as a feedback signal for the OPA which converts the input current signal into an output potential on node 50.

The current on wire 36 similarly flows into the resistor 45 of RC circuit 48. This creates a potential difference between the negative input and positive output of the OPA. This difference serves as a feedback signal for the OPA which converts the input current signal into an output potential on node 52.

In combination, the potentials on nodes 50 and 52 serve to define an output voltage VBB-

The capacitors 43 and 47 serve to low-pass filter the current I_BB (i.e. to filter out the high frequency component of the current I_BB)- Consequently the output voltage V_BB of the transimpedance amplifier contains only the component which is at baseband.

While the above arrangement has been described with a transimpedance amplifier, any other device which has a low input impedance and is capable of buffering a current can be used in its place. For example, a common gate amplifier may be used. Moreover, the above arrangement can be used for both low-IF as well as zero-IF receivers.

The above described receiver is a current-mode passive mixer. This, for the reasons described below, provides advantages over other circuits.

In particular, due to the virtual ground present at the input of the OPA of the transimpedance amplifier, the drains and sources of the switching transistors of the mixer are kept at constant potential and thus the voltage swing across these switching transistors is reduced. This means that the switching transistors generate low amount of distortion and a high degree of mixer linearity can be achieved, even at a low supply voltage. The ability to operate at low supply voltage is useful in sub micron CMOS processes, where the supply voltages are only in the order of 1 V. In contrast, a voltage-mode passive mixer presents voltage swing at the output of the switching quad. As a result, the switching quads of a voltage-mode passive mixer can contribute significant amount of distortion at the mixer output.

Moreover, since in a current-mode mixer, any voltage swing is present at the OPA output instead of at the output of the switching quad, there is low distortion in the output of the circuit. This is due to the fact that high-gain operational amplifiers with negative feedback loops can swing rail-to-rail with very low distortion.

It is possible to construct both active and passive current-mode mixers. However, switching transistors in an active current-mode CMOS mixer can contribute a significant amount of 1/f noise to the mixer output. This is due to the switching transistors, in an active mixer, being biased at DC current and thus commutating both signal and DC current. Moreover, active CMOS mixers may contribute DC voltage offsets to the output signal.

In contrast, the passive current-mode mixer, as shown in this arrangement, does not commutate DC current, consequently it is generally free of 1/f noise and DC voltage offsets. Passive mixers are also capable of achieving very high linearity, even at low supply voltage. Thus, since passive current-mode CMOS mixers can achieve both low 1/f noise and high linearity simultaneously, they are suited for CMOS zero-IF receivers, and in particular for cellular applications.

An active filter can contribute the largest noise of all the elements in a zero-IF receiver. For this reason, it is useful to reduce the noise contribution of the active filter following the mixer.

The arrangement of Figure 2 is able to reduce the noise contribution of the filter by utilizing a current-mode interface between the mixer and filter, or in other words by employing a filter with a current-mode input stage (i.e. a transimpedance amplifier). If in the current-mode interface between the input to the transimpedance amplifier (i.e. the input to the OPA) and the mixer output, the input-referred noise voltage at the OPA is Vni_N, the output noise voltage (Vn_Ouτ) at the filter passband is given by:

2R

Vn_ouτ = (1 + _p )^VΠIN

where R_Ouτ is the output resistance of the mixer and is represented in Figure 2 as dashed resistor 54.

The noise contribution due to the OPA can therefore be reduced by making the output resistance of the mixer high. Thus, in the current-mode interface between mixer output and analogue baseband input shown in Figure 2, it is useful to provide a large mixer output impedance compared to the resistance R in the feedback loop of the OPA.

in fact, if ROUT » R then Vnouτ ~ VΠJN. Thus, the input-referred noise voltage of OPA experiences very little voltage gain from the OPA input to the OPA output. In comparison, the noise voltage gain in a voltage-mode mixer can be significant.

The current-mode interface between the passive mixer output and transimpedance amplifier's input is also beneficial when considering any DC offset voltage of the OPA in the transimpedance amplifier. Again, provided that the mixer output resistance is sufficiently large compared to that of the resistors 41 and 45, the input-referred offset DC voltage of the OPA in the transimpedance amplifier experiences no voltage gain to the OPA output. Again, in comparison, the corresponding gain in a voltage-mode mixer can be significant.

The output resistance R_Ouτ of the passive mixer may be substantially affected by two mechanisms. Firstly, parasitic capacitance (C_p) located at the output transconductance amplifier. This is shown in Figure 2 as dashed capacitors 56 and 58. Due to the switched capacitor effect, this capacitance may be represented as an equivalent resistor at the point of resistance of ROUT- R_^ eq = ^■

4ηf_LOc_p

Here, f_Lo is the LO frequency; and η is the duty cycle of the LO signal (usually represented as a percentage), η is defined as:

η =

T L₁ O

where τ is the time period for which a switch is in a conducting state; and TLO is the period of the local oscillator signal.

Secondly, ROUT is affected by the fact that, at RF frequencies, the LO signal driving the mixer is not an ideal square wave. Figure 3 shows simulated LO voltage waveforms for a mixer such as that shown in Figure 2. Here, the LO voltage waveform is a rail-to-rail signal with supply voltage of 1.2 V and duty cycle of 50%. It is seen that in the transition point (when the switches turn on and off), there is a short time period, during which all the switches are in conducting state. Thus, during this time period, there is a low impedance path between the inputs of the OPA.

This low impedance path will result in ROU_T being lowered. This mechanism, referred to as switch cross-over, may be reduced by increasing the slope of the LO signal and by selecting the common-mode level of the LO signal accordingly.

The voltage gain of the circuit is another design objective. In the passive current- mode mixer, the switching transistors operate in the triode region and they are not biased at DC current. If instantaneous switching of the switching quad is assumed (i.e. that the current-mode RF signal is multiplied by the square wave toggling at the LO frequency), the voltage conversion gain of the circuit (Av) may be calculated

'BB ^: -K π

therefore V₈B

Av - ² α R v_RF π

where

2/π represents the conversion loss through the switching quad; g_m is the transconductance of the transconductance amplifier; R is the resistance of the resistor 41 or 45 in the RC circuit.

To provide the two signals V_Ouτ,ι and VOUT.Q as described in figure 1 , two single phase mixers as described above may be used in the I and Q paths respectively .

An embodiment of the present invention will now be described with reference to Figure 4. This embodiment shows a current-mode passive mixer arrangement capable of vector demodulation.

This arrangement shares many common components with the single phase mixer described above. Accordingly, similar components have the same reference numerals. As will become apparent from the description, this embodiment of the invention provides two mixing paths, referred to as an in-phase (I) and quadrature (Q) paths. These paths contain components similar in both design and operation to those found in Figure 2. These components have been given reference numerals XXO for the I path and XX5 for the Q path, where XX is the reference numeral used in Figure 2.

This mixer arrangement is comprises a transconductance amplifier 14, which in similar operation to the arrangement in Figure 2, converts V_RF to l_RF on wires 20 and 22. Each IRF wire is split. Wire 20 divided into wires 200 and 205, wire 22 divided into wires 220 and 225. Wires 200 and 220 are provided to a first mixer 400, and wires 205 and 225 are provided to a second mixer 405. Capacitors 240, 245, 260 and 265 may be provided on wires 200, 205, 220 and 225 respectively. These oppose the flow of DC current through the mixers. Alternatively or additionally the output of the transconductance amplifier 14 may be biased at the same potential as the input potential of the OPAs 440 and 445.

Thus wires 200 and 220 provide signal current l_RF,ι to I mixer 400 (and in particular to nodes 290 and 330 of this mixer) and wires 205 and 225 provide signal current I_RF.Q to Q mixer 405 (and in particular to nodes 295 and 335 of this mixer).

The ratio at which I_RF is split between the wires to produce l_RF,ι and I_RF,Q will be dependent on, at least, the switches in the respective mixers. As such, the ratio of the currents may not be constant.

The two mixers have similar operation to the arrangement described in Figure 2. However the first mixer is driven by a first LO signal LOI and the second mixer is driven by a second LO signal LOQ.

The arrangement in Figure 2 comprised only of NMOS transistors. Embodiments of the present invention may comprise NMOS transistors, alternatively or additionally PMOS, an other equivalent element, or an alternative transistor type may be used for switching the current. The mixer may comprise a combination of different types of switching element, (e.g. comprise two NMOS transistors and two PMOS transistors). It will be clear to the skilled person that the mixing signals LOI and LOQ are described for driving a NMOS transistor, and that they may be appropriately changed to drive any alternative switches.

The first mixer downconverts the I_RF signal to a first baseband signal current which is provided to transimpedance amplifier 420. This transimpedance amplifier is similar in arrangement to the transimpedance amplifier described in relation to Figure 2 and converts the signal current into a first signal voltage VBBJ-

The second mixer downconverts the IRF signal to a second baseband signal current which is provided to transimpedance amplifier 425. This transimpedance amplifier is again similar in arrangement to the transimpedance amplifier described in relation to Figure 2 and converts the signal current into a second signal voltage V_BB,Q-

This mixer arrangement is known as a quadrature mixer since two mixers (a first operating in-phase and a second operating in quadrature) receive the same signal from a transconductance amplifier.

The operation of the quadrature mixer is similar to that of the single phase mixer described with reference to Figure 2, with the switching quads operating to commutate the input signal current in dependence on the mixing signal.

A plot of signals LOi and LOQ is shown in Figure 6. Each signal comprises a positive and a negative element which have a half period or 180 degrees phase shift between them. These signals are represented by LOI- and LOI+, LOQ- and LOQ+.

The two signals, LOI and LOQ are in quadrature, i.e. they are a quarter period or 90 degrees out of phase with each other. Moreover the signals have a 25% duty cycle (duty cycle being defined above). LOi and LOQ, being in quadrature and having a 25% duty cycle means that, assuming the signals are ideal square waves, when any one of signals LOI+, LOI-, LOQ+ and LOQ- are positive, all of the other signals are at zero. This means that the switches driven by LOi wilf be open when the switches of LOQ are closed and vice versa. To put this another way, only one of the two mixers will be conducting current at any given time in the operation of the quadrature mixer.

It will be clear to the skilled person that the mixing signal will not be an ideal square wave. It will also be clear that exactly 25% duty cycle is not essential to the invention. Consequently there may be an overlap between the signals. However embodiments of the invention will demonstrate a period of time in which one mixer is in a substantially stable conducting state and the other mixer will be in a substantially stable non-conducting state.

It will be well understood that both the phase shift and the 25% duty cycle signals can be produced by many methods known in the art. These known techniques allow very accurate generation of 25% duty cycle with little sensitivity to process, temperature or supply variations.

The signals may be derived from one or many local oscillator signals. The signals may be derived in stages (i.e. LOI and LOQ are produced by first circuitry, and further circuitry then converts these signals into LOI+, LOI-, LOQ+ and LOQ-), or produces as four separate signals from the same circuitry.

The separate signals may be produced by phase shifting this LO signal by differing amounts to produce the separate signals. For example, to produce LOI and LOQ, a LO signal may be maintained in-phase to produce LOI and phase shifted by 90 degrees to produce LOQ. Alternatively or additionally, the same LO signal may be phase shifted by +45 degrees and -45 degrees to produce LOI and LOQ respectively. Other combinations of angles will be obvious to the skilled person.

In one embodiment, the four signals may be produced by phase shifting a LO signal with 25% duty cycle by O, 90, 180 and 270 degrees respectively.

In some embodiments of the invention the LOI and LOQ signals may be considered as single signals comprising three levels as opposed to the two shown in Figure 6. These levels may be analogous to +1 , O and -1 , where (in respect of the LOI signal) +1 indicates LOI+ being high and LOl- being low; -1 indicates LOI- being high and LO+ being low; and O indicates both LOI+ and LOI- being low. These signals may be converted before driving the switches in the mixers, alternatively or additionally the signals may directly drive the switches.

The LO signal may have a 50% duty cycle, and be later converted into a signal with a 25% duty cycle. Alternatively or additionally, the LO signal may have a frequency double that of the desired LOI and LOQ signals, the separate signals being produced from this higher frequency signal.

In one embodiment, the 25% duty cycle signals may be produced by combining a first signal at the desired frequency with a second signal at twice the desired frequency.

While variations in the offset angles may provide disadvantageous, the use of the exact angle described above is not essential to the invention. Variations in angle away from the desired angle may be by design or be as a result of manufacturing tolerances.

Additionally, in current-mode quadrature passive mixers with nominal duty cycle of 25%, a ±5% variation in the duty cycle has little effect on the mixer performance. It will therefore be apparent that the use in the description of 25% duty cycle is purely exemplary and that variations away from this nominal figure are within the scope of the present invention. Again this variation may be due to design and/or manufacturing tolerances.

Generally, the generation of 25% LO duty cycle (with I and Q phases) requires only small amount of extra power consumption or silicon area compared to generation of 50% I and Q LO signals.

Advantageously, with 25% duty cycle in the LO waveform, high mixer output impedance, high linearity, low DC offset, and low 1/f noise are all available simultaneously. The mixer output impedance with 25% duty cycle may be even higher than a single-phase mixer with 50% duty cycle. Linearity may be measured in terms of 1 dB input compression point (ICP), second-order intercept point (I1P2), and third-order intercept point (IIP3). The above described embodiment provides high linearity in all these measurements.

The advantages of this embodiments of the invention will be discussed in more detail below.

Previous mixers typically comprised two transconductance amplifiers which would separately provide the inputs to the two switching quads. Moreover, both single-phase and quadrature mixers are typically operated with LO voltage waveforms having duty cycle of 50%.

The quadrature mixer show in the embodiment in Figure 4 has only one transconductance amplifier, this saves one integrated inductor compared to previous techniques. This provides a first advantage as it results in lower silicon area and cost.

However, simply providing a single transconductance amplifier with a traditional 50% duty cycle has disadvantages.

Figure 5 shows the simulated LO voltage waveforms for a mixer shown in Figure 4 assuming the LO signal has 50% duty cycle, tt is apparent that at any given time, two of the LO signals are high. For example, when LOi+ is high, either LOQ- or LOQ+ is also high.

With reference to the circuit shown in Figure 4, if, for example, LOI+ and LOQ- are both high, switches 280, 320, 305, and 345 are all in a conducting state and there is a low-impedance path between the inputs of the two transimpedaπce ampiifiers. Specifically, input 370 of OPA 440 will be connected to input 395 of OPA 395, and input 390 of OPA 440 will be connected to input 375 of OPA 395. In other words, OPA 440 in the f branch sees the virtual ground of the OPA 445 in the Q branch and vice versa. A low impedance path such as the above one will exist at all time except for the short periods in which the switches are switching. This short circuit lowers the output resistances Rouτ,ι and ROUT.Q of the current- mode quadrature passive mixer with 50% LO duty cycle compared to the output resistance of a single-phase mixer, such as the one shown in Figure 2. This is a disadvantage as it increases the noise contribution of the circuit (in accordance with the equations above).

Figure 6 illustrates the LO voltage waveforms in the quadrature mixer when operated with 25% duty cycle of the currently presented embodiment. Again, I and Q phases differ by 90 degrees. It seen that during all times, only a single LO signal is high at time. As a result, only a single pair of switches is conducting at time and there is no overlap between I or Q switches. In other words, if a single pair of ] switches is conducting at time, all Q switches are in non-conducting stage and vice versa. For instance, when LOI+ is high, only transistors 280 and 320 are conducting. Thus, inputs of the I and Q OPAs are disconnected at all times.

As a result, high mixer output resistance can be achieved. In particular R_Ouτ is increased compared to a similar quadrature mixer with a 50% duty cycle. This beneficially reduces the noise contribution of the transimpedance amplifier.

Moreover, the OPAs see the parallel parasitic capacitance (C_p), which results in R_eq described above, only half of time compared to the mixer with 50% duty cycle. From the equation for R_eq above , it is clear that this has the effect of doubling R_eq.

If it is assumed that the switched capacitor effect (which leads to R_eq) dominates ROU_T, twice as high mixer output resistance ROUT may be achieved in a quadrature current-mode passive mixer with 25% duty cycle compared to a single-phase mixer with 50% duty cycle. This, as described above, may reduce noise.

The quadrature mixer of this embodiment provides a further advantage over a quadrature mixer using a 50% LO duty cycle. In the quadrature mixer with 50% LO duty cycle, the mixer RF input transconductance amplifier always sees a pair of switches from both the I and Q mixers conducting at a time. As a result, the current through the I and Q branches of the circuit (IRF.I and I_RF,Q) is halved. I.e. since: I_{RF I} + I_{RF Q} = V_RFg_m and assuming: I_{RF I} = I_{RF Q}

IRF.I and IR_F.Q are both half of the current in the single phase mixer.

Therefore, for a given g_m and mixer load resistance R, the voltage conversion gain of the quadrature mixer (with 50% LO duty cycle) is half that of the a single- phase mixer, i.e.:

A_v = ^ = -g_mR

V_RF π

This equates to a 6dB reduction in voltage gain.

In contrast, in the quadrature mixer with 25% LO duty cycle, the mixer RF input transconductance amplifier sees only a single pair of switches from the I or Q mixers conducting at a time.

As a result, the current through any given mixer and thus to the corresponding transimpedance amplifier is the same as that for the single phase mixer as shown in Figure 2. As a result, it can be shown that for a given g_m and mixer load resistance R, the voltage conversion gain of the quadrature mixer (with 25% LO duty cycle) is V2 (or 3dB) higher than the gain of quadrature mixer with 50% duty cycle:

It will be recognized by the skilled person that a voltage-mode mixer will not suffer from the loss of power described above since the voltage will be the same across the two mixers, even when they are both in a conductive state. However, voltage-mode mixers have associated problems and disadvantages as described above.

In the current-mode quadrature passive mixer with 25% LO duty cycle, high linearity is also available. This is due to the fact that during all times, a single pair of switches is conducting and the outputs of the transconductance amplifier are constantly connected to the virtual ground (low impedance node) of one of the I or Q OPA. Thus, switches in the I and Q mixers see a low voltage swing across them and thus low distortion is created.

It will be apparent to the skilled person that if separate single-phase I and Q mixers were used (i.e. each mixer was driven by a separate transconductance amplifier) with a 25% duty cycle, the output of respective transconductance amplifiers would not see low impedance node for a full LO period. During this time period, when low impedance is not seen, distortion may be generated.

Thus, 25% LO duty cycle is beneficial only when embedded in quadrature mixer. Furthermore, as mentioned above, use of a 25% duty cycle reduces noise by increasing R_Ouτ- It will therefore be apparent that this embodiment of the invention offers a benefit which cannot be realized in a conventional single phase mixer without an unacceptable increase in distortion.

In some cases, it may be advantageous or necessary to use an inductor- capacitor (LC) tuned circuit in the transconductance amplifier. If this is necessary, the use of quadrature current-mode passive mixer in the zero- or low- fF receiver allows saving integrated inductors compared to the use of separate single-phase mixers for I and Q branches. As is known in the art, these may contribute significantly to both silicon area and cost. Moreover, by employing 25% instead of 50% LO duty cycle, the quadrature current-mode passive mixer can be implemented with high output resistance or without otherwise deteriorating the mixer performance (noise or linearity).

An alternative embodiment of the invention will now be described with reference to Figure 7. This embodiment in similar in structure to the embodiment described with reference to Figure 4. In addition to the circuitry provided in the previous embodiment there are provided switches 210, 215, 230 and 235 on current wires 200, 205, 220 and 225 respectively. Alternatively or additionally these switches may be provided on current wires 360, 365, 380 and 385.

The switches are driven by a signal HLO with a frequency twice that of the LO signal driving the mixers. This higher frequency LO signal preferably has alternate rising edges coinciding with the rising edges of the lower frequency signal. The switches are preferably configured such that when the switches which are connected to the first mixer are open, the switches connected to the second mixer are closed and vice versa.

The first and second mixers may then be driven by a LO signal with a 50% duty cycle.

In operation, this alternative circuit will have the same operational effect and thus the same benefits of the 25% duty cycle system. In particular, the additional switches prevent current being conducted by the I path when the Q path is conductive and vice versa. The embodiment provides an advantage in that it is less complex to produce a 50% duty cycle signal, moreover, this embodiment may enable the same 50% duty cycle signal to be provided to the I and Q mixers, thereby reducing the need for phase shifting circuitry for generating the separate LOI and LOQ signals. However this circuit also requires additional switches.

Normally, since the output resistance and the voltage conversion gain of the current-mode quadrature passive mixer operating with 50% LO duty cycle are significantly lower compared to the single-phase mixer, the use of quadrature mixer with 50% LO duty cycle in zero- or low-IF receivers is avoided and separate I and Q single-phase mixers are employed instead. Embodiments of the present invention enable the benefits of quadrature mixers to be realized while mitigating these problems.

However, it will be apparent to the skilled person that, even if separate I and Q current-mode CMOS passive mixers are used, two mechanisms, i.e. switched capacitor mechanism and switch cross-over, remain. These lower the mixer output resistance, a problem mitigated by embodiments of the present invention. Furthermore, as mentioned above, a single phase mixer cannot be run with a 25% duty cycle without creating additional distortion.

Current-mode passive mixers offer benefits when used in CMOS zero-IF radio receivers such as wireless communication devices (such as mobile stations, PDAs laptop computers, computers and mobile telephones). Firstly, passive mixers are substantially free of 1/f noise and DC offsets.

Secondly, the current-mode interface at the output of the current-mode mixer improves the receiver performance in many ways. Since current-mode mixers commutate current instead voltage, there is no voltage swing at the mixer output and thus the distortion due to the switches is reduced. As a result, very high mixer linearity is available, even at low supply voltage. The current-mode interface also reduces the noise contribution and DC offsets of the analogue BB, provided that sufficiently large ROU_T in the mixer can be provided. For this reason, the R_Ouτ of the current-mode mixer should be made high.

According to embodiments of the invention a technique for increasing the ROUT of a current-mode quadrature passive mixer is proposed. In the quadrature mixer, separate I and Q mixers are combined to form a single mixer, in which a common transconductance amplifier drives both I and Q switches in parallel. Since only a single transconductance amplifier is needed, lower silicon area and cost are available.

It is shown that, if the switches in the current-mode quadrature passive mixer are driven by an LO signal having duty of 25% as herein proposed, high ROUT and a 3dB larger gain may be achieved in comparison to a quadrature mixer with a duty cycle of 50%.

Moreover, up to twice as high R_Ouτ may be achieved in the current-mode quadrature passive mixer with duty cycle of 25% compared to the conventional single-phase mixer with duty cycle of 50%. The use of a single quadrature mixer instead of two separate single-phase mixers result in a lowering of cost.

Figure 8 shows an example of a mobile telecommunications network in which embodiments of the invention may be implemented.

A base station 60 may transmit an RF signal. User equipment 62, may then receive this RF signal at aerial 64. This received signal may be passed to a mixing apparatus 66 which comprises an embodiment of the present invention. The mixing apparatus downconverts the received RF signal to a baseband signal in accordance with embodiments of the invention. The downconverted signal may then be passed to processing circuitry 68.

This processing circuitry may be contained within the same integrated circuit as the mixing apparatus. Alternatively some or all of the processing circuitry may be separate from the mixing apparatus.

This processing circuitry may further process the received signal. This further processing may include signal processing. The processing circuitry may convert the received signal into an audio signal which may be passed to an amplifier. Alternatively or additionally the processor may convert the received signal into a data format which may be displayed on a screen of the user equipment.

The user equipment may further transmit a RF signal, this may be received at the base station. The base station may be connected to one or more networks 70. These networks may include GSM, GPRS, UTMS, EDGE or 3G networks. Alternatively or additionally the networks may include a connection to the internet.

It is advantageous in such systems to provide small and efficient user equipment such as that shown in Figure 8. For example, it is particularly advantageous to be able to realize single-chip multi-radio modems on pure digital CMOS for wireless applications. The technique proposed in this report for realizing the quadrature passive current-mode mixer, aids this realization. The present invention has been described with reference to exemplary embodiments, however it will be clear to the skilled person that the present invention is not restricted to these embodiments and that substantial modification is possible within the scope of the invention.

Claims

1. A mixing arrangement comprising: an input configured to receive an input signal at a first frequency; first and second outputs configured to provide output signals; a first mixer connected between said input and said first output, said first mixer configured to mix said input signal with a first mixing signal; a second mixer connected between said input and said second output, said second mixer configured to mix said input signal with a second mixing signai; wherein said first mixer and said second mixer are configured such that in use, for at least part of the time only one of said first and second mixers is arranged to mix said input signai with a respective mixing signal.

2. The mixing arrangement of any claim 1 wherein said first and second mixers are substantially the same.

3. The mixing arrangement of claims 1 or 2 wherein said input signal comprises a current signal.

4. The mixing arrangement of any preceding claim wherein said mixing arrangement is configured such that said first and second mixing signals have the same frequency.

5. The mixing arrangement of any preceding claim wherein said mixing arrangement is configured such that there is a difference in phase between said first and second mixing signals,

6. The mixing arrangement of claim 5 wherein said difference in phase is substantially equal to a quarter of a period of said mixing signals.

7. The mixing arrangement of any preceding claim wherein said first and second mixers comprise a plurality of switching elements.

8. The mixing arrangement of claim 7 wherein said mixers are configured such that said switching elements are operable for less than 35% of a period of said mixing signal.

9. The mixing arrangement of claims 7 or 8 wherein said mixers are configured such that said switching elements are operable for substantially 25% of a period of said mixing signal.

10. The mixing arrangement of any preceding claim wherein said mixers comprise transistors.

11. The mixing arrangement of claim 10 wherein said first and second mixers each comprise four transistors.

12. The mixing arrangement of claim 11 wherein said four transistors of each of said first and second mixers are arranged as pairs of transistors such that each pair of transistors operate concurrently.

13. The mixing arrangement of claim 12 wherein said mixing arrangement is configured such that in use, for at least part of the time, a first pair of transistors of one of either said first or second mixer is on, and a second pair of transistors of said one of said first or second mixer is off.

14. The mixing arrangement of claims 10 to 13 wherein said mixing arrangement is configured such that in use, for at least part of the time, the four transistors of one of said first and second mixer are off.

15. The mixing arrangement of claims 11 to 14 wherein said transistors are driven by said mixing signals.

16. The mixing arrangement of any preceding claim further comprising a transconductance amplifier, the output of which is connected to said input.

17. The mixing arrangement of any preceding claim further comprising a first and a second transimpedance amplifier respectively connected to said first and second outputs.

18. The mixing arrangement of any preceding claim wherein said mixers comprise complementary metal oxide semiconductor elements.

19. An integrated circuit comprising a mixing arrangement as claimed in any preceding claim.

20. A radio receiver comprising said mixing arrangement as claimed in any preceding claim.

21. The radio receiver of claim 20 wherein said radio receiver is a mobile communications device or a base station in a communications network.

22. A method of mixing comprising: receiving at an input an input signal at a first frequency; mixing at a first mixer connected to said input, said input signal with a first mixing signal; mixing at a second mixer connected to said input, said input signal with a first mixing signal; wherein for at least part of the time only one of said first and second mixers is arranged to mix said input signal with a respective mixing signal.