AN ATTENUATOR

Abstract

An attenuator for attenuating a signal is disclosed. The attenuator comprises a differential input port with a positive input node and a negative input node to receive the signal; and a differential output port with a positive output node and a negative output node to output the attenuated signal. The attenuator further comprises a first switched resistor network connected between the positive input node and the positive output node; and a second switched resistor network connected between the negative input node and the negative output node. Further a pair of compensation paths is connected to the first and second switched resistor networks for cancellation their parasitic leakages, where a first compensation path is connected between the positive input node and the negative output node, and a second compensation path is connected between the negative input node and the positive output node. The attenuator further comprises a control circuit to generate control signals for controlling the first and second switched resistor networks.

Claims

1. An attenuator for attenuating a signal comprising: a differential input port with a positive input node and a negative input node to receive the signal; a differential output port with a positive output node and a negative output node to output the attenuated signal; a first switched resistor network connected between the positive input node and the positive output node; a second switched resistor network connected between the negative input node and the negative output node; a pair of compensation paths for cancellation of parasitic leakages in the first and second switched resistor networks, wherein a first compensation path is connected between the positive input node and the negative output node, and a second compensation path is connected between the negative input node and the positive output node; and a control circuit to generate control signals for controlling the first and second switched resistor networks.

2. The attenuator according to claim 1, wherein the attenuator further comprises a first pair of inductors connected at the differential input port and a second pair of inductors connected at the differential output port.

3. The attenuator according to claim 1, wherein the control circuit is further configured to generate control signals for controlling the pair of compensation paths.

4. The attenuator according to claim 1, wherein each of the first and second switched resistor networks is a Pi type resistor network comprising two switchably variable parallel resistors and one switchably variable series resistor.

5. The attenuator according to claim 1, wherein each of the first and second switched resistor networks is a T type resistor network comprising one switchably variable parallel resistor and two switchably variable series resistors.

6. The attenuator according to claim 4, wherein the switchably variable series resistor comprises one or more switched resistor branches and a by-pass path.

7. The attenuator according to claim 6, wherein the by-pass path comprises a switch, and wherein a bootstrap path comprising a capacitor in series with a resistor is connected between a gate and a drain or source of the switch.

8. The attenuator according to claim 4, wherein the switchably variable parallel resistor comprises one or more switched resistor branches connected in series with a resistor.

9. The attenuator according to claim 6, wherein each of the one or more switched resistor branches comprises a resistor in series with a switch, and wherein for one or more switches, a bootstrap path comprising a capacitor in series with a resistor is connected between a gate and a drain or source of the switches depending on size of the one or more switches.

10. The attenuator according to claim 1, wherein each of the pair of compensation paths comprises one or more switched capacitor branches, and each switched capacitor branch comprises a capacitor connected in series with a resistor.

11. The attenuator according to claim 10, wherein each of the switched capacitor branch further comprises a switch connected in series with the capacitor and the resistor.

12. The attenuator according to claim 2, wherein the first and second pair of inductors are mutually coupled inductors.

13. An electronic device comprising an attenuator according to claim 1.

14. The electronic device according to claim 13, wherein the electronic device is a transceiver.

15. The electronic device according to claim 13, wherein the electronic device is a radio base station.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Examples of embodiments herein are described in more detail with reference to attached drawings in which:

[0010] FIG. 1 is a general block view of an attenuator according to embodiments herein.

[0011] FIG. 2 is a schematic block view illustrating an attenuator with Pi-type switched resistor networks according to embodiments herein.

[0012] FIG. 3 is a schematic block view illustrating an attenuator with T-type switched resistor networks according to embodiments herein.

[0013] FIG. 4 is a schematic diagram illustrating a switchably variable series resistor according to embodiments herein.

[0014] FIG. 5 is a schematic diagram illustrating a switchably variable parallel resistor according to embodiments herein.

[0015] FIG. 6 is a schematic diagram illustrating a compensation path according to embodiments herein.

[0016] FIG. 7 is a simplified schematic diagram illustrating an equivalent circuit for the Pi-type switched resistor network in FIG. 2.

[0017] FIG. 8 is a diagram showing frequency response for an attenuator according to one embodiment herein.

[0018] FIG. 9 is a diagram showing frequency response for an attenuator according to another embodiment herein.

[0019] FIG. 10 is a drawing showing an example layout of compensation inductors according to embodiments herein.

[0020] FIG. 11 is a block diagram illustrating an electronic device in which embodiments herein may be implemented.

DETAILED DESCRIPTION

[0021] A general view of an attenuator 100 for attenuating a signal according to embodiments herein is shown in FIG. 1. The attenuator 100 comprises a differential input port Inp/Inn with a positive input node Inp and a negative input node Inn to receive the signal, and a differential output port Outp/Outn with a positive output node Outp and a negative output node Outn to output the attenuated signal.

[0022] The attenuator 100 further comprises a first switched resistor network 102 connected between the positive input node Inp and the positive output node Outp, and a second switched resistor network 104 connected between the negative input node Inn and the negative output node Outn. The first and second switched resistor networks 102/104 are configured to create desired attenuation.

[0023] The attenuator 100 further comprises a pair of compensation paths for cancellation of parasitic leakages in the first and second switched resistor networks 102/104. As shown in FIG. 1, a first compensation path 106 referred to as Neutralization Compensation NC 106 in FIG. 1, is connected between the positive input node Inp and the negative output node Outn, and a second compensation path 108 referred to as NC 108 in FIG. 1, is connected between the negative input node Inn and the positive output node Outp.

[0024] Further, a control circuit 110 is also comprised in the attenuator 100 to generate control signals for controlling the first and second switched resistor networks 102/104.

[0025] According to some embodiments, the control circuit 110 in the attenuator 100 may be further configured to generate control signals for controlling the pair of compensation paths 106/108.

[0026] The control signals CtrIS, CtrIP and CtrINC are generated by the control circuit 110 through digital data interface. The first and second switched resistor networks 102/202 are controlled by the control signals CtrIP and CtrIS so that the resistance values of the first and second switched resistor networks 102/202 are tunable. The control signal CtrINC is to control the compensation path 106/108.

[0027] According to some embodiments, the attenuator 100 may further comprise a first pair of inductors L1a/L1b connected at the differential input port Inp/Inn and a second pair of inductors L2a/L2b connected at the differential output port Outp/Outn. In these embodiments, the first and second switched resistor networks 102/104 are coupled to the differential input/output port via the first and second pair of inductors, respectively. The first and second pair of inductors L1a/L1b, L2a/L2b are used to compensate the bandwidth of the attenuator 100. According to some embodiments, the first and second pair of inductors are mutually coupled inductors, with coupling coefficients M1 and M2 respectively.

[0028] According to one embodiment, the attenuator 100 may be implemented by circuits shown in FIG. 2, where the first and second switched resistor networks 102/104 may be implemented by Pi-type switched resistor networks 202/204.

[0029] As shown in FIG. 2, the Pi-type switched resistor networks 202/204 each comprises two switchably variable parallel resistors Rp and one switchably variable series resistor Rs.

[0030] According to one embodiment, the attenuator 100 may be implemented by circuits shown in FIG. 3, where the first and second switched resistor networks 102/104 may be implemented by T-type switched resistor networks 302/304.

[0031] As shown in FIG. 3, the T-type switched resistor networks 302/304 each comprises one switchably variable parallel resistor Rp and two switchably variable series resistors Rs.

[0032] According to some embodiments, the switchably variable series resistor Rs may be implemented by circuits shown in FIG. 4, where FIG. 4(a) shows a symbol of the switchably variable series resistor Rs, FIG. 4(b) shows an example structure of the switchably variable series resistor Rs and FIG. 4(c) shows each branch of the switchably variable series resistor Rs in more detail.

[0033] As shown in FIG. 4(b), the switchably variable series resistor Rs comprises one or more switched resistor branches 400, 401, . . . 40n, each of the one or more switched resistor branches 400, 401, . . . 40n comprises a resistor Rs.sub.0, Rs.sub.1, . . . Rs.sub.n in series with a switch. The switchably variable series resistor Rs further comprises a by-pass path 40b.

[0034] As shown in FIG. 4(c), for one or more switches, a bootstrap path comprising a capacitor in series with a resistor is connected between a gate and a drain or source of the switch depending on size of the one or more switches. Using bootstrap path can effectively improve the linearity of the switch. However, for smaller size of the switches, the bootstrap path may not be needed, for example, in branch 400, there is no bootstrap path connected between a gate and a drain or source of the switch Ts.sub.0.

[0035] As shown in FIG. 4(c), the by-pass path 40b comprises a switch Tb, and a bootstrap path comprising a capacitor Cb in series with a resistor Rb is connected between a gate and a drain or source of the switch Tb.

[0036] When the switchably variable series resistor Rs comprises multiple switched resistor branches 400, 401, . . . 40n, the control signal CtrIS generated by the control circuit 110 comprises multiple control signlas Ctrls0, Ctrls1, . . . , Ctrlsn to control the switch in each branch, where n is positive integer. The control signal CtrIS further comprises Ctrlspass to control the by-pass path 40b.

[0037] According to some embodiments, the switchably variable parallel resistor Rp may be implemented by circuits shown in FIG. 5, where FIG. 5(a) shows a symbol of the switchably variable parallel resistor Rp, FIG. 5(b) shows an example structure of the switchably variable parallel resistor Rp and FIG. 5(c) shows each branch of the switchably variable parallel resistor Rp in more detail.

[0038] As shown in FIG. 5(b), the switchably variable parallel resistor Rp comprises one or more switched resistor branches 500, 501, . . . 50m, and the switchably variable parallel resistor Rp further comprises a resistor (RPb) connected in series with the switched resistor branches 500, 501, . . . 50m. Each of the one or more switched resistor branches 500, 501, . . . 50m comprises a resistor Rp.sub.0, Rp.sub.1, . . . Rp.sub.m in series with a switch. As shown in FIG. 5(c), for one or more switches, a bootstrap path comprising a capacitor in series with a resistor is connected between a gate and a drain or source of the switch depending on size of the one or more switches. For example, for smaller size switch Tp.sub.0 in branch 500, there is no bootstrap path connected between a gate and a drain or source of the switch Tp.sub.0.

[0039] When the switchably variable parallel resistor Rp comprises multiple switched resistor branches 500, 501, . . . 50m, the control signal CtrIP generated by the control circuit 110 comprises multiple control signals CtrIp0, Ctrlp1, . . . , Ctrlpm to control the switch in each branch, where m is positive integer.

[0040] According to some embodiments, the compensation path 106/108 may be implemented by circuits shown in FIG. 6, where FIG. 6(a) shows a symbol of the compensation path 106/108, FIG. 6(b) shows an example structure of the compensation path 106/108, and FIG. 6(c) shows another example structure of the compensation path 106/108. As shown in FIGS. 6(b) and (c), each compensation path 106/108 comprises one or more switched capacitor branches, and each switched capacitor branch comprises a capacitor, Cnc, Cnc.sub.0, C.sub.nc1, . . . Cnc.sub.k connected in series with a resistor, Rnc, Rnc.sub.0, Rnc.sub.1, . . . Rnc.sub.k. In by-pass mode, the compensation paths for cancelling the parasitic leakages can be switched off as to reduce the insertion loss.

[0041] According to some embodiments, each switched capacitor branch in the compensation path 106/108 further comprises a switch connected in series with the capacitor Cnc, Cnc.sub.0, Cnc.sub.1, . . . Cnc.sub.k and the resistor Rnc, Rnc.sub.0, Rnc.sub.1, . . . Rnc.sub.k.

[0042] When the compensation path 106/108 comprises multiple switched capacitor branches, the control signal CtrINC comprises multiple control signals CtrInck, CtrInc1, . . . CtrInc0 to control the switch in each branch, where k is an integer larger or equal to 0.

[0043] Implementation details, performance and advantages of the attenuator 100 according to embodiments herein are described below. As described above, the switched resistor network 102/104 are used to set the desired attenuation. In order to show the relation between the desired attenuation and the resistor value, example design equations are derived for the Pi-type switched resistor network 202/204. A simplified equivalent circuit of the Pi-type switched resistor network 202/204 without the compensation path is shown in FIG. 7, where FIG. 7(a) shows the Pi-type switched resistor network 202/204 connected in differential, and FIG. 7(b) shows its singleended part.

[0044] For simplicity, the Pi-type switched resistor network 202/204 connected in single-ended in FIG. 7(b) is analysed, where the input and output port impedances R.sub.L=50 are added to it. In an ideal condition for matching requirement, the switched resistor network 202/204 should match the input and output port impedance R.sub.L=50 ohm, that is

R.sub.L=Rp(R.sub.LRp+Rs)Eq. (1)

[0045] While in matched condition, the voltage gain vg is given by

[00001]$\begin{array}{cc}\mathrm{vg}=\frac{\mathrm{Rp}{R}_L}{\mathrm{Rs}+\mathrm{Rp}{R}_L}=\frac{\mathrm{Rp}.Math.R_L}{\mathrm{Rs}.Math.\mathrm{Rp}+\mathrm{Rs}.Math.R_L+\mathrm{Rp}.Math.R_L}& \mathrm{Eq}..Math.\left(2\right)\end{array}$

[0046] From matching requirement, it is given

Rs.Math.Rp.sup.2.Math.2.Math.R.sub.L.sup.2.Math.Rp.Math.R.sub.L.sup.2.Math.Rs=0Eq. (3)

[0047] So Rs may be expressed as

[00002]$\begin{array}{cc}\mathrm{Rs}=\frac{2.Math..Math.R_L^2.Math.\mathrm{Rp}}{{\mathrm{Rp}}^2-R_L^2}& \mathrm{Eq}..Math.\left(4\right)\end{array}$

[0048] Replace Rs in the attenuation or gain expression (2) with Rs expression in (4), which gives

[00003]$\begin{array}{cc}\mathrm{vg}=\frac{\mathrm{Rp}-R_L}{\mathrm{Rp}+R_L}=\frac{1-}{1+}& \mathrm{Eq}..Math.\left(5\right)\end{array}$

[0049] Where

[00004]$=\frac{R_L}{\mathrm{Rp}}<1,$

for an ideal pure resistor case.

[0050] Rp may be written as a function of desired attenuation

[00005]$\begin{array}{cc}\mathrm{Rp}=\frac{R_L.Math.\left(1+\mathrm{vg}\right)}{1-\mathrm{vg}}& \mathrm{Eq}..Math.\left(6\right)\end{array}$

[0051] Where

[00006]$\mathrm{vg}={10}^{-\frac{\mathrm{VG}}{20}},$

and VG is the required or desired attenuation in unit of dB.

[0052] It should be noted that above equations are example design equations for Pi-type switched resistor network, the skilled person will appreciate that for T-type switched resistor network, design equations will be different.

[0053] Turning back to FIG. 4, where the structure of the switchably variable series resistor Rs according to one embodiment herein is shown. As described above, the switchably variable series resistor Rs comprises a by-pass path 40b, as shown in dashed-line box at the bottom of FIGS. 4(b) and (c). When there is no need to attenuate the input signal, the attenuator 100 is set to a by-pass mode and all switched resistor branches 400, 401, . . . , 40n are by passed by turning on the switch Tb in the by-pass path 40b. The control signal CtrlSpass is set to logic high during the by-pass mode. As shown in FIG. 4(c), in the by-pass path 40b, a bootstrap path comprising a capacitor Cb in series with a resistor Rb is connected between the gate and drain or source of the switch Tb. The purpose of the capacitor Cb is to bootstrap the gate voltage of the switch Tb by using the input signal at its drain or source, and to keep the voltage between the gate and drain or source close to a constant DC voltage provided by the control signal CtrlSpass. It means that the conducting resistor Ron of the MOS switch transistor Tb is close to constant, thus improving the linearity of the switch Tb. Resistor Rb inserted in the bootstrap path is to avoid increasing capacitive load in the signal path and also reduce the leakage caused by the parasitic capacitance of the switch transistor Tb. Resistor Rg is a resistor in series between the control node and the gate of the switch transistor Tb, and the time constant of RgCb should be much larger than the period of the lowest operating frequency. In this way, lower insertion loss for the input signals at higher frequencies is achieved. In other attenuation modes, the control signal CtrlSpass is set to logic low, the switch Tb is turned off, and the by-pass path is disconnected.

[0054] The switched resistor branches 400, 401, . . . , 40n are used for different attenuation settings including the by-pass mode. In the by-pass mode, all control signals Ctrls0, Ctrls1, . . . , Ctrlsn are set to logic high, so all switches in the switched resistor branches 400, 401, . . . , 40n are conducting, this reduces the insertion loss further.

[0055] While in other attenuation settings, only some of the switched resistor branches 400, 401, . . . , 40n conduct, and provide a resistance value Rs according to the expression given by Eq. (4). Due to silicon process variations, resistance value for an on-chip resistor may vary in a range of 25%. Therefor in practice, more control bits may be added for trimming the resistance.

[0056] The resistors in the switched resistor branches 400, 401, . . . , 40n may be designed, e.g. in a binary weighted size, i.e. Rs.sub.0=R, Rs.sub.1=R/2, Rs.sub.n=R/2.sup.n. For the switched resistor branch 40n, it may be viewed as 2.sup.n branches of unit resistor cell R in parallel. So Rs.sub.n is the smallest one in resistance, and switch Ts.sub.n is the largest one in size. Enlarging the size of Ts.sub.n may reduce the impact of the conducting resistance Ron of the switch Ts.sub.n, so as to improve the linearity, but at a cost of introducing larger parasitic capacitances in the signal path. Therefore, bootstrap path used here to improve the linearity is more effective for the larger size switches, as shown in FIG. 4(c). However, for smaller size switches, the bootstrap path may not be needed. As seen from FIG. 4(c), for the switched resistor branch 400, there is no bootstrap path for switch Ts.sub.0.

[0057] For the switchably variable parallel resistor Rp, as shown in FIG. 5, binary weighted resistors may also be used in the switched resistor branches 500, 501, . . . 50m. In the by-pass mode, all switched resistor branches are disconnected, so the insertion loss is minimized.

[0058] In other attenuation modes, the value of Rp may be chosen according Eq. (6). Bootstrap paths are also optionally used depending on size of the switch, e.g. a bootstrap path is used for largest switch Tp.sub.m in the switched resistor branch 50m, and ignored for the smallest switch Tp.sub.0 in the switched resistor branch 500, as shown in FIG. 5(c).

[0059] In practice, due to parasitic capacitances between the three nodes, i.e. gate, drain and source of the switch Tb, parasitic leakages exist, and these leakages damage attenuation settings in deep attenuation steps. FIG. 8 shows attenuation level versus frequency for the by-pass mode and different attenuation steps, where the desired attenuation levels are plotted in solid lines, while the real attenuation levels due to the parasitic capacitances are plotted in dotted lines. It may be seen that the parasitic leakages due to the parasitic capacitances ruin the desired attenuation levels, especially at higher frequencies.

[0060] As shown in FIGS. 1-3, the attenuator 100 is in a differential structure, this makes it possible to add a pair of compensation path NC 106/108 to neutralize or cancel the leakages. The cancellation is achieved by cross-coupling the pair of compensation paths 106/108, i.e. the first compensation path 106 is connected between the positive input node Inp and the negative output node Outn, and the second compensation path 108 is connected between the negative input node Inn and the positive output node Outp. In this way, any leakage signals at the positive input node are coupled to the negative output node, thereby cancel any leakage signals at the negative output node. In the same way, any leakage signals at the negative input node are coupled to the positive output node, thereby cancel any leakage signals at the positive output node. This results in a good attenuation performance, especially at high frequencies and in deep attenuation steps, as shown in the solid lines in FIG. 8, where the slopes of attenuation levels for all steps are the same after the compensation.

[0061] As described above, the compensation path 106/108 comprises one or more switched capacitor branches. When only one switched capacitor branch is used, as shown in FIG. 6(b), to reduce the insertion loss, the switch transistor is turned off in the by-pass mode, so the compensation path is disconnected from the switched resistor networks. However, it should be noted that switch transistors in the compensation paths are optional, i.e. the compensation paths may only comprise capacitors and resistors, and may always connected to the switched resistor networks in the by-pass mode and at other attenuation settings.

[0062] In FIG. 6(c), several switched capacitor branches are used, so it will give more freedom to tune compensation for different attenuation settings or levels. As it is a programmable approach, it may also deal with process variations and design errors due to inaccurate extraction of the values for the capacitors and resistors in design tools.

[0063] From FIG. 8, it may be seen that the attenuation levels have slopes along the frequency axis. This means that insertion loss at higher frequencies is higher than that of at lower frequencies, this leads to a narrower bandwidth for the attenuator. To improve the bandwidth, a pair of inductors with mutual coupling may be employed at both the input port and output port. As shown in FIGS. 1-3, the first pair of inductors L1a/L1b is connected at the differential input port Inp/Inn to resonate with the input parasitic capacitance at desired frequencies, and the second pair of inductors L2a/L2b is connected at the differential output port Outp/Outn to resonate with the output parasitic capacitance at desired frequencies. In this way, the slops of the attenuation levels are reduced, and the bandwidth of the attenuator 100 is extended. This is illustrated in FIG. 9, where the solid curves are frequency responses, i.e. attenuation levels versus frequency, with inductor compensation, and the dotted lines are uncompensated frequency responses.

[0064] An example layout of the two pairs of mutual coupling inductors is shown in FIG. 10, where the layout of L1a/L1b and L2a/L2b is designed in an interleaved way, as to save silicon area and increase the Q factor by utilising mutual inductance between the two parts. Of course other types of separated inductors may also be used, but at a cost of large silicon area and a bit more insertion loss.

[0065] To summarise the discussions above, advantages of various embodiments of the attenuator 100 include: [0066] Accurate attenuation levels or steps: The switched resistor networks in Pi or T structure make resistances tunable and may also cope with process variations, so accurate attenuation levels and steps are provided. [0067] Larger range of attenuation: Compensation paths are used to cancel the parasitic leakages so that attenuation levels or steps at deep attenuation are more accurate, therefore the attenuator 100 achieves a larger range of attenuation. [0068] Higher linearity and low insertion loss: Bootstrap paths are used in some embodiments for larger size switch transistors and for the by-pass path so that linearity is improved and insertion loss is reduced. [0069] Wide bandwidth: Compensation inductors are used in some embodiments at both input port and output port to resonate with the input and output parasitic capacitances respectively, so the bandwidth of the attenuator 100 is extended.

[0070] The attenuator 100 according to the embodiments herein may be employed in various electronic devices. FIG. 11 shows a block diagram for an electronic device 1100, which may be, e.g. a radio frequency transceiver, a radio frequency frontend, a wireless communication device, such as a user equipment or a mobile device and/or a base station, a multi-antenna systems in a radio base station, or any general electronic circuit or equipment, such as an automatic gain control circuit, a measurement equipment etc. The electronic device 1100 may comprise other units, where a processing unit 1110 is shown, which may interactive with the control circuit 110 in the attenuator 100 for different attenuation settings or operating modes.

[0071] Those skilled in the art will understand that although switch transistors in the switched resistor array Rs, Rp, the by-pass path 40b and the compensation path 104/106 of the attenuator 100 as shown in FIGS. 4-6 are Field-Effect Transistors (FET), any other types of transistors, e.g. Metal-Oxide-Semiconductor FET (MOSFET), Junction FET (JFET), Bipolar Junction Transistors (BJT) etc., may be comprised in the attenuator 100. When using the word comprise or comprising it shall be interpreted as non-limiting, i.e. meaning consist at least of.

[0072] The embodiments herein are not limited to the above described preferred embodiments. Various alternatives, modifications and equivalents may be used. Therefore, the above embodiments should not be taken as limiting the scope of the invention, which is defined by the appending claims.