METHOD TO IMPROVE SELECTIVITY OF TRANSFERRED- IMPEDANCE FILTER AND A TRANSFERRED- IMPEDANCE FILTER

Technical field

The present invention relates to a transferred- impedance filter with a feedback loop, a radio receiver utilizing such a filter, a mobile communication apparatus comprising such a radio receiver, and a method for increasing selectivity for such a filter. Background of the invention

RF receivers must tolerate high blocking signals while maintaining their own performance. This requires filtering for RF-signals prior to a LNA (low noise amplifier) and in many systems also after the LNA. This is especially true in code division multiple access systems (e.g., CDMA2000 and WCDMA) where a transmitter usually sends its high-level signal while a receiver receives a very low-level signal. A transferred-impedance filter (TIF) can be used to improve tolerance of blocking signals in this context. This is described in WO2006/097835 A2.

A TIF transfers a baseband impedance to radio frequency (RF) range around a center frequency indicated by a reference frequency provided to the filter. The reference frequency can be a local oscillator (LO) frequency provided to the TIF. The TIF and the impedance in parallel form a bandpass or bandstop RF filter depending on the transferred impedance. The center frequency of the filter is determined by the reference frequency of the filter, and can thus be changed by changing the reference frequency. It is desireable to improve selectivity of the TIF, i.e. difference in gain between frequencies to be blocked and frequencies to be passed.

Summary of the invention

In view of the above, an objective of the invention is to solve or at least reduce the problems discussed above. In particular, an objective is to improve selectivity of the TIF.

The present invention is based on the understanding that two branches (inphase (I) and quadrature (Q) branches) having phase shifts of π/2 in the corresponding LO signals can be used to process a received I/Q- modulated signal. A TIF generally consists of three parts. (1) The transferred-impedance stage consists of impedances that are transferred. (2) A switching stage, which transfers the impedances to RF frequencies, consists of two stages. In a balanced topology, both stages may contain four switches. (3) The LO generation circuit generates the signals, preferably pulse trains, that control the switches of the switching stage. In order to prevent inphase and quadrature branches from interfering with each other, the time when a switch of the TIF is on, i.e. conducts, during a period of the signal controlling the switch can be selected to 25% of the period. Thus, in the general case, four signals having phase shifts of π/2 are needed and the time when a switch is on during one period can be selected to 25% of the period. The inventor found out that the selectivity of a TIF can be improved by decreasing the time when a switch is on during one period of the signal controlling the switch. Conventionally, the time when a switch is on during one period has been set to approximately 25% in TIF implementations. However, the inventor found that when the time when a switch is on during one period is reduced below that value, both the passband and stopband impedances increase, but the relation between these impedances increases. For example, when the time when a switch is on during one period of the signal controlling the switch decreases from 25% of the period to 20% of the period, the selectivity of a TIF, which uses capacitors

as the transferred impedances, is improved by approximately 3dB.

According to a first aspect of the invention, there is provided a transferred-impedance filter comprising a transferred-impedance stage comprising impedances; a switching stage configured such that said impedances are transferred to radio frequency domain; an oscillator configured to produce four signal pulse trains, each being phase shifted to each other with a distribution of π/2, wherein said signal pulse trains are connected to switches of said switching stage to control switches of said switching stage; and a feedback circuit configured to measure a difference between a produced direct current component related to one or more of said signal pulse trains and a reference value, and to control at least one of said signal pulse trains to reduce the difference such that the time when a switch of said switching stage conducts during one period of the signal controlling the switch to a value indicated by said reference value.

The feedback circuit may be configured such that the time when a switch of said switching stage conducts during one period of the signal controlling the switch is less than 25%, preferably less than 24%, preferably less than 23%, preferably less than 22%, preferably less than 21%, preferably less than 20%.

The feedback circuit may comprise a measurement resistor connected between a node of said oscillator and a node of said feedback circuit for measurement of said direct current component, wherein said configuration of said feedback circuit depends on said measurement resistor .

The produced direct current component may be produced by a lowpass filter. The control of said signal pulse trains may comprise a level shifter structure, wherein said level shifter

structure is arranged to generate a signal related to said produced direct current component.

The feedback circuit may comprise a level shift resistor connected in between a node in said feedback circuit and a voltage supply, wherein said configuration of said feedback circuit depends on said level shift resistor .

According to a second aspect of the present invention, there is provided a radio receiver comprising a filter arranged in a signal path of said receiver, wherein said filter comprises a transferred-impedance filter according to the first aspect of the present invention.

According to a third aspect of the present invention, there is provided a radio transmitter comprising a filter arranged in a signal path of said transmitter, wherein said filter comprises a transferred- impedance filter according to the first aspect of the present invention. Here, it should be noted that the second and third aspects of the invention may be combined to form a transceiver comprising the filter of the first aspect of the present invention.

According to a fourth aspect of the present invention, there is provided a mobile communication apparatus comprising a radio receiver according to the second aspect of the present invention.

According to a fifth aspect of the present invention, there is provided a mobile communication apparatus comprising a radio transmitter according to the third aspect of the present invention.

Here, it should be noted that the fourth and fifth aspects of the present invention may be combined to form a mobile communication apparatus comprising both a receiver according to the second aspect of the present invention and a transmitter according to the third aspect of the invention, or a mobile communication apparatus

comprising a transceiver, as demonstrated above, comprising a filter according to the first aspect of the present invention.

According to a sixth aspect of the present invention, there is provided a method for increasing selectivity of a transferred-impedance filter, comprising providing four signal pulse trains, each being phase shifted to each other with a distribution of π/2, wherein said signal pulse trains are connected to switches of a switching stage of said transferred-impedance filter to control switches of said switching stage; measuring a difference between a produced direct current component related to at least one of said signal pulse trains and a reference value; and controlling at least one of said signal pulse trains to reduce said difference such that the time when a switch of said switching stage conducts during one period of said signal controlling the switch to a value indicated by said reference value.

The method may further comprise selecting said reference such that the time when a switch of said switching stage conducts during one period of said signal controlling the switch is less than 25% of said period, preferably less than 24%, preferably less than 23%, preferably less than 22%, preferably less than 21%, preferably less than 20%.

The method may further comprise lowpass-filtering at least one of said pulse trains to produce said direct current component.

Generally, all terms used in the claims are to be interpreted according to their ordinary meaning in the technical field, unless explicitly defined otherwise herein. All references to "a/an/the [element, device, component, means, step, etc]" are to be interpreted openly as referring to at least one instance of said element, device, component, means, step, etc., unless explicitly stated otherwise. The steps of any method

disclosed herein do not have to be performed in the exact order disclosed, unless explicitly stated.

Other objectives, features and advantages of the present invention will appear from the following detailed disclosure, from the attached dependent claims as well as from the drawings . Brief description of the drawings

The above, as well as additional objects, features and advantages of the present invention, will be better understood through the following illustrative and non- limiting detailed description of preferred embodiments of the present invention, with reference. to the appended drawings, where the same reference numerals will be used for similar elements, wherein: Fig. 1 illustrates a transferred-impedance filter according to an embodiment of the present invention;

Fig. 2 illustrates switching stage and impedance stage of a transferred-impedance filter according to an embodiment of the present invention; Fig. 3 illustrates signal output from NAND circuit;

Fig. 4 illustrates signal output from LO buffer stage;

Fig. 5 illustrates gain to frequency characteristics for some examples of MOSFET switch width (W) and the relation between the time when a switch is on {t _0N ) during one period (T) and the period;

Fig. 6 illustrates a feedback circuit according to an embodiment of the present invention;

Fig. 7 illustrates signals In Fig. 6; Fig. 8 illustrates a part of a joint feedback structure according to an embodiment of the present invention;

Fig. 9 illustrates a part of a joint feedback structure with a further loop filter stage according to an embodiment of the present invention;

Fig. 10 illustrates LO buffer stage and feedback circuit according to an embodiment of the present invention;

Fig. 11 illustrates a transferred impedance filter according to an embodiment of the present invention;

Fig. 12 illustrates a transferred impedance filter according to an embodiment of the present invention; and

Fig. 13 illustrates a transferred impedance filter according to an embodiment of the present invention.

Detailed description of preferred embodiments

Improvements of selectivity of a transferred- impedance filter (TIF) , which filter structure is thoroughly described in WO2006/097835, which is hereby incorporated by reference, have a large impact on , performance of radio receivers. A TIF can be used in radio frequency integrated circuits (RFICs) to improve the tolerance of blocking signals. In a radio receiver IC, the downconversion mixers, which follow the low-noise amplifier (LNA) , typically limit out-of-band linearity. When a TIF is implemented in the signal path of the receiver before the downconversion mixers, (for example, at a load of the LNA) , out-of-band signals, like the transmitter leakage in a Wideband Code Division Multiple Access (WCDMA) mobile station receiver, can be attenuated before the mixers, which results in a significant improvement in the out-of-band linearity. A TIF can also be located elsewhere in the receiver signal chain and more than one TIF can be used. One or more TIFs can also be used in a radio transmitter. A TIF can be integrated on the receiver IC, transmitter IC, or transceiver IC.

A TIF transfers a baseband impedance to RF frequencies around a center frequency indicated by the reference frequency provided to the filter. A TIF and the impedance in parallel with the TIF form a bandpass or bandstop RF filter depending on the impedance, which is transferred. The center frequency of the filter is

determined by the reference frequency of the filter and, therefore, can be changed simply by changing the reference frequency. Two branches (I and Q branches) having π/2 phase shifts in the corresponding pulse train signals controlling the switches are desireable to be used to process a received I/Q-modulated signal. A TIF consists of three parts:

(1) The transferred-impedance stage consists of impedances that are transferred. (2) A switching stage, which transfers the impedances to RF frequencies, consists of two stages. In a balanced topology, both stages can contain four switches, e.g. MOSFET switches.

(3) The LO generation circuit generates the pulse train signals that control the switches of the switching stage. Four pulse trains having π/2 phase shifts are normally needed and the time when a switch of the TIF is on, i.e. conducts, during one period of the signal controlling the switch can be selected to 25% of the period in order to prevent inphase and quadrature branches from interfering with each other.

There are some characteristics of a TIF that have great impact on performance, e.g. when used in a radio receiver, namely selectivity of the TIF. The ability to hold the relation between the time when a switch is on during one period and the period constant is also advantageous .

The selectivity of the TIF is the difference between the passband gain and the stopband gain at specified frequency offsets from the center frequency of the TIF. It was found out that the selectivity of a TIF can be improved by decreasing the time when a switch is on during one period of the signal controlling the switch. Conventionally, the time when a switch of the TIF is on during one period of the signal controlling the switch has been selected to approximately 25% of the period in

TIF implementations to prevent inphase and quadrature branches from interfering with each other.

If the switch is on when the pulse train signal controlling the switch has a high value, e.g. equal to a positive supply voltage, the relation between the time when the switch is on during one period and the period is approximately equal to the duty cycle (if the rise and fall times of an ideal pulse train signal are equal to zero, the relation between the time when the switch is on during one period and the period is equal to the duty cycle) . For example, this is the case with a NMOS transistor when the control signal is connected to the gate of the transistor. In the case of a PMOS transistor, for example, having the control signal connected to the gate, the relation between the time when the switch is on during one period and the period is approximately equal to 1 minus the duty cycle.

The concept of duty cycle is used here only to explain the operation of the circuit. The objective of the invention is set the relation between the time when a switch of the TIF is on during one period and the period to a desired value, which is less than 25%. This leads to an improved selectivity of the TIF compared to conventional implementations. The simplified output signal of the LO buffer (V _L0 ) varies between OV and the supply voltage (V _DD ) . Here, the rise time (t _R ) is the time it takes for the signal to increase from OV to V _DD . The fall time (t _F ) is the time it takes for the signal to decrease from V _D D to OV. The time when the signal is high (t _HIGH ) , i.e. V _L0 > 0.5V _D D, divided by the period of the signal (T) is the duty cycle (t _H i _GH /T) . The direct current (DC) component is the mean value of the signal:

(1)

Therefore, the duty cycle of the simplified pulse train signal is equal to the DC component divided by the supply voltage:

¹ tHlGH _ VLOβC _/o ,

T ¹ V* DD

In practice, the signal waveform differs from the simple model and, therefore, the DC component gives an approximation of the duty cycle. However, the DC component gives an accurate and sufficient approximation of the duty cycle of a practical pulse train signal according to transistor-level simulations.

A feedback loop that sets the duty cycle of the output pulse train signal to a desired value is shown in Fig. 6. The pole formed by R2 and C2 measures the DC component of the output signal (V _OOT , _DC ) by attenuating the LO frequency and its harmonics. The feedback amplifier adjusts the DC voltage at node Nl, illustrated in Fig. 6, in such a way that the DC component of the output signal becomes equal to the reference voltage (V _REF ) _I which is derived from the supply voltage with the resistive voltage divider formed by R3 and R4. Now, the duty cycle is

HIGH _ v ^V LO, DC _ v ^v REF _ R4

T V _DD V _DD R3 + R4

It can be seen that the duty cycle does not depend on the supply voltage. The feedback loop does not change the breakover voltage (or breakover point) of the input of the LO buffer, which is determined by the component dimensions, process parameters, supply voltage, and temperature. When the input signal of the LO buffer crosses the breakover point, the output signal changes

its state. In principle, the feedback loop controls the

DC component of the input signal at node Nl without affecting the AC component of the input signal, as shown in Fig. 7. As discussed above, four similar pulse trains with π/2 phase shifts are preferred in a TIF, and a single feedback loop can be used instead of four separate loops to reduce silicon area and power dissipation. The four resistors (R _MEAS ) connected to the four outputs can be connected to a common node for the measurement of the average value of the DC components of the four output signals (V _ODT , _DC ) r ^as shown in Fig. 8. In this case, the value of V _0UT , _DC = t _H iGH/T-V _DD where t _H iGH/T is the duty cycle at one output. For example, when the duty cycle is 20%, the input voltage of the opamp is 0.20V _DD . However, the optimal input voltage for a feedback amplifier can be higher than 0.20V _DD . Therefore, level shifting of V _OOT , _DC to a higher voltage that is suitable for the feedback amplifier may be needed. The level shifting can be implemented with a resistor (R _LS ) connected between the positive supply and the common node used for the measurement of the average value of the DC components of the four output signals, as shown in Fig. 8. Level shifting to a lower voltage can be implemented by connecting a resistor between the negative supply and the common node used for the measurement of the average value of the DC components of the four output signals. Other techniques to implement the level shifting exist, for example, a current source can be used. When the level shifting resistor is connected to the positive supply, the value of R _LS as a function of duty cycle, V _DD , and V _RE F can be written as

The factor four in the denominator is due to the four output signals. Since the relation between R _L s and R _MEAS and the relation between V _DD and V _REF (V _REF can be generated from V _DD with a resistive voltage divider, as illustrated in Figs 6 and 10) are accurate (however limited by device matching), it can be seen in eq. (4) that the duty cycle (t _H i _GH/ T) also remains accurate in spite of variations in V _DD -

If the duty cycle of the output signal" is less than 25% and Cl is excluded in Fig. 8, V _OUT , _DC is a rectangular pulse train, which has a frequency and duty cycle that are four times those of one output signal. Therefore, if the LO frequency is f _L θ _f V _OUT , _DC contains in principle a DC component, a fundamental component at 4f _L0 , and the harmonics of 4f _L0 . Signals at frequencies f _L0 and its harmonics may also exist due to device mismatches . The filter of the feedback loop preferably attenuate the high-frequency components of V _OOT , _DC - A single pole is not a very selective filter and if a high attenuation is required at high frequencies, the pole frequency is preferably orders of magnitude lower than the reference frequency of the TIF. Naturally, a higher order passive or active filter can be used instead of a single pole to improve the high frequency attenuation. For example, a higher attenuation at the frequency of 4f _L0 with an equal total capacitance can be achieved with the topology shown in Fig. 9. Two capacitors Cl, C2 are used instead of a single one Cl as in Fig. 8, and resistor R2 is added. Since R2 is used only for filtering and its value does not affect the direct current component of V ₀ UT,D _O R2 can

be a minimum-width resistor occupying only a small silicon area. With the filter topology shown in Fig. 9, the -3dB frequency of the filter can be maintained high even when a high attenuation at 4f _L0 is achieved. This improves the stability of the feedback loop.

A possible implementation of the invention is shown in dashed lines in Fig. 10. The circuit implements the block referenced as 105 and 110 in Fig. 1. The circuit can be implemented with CMOS and BiCMOS technologies and can be operated from low supply voltages, like 1.2V. The nodes LO_0, LO__90, LO_180, and LO_270 are connected to the gates of the NMOS switches of the TIF. AC couplings can be used before the gates of the switches if necessary. Since a NAND circuit is used before the circuit of Fig. 10 and NMOS switches are used in the switching stage, an odd number of cascaded inverters 1002 is needed to get output signals with a duty cycle of less than 25%. This means that the time when a NMOS switch of the TIF is on during one period is less than 25% of the period. This leads to an improved selectivify of the TIF compared to the conventional case where the time when a switch is on during one period is approximately 25% of the period. Three cascaded inverters 1002 are used in this example to get a sharp-edge rail-to-rail output signal. The reference voltage can be derived from the supply voltage of the LO buffers with a resistive voltage divider 1016. A single feedback loop is used instead of four separate loops to reduce silicon area and power dissipation. The average of the DC components of the four LO signals (V _OU T, _DC ) can be measured with a passive RC filter formed by four measurement resistors 1012, level shifting resistor 1014, RC structure 1006, and capacitor 1010. An amplifier 1008 can be implemented, for example,

as a Miller-compensated operational amplifier with a PMOS differential pair at the input. If duty cycle is low, the DC component and the reference voltage are close to the negative supply, which may not be optimal input voltages for the feeedback amplifier. The input voltages of the amplifier can be increased to a higher and more suitable value by using a level shifting resistor 1014. The reference voltage V _REF has to be selected accordingly.

Fig. 1 is a schematic view of a TIF according to an embodiment of the present invention. As indicated above, four pulse trains having π/2 phase shifts are needed in a TIF and the time when a switch is on during one period can be selected to less than 25% of the period. These four signals can be generated from a double-frequency LO (not shown) by using a divide-by-two circuit 100 followed by a NAND circuit 102, which can be both implemented with, for example, source-coupled logic (SCL) technique. The four LO signals can also be generated by using, for example, a digital divider (not shown) , or some other technique. In practice, the four output signals 104 of the NAND circuit are not sharp-edge rail-to-rail pulses, as illustrated in Fig. 3, and, therefore, they are preferably amplified and limited before the switches 106 of the TIF. Since the output signals 104 of the NAND 102 are not sharp-edge rail-to-rail pulses, the DC voltage, i.e. biasing, at the inputs of the four buffers 105 affects the duty cycle of the four output signals 108. This means that the supply voltage, temperature, and process parameters affect the duty cycle. The duty cycle of the signals 108 controlling the switches of the TIF affects the relation between the time when a switch is on during one period and the period and, therefore, the passband and stopband impedances, and selectivity. The passband impedance of a TIF affects the gain at the input of the TIF. By employing a feedback loop 110, it is possible to maintain a constant duty cycle regardless of

the LO frequency, input signal swing of the double- frequency LO, supply voltage, temperature, and process parameters . The decreased variation in duty cycle means a decreased variation in the relation between the time when a switch is on during one period and the period.

Conventionally, in TIF implementations, the time when a switch is on during one period has been approximately 25% of the period. The invented structure improves the selectivity of a TIF by controlling the pulse train signals controlling the switches of the TIF in such a way that the time when a switch is on during one period is less than 25% of the period. The relation between the time when a switch is on during one period and the period can also be set to desired value determined by a reference signal. In addition, the invented structure maintains the duty cycle of the rail- to-rail LO signals controlling the switches of the TIF constant regardless of the LO frequency, input signal swing of the double-frequency LO, supply voltage, temperature, and process parameters. This means that the variation of the relation between the time when a switch is on during one period and the period is reduced, which decreases the unwanted variations in passband and stopband impedances and selectivity due to the reasons mentioned above.

The relation between the DC component of the rail- to-rail LO signal and the supply voltage is equal to the duty cycle. If the input signals of the buffers, which produce the LO signals, are not sharp-edge pulses, DC level, i.e. biasing, at the inputs of the buffers affects the duty cycle of the LO signals. Therefore, the feedback loop 110, which measures the difference between the DC component of the LO signal and a reference value and controls the bias voltage at the input of the LO buffer in such a way that the difference is minimized, sets the duty cycle to the desired value. The relation between the time when a switch is on during one period and the period

is also set to the desired value. A single feedback loop can be used even when there is more than one LO buffer, as is illustrated in Figs 8 to 10. The average of the DC components of all LO signals can thus be measured and compared to a single reference value. In addition, a single bias voltage controlled by the feedback loop can be used at all LO buffer inputs.

The invention can be implemented with CMOS and BiCMOS technologies and can be operated from low supply voltages, like 1.2V. The nodes LO_0, LO_90, LO_180, and LO_270 in Fig. 10 are connected to the gates of the switches of the TIF, as shown in Figs 1 and 2. AC couplings can be used before the gates of the switches if necessary. Since a NAND circuit, which generates the signals shown in Fig. 3, is used before the circuit of Fig. 10 and NMOS switches are used in the TIF, an odd number of cascaded inverters 1002 enables to get output signals with a duty cycle of 25% or less. In this case, the duty cycle is approximately equal to the relation between the time when a switch is on during one period and the period. Three cascaded inverters 1002 are used to get a sharp-edge rail-to-rail output signal. The output signals of the four LO buffers are shown in Fig. 4. The reference voltage can be derived from the supply voltage of the LO buffers with a resistive voltage divider 602, as illustrated in Fig. 6. A single feedback loop 1004 can be used instead of four separate loops, as illustrated in Fig. 10, to reduce silicon area and power dissipation. Of course can separate feedback loops be used as well. The average of the DC components of the four LO signals

Vooτ,Dcr as illustrated in Fig. 10, can be measured with the passive RC filter formed by the four measurement resistors 1012, level shifting resistor 1014, RC structure 1006, and capacitor 1010. The amplifier 1008 can be implemented, for example, as a Miller-compensated opamp with a PMOS differential pair at the input.

The advantages of the invention include (but are not limited to) :

(1) The selectivity of a TIF is improved, which means a better tolerance of blocking signals. (2) The duty cycle of the signals controlling the switches of the TIF affects the passband and stopband impedances, and selectivity. By employing the feedback loop 1004, it is possible to maintain a constant duty- cycle regardless of LO frequency, input signal swing of the double-frequency LO, supply voltage, temperature, and process parameters. This means that the variation of the relation between the time when a switch is on during one period and the period is reduced, which decreases the unwanted variations in passband and stopband impedances and selectivity.

(3) The invention can be implemented with a low supply voltage, like 1.2V, which is important in submicron CMOS processes. To be able to realize single- chip radio IC including RF, analog, and digital circuitry on the same chip, modern CMOS processes can be utilized and RF circuits are then preferably able to operate at low voltage.

(4) The implementation of the invention requires only a small amount of silicon area and (5) increase in the power dissipation is very small. (6) The noise contribution of the additional circuit required in the implementation of the invention is insignificant.

Fig. 11 schematically illustrates a transferred impedance filter 1100 according to an embodiment of the present invention, where a DC component related to at least one of the signal pulse trains is measured 1102 by directly measuring the signal pulse trains 1104 that is used for controlling switches 1106 of a switching stage 1108 of the transferred impedance filter 1100.

Fig. 12 schematically illustrates a transferred impedance filter 1200 according to an embodiment of the

present invention, where a DC component related to at least one of the signal pulse trains is measured 1202 by measuring a separate signal 1203 related to the signal pulse trains 1204 that is used for controlling switches 1206 of a switching stage 1208 of the transferred impedance filter 1200. The separate signal 1203 can be provided by the oscillator 1210 providing the signal pulse trains 1204.

Fig. 13 illustrates a transferred impedance filter according to an embodiment of the present invention, which is similar to the embodiment demonstrated with reference to Fig. 1, but in the embodiment illustrated in Fig. 13, signals IIP, UN, IQP, IQN are illustrated to be generated in an arbitrary fashion, in contrary to the more specific fashion illustrated with reference to Fig. 1.

It should be noted that any of the disclosed ways of generating signals, providing feedback, measurement of particulars of signals, such as DC component, etc., of the demonstrated embodiments can be combined in any- feasible fashion.

Fig. 2 illustrates an exemplary switching stage 202 and an example of the impedances 204 of a TIF. The switching stage 202 consists of eight switches 206 and two similar transferred impedances 208, which in this case are formed by capacitors. A TIF is usually a balanced circuit, but can be single-ended as well, since balanced topologies are generally used in mixed-mode RFICs to reject the interference from digital circuits, supply voltage, or silicon substrate.

The relation between the time when a switch of the TIF is on during one period of the signal controlling the switch and the period affects the passband and stopband impedances of the TIF. If this relation decreases, both the passband and stopband impedances increase. However, the relation between the passband and stopband impedances increases. When the relation is decreased, the widths of

the switches can be increased accordingly to maintain constant resistive passband impedance. The result is an increased selectivity. An example is shown in Fig. 5. Conventionally, in TIF implementations , the time when a switch is on during one period has been approximately 25% of the period to prevent inphase and quadrature branches from interfering with each other. However, when this relation is reduced from 25% to 20%, the selectivity of a TIF, which uses capacitors as the transferred impedances, is improved by approximately 3dB. If the circuit following the TIF limits the out-of-band linearity of the receiver IC and the improvement in the selectivity of the TIF due to the invention is δL, the improvements in the out-of-band IIP3 and IIP2 of the receiver are 1.5*δL and 2 ^χ δL, respectively. If δL = 3dB, the out-of-band IIP3 and IIP2 of the receiver are improved by 4.5dB and 6.OdB, respectively.