Method and apparatus for an LNA with high linearity and improved gain control

Abstract

An analog signal processing circuit comprises a bias circuit, a first circuit including a control input that communicates with the bias circuit, a first terminal that generates an output current, and a second terminal, and a device that communicates with the second terminal of the first circuit, that includes a variable resistor, and that has a resistance that is modulated in response to an input signal to the analog signal processing circuit.

Description

TECHNICAL FIELD

An aspect of the invention relates to Metal Oxide Semiconductor Field Effect Transistor (MOSFET) amplifiers.

BACKGROUND

There is a growing demand for mobility in today's world. The rapid progress in the wireless industry makes the ubiquitous connection possible. Radio Frequency (RF) transceivers are important components for wireless devices. The majority of the RF ICs used in the wireless communication were implemented using either GaAs or silicon bipolar technologies. Not until recently, when the continuous scaling of CMOS technology brought the cutoff frequency (fT) of MOS transistors up to multi-tens of GHz, were such circuits built in CMOS technology possible. The advantage of using Complementary Metal-Oxide-Semiconductor (CMOS) RF is that it can be integrated with digital functions easily. As a result, it is possible to incorporate the whole system on one single chip which yields low cost, small form factor wireless devices. A Low Noise Amplifier (LNA) is an important building block in the wireless transceiver. For LNAs, the gain linearity applied to a signal is an important operating characteristic, especially when the incoming signal is large. Under that condition, amplification by the LNA actually could be greater or smaller than one, and the noise contribution from the LNA may be negligible compared to the input signal. In fact, the linearity of the LNA becomes the most important figure of merit. Gain linearity is generally characterized as a 1 dB compression point or third order Input Intercept Point (IIP3). The gain linearity is typically related to the transconductance of a MOSFET in an input stage of the amplifier. For example, the transconductance of a MOSFET operating in the saturation region is constant only when the input signal is small. When the input signal is large, the transconductance may vary as a function of the input signal, leading to nonlinear amplification of the signal. Source degeneration may be employed at lower frequencies to increase the linearity of the input stage. However, at higher frequencies source degeneration may not be effective due to the large parasitic capacitance of the device. Also, source degeneration may increase power consumption due to the relative low gm/Id for the MOSFET in comparison with a bipolar device. In addition, Gain control is also very important in practical applications since the gain of the LNA could vary with process and temperature if not properly controlled.

SUMMARY

An LNA comprising an input stage to amplify an input signal. The LNA being particularly suitable for amplifying large input signals. The input stage includes a linearized transconductance and has reduced gain variations in response to changes in process and environmental conditions.

The details of one or more embodiments of the invention are set forth in the accompanying drawings and the description below. Other features, objects, and advantages of the invention will be apparent from the description and drawings, and from the claims.

DESCRIPTION OF DRAWINGS

FIG. 1 is a block diagram of an aspect of a transceiver.

FIG. 2A is a schematic diagram of an aspect of an LNA.

FIG. 2B is a schematic diagram of an aspect of an LNA.

FIG. 2C is a schematic diagram of an aspect of an LNA.

FIG. 2D is a schematic diagram of an aspect of an amplifier.

FIG. 3 is a schematic diagram of an aspect of a bias circuit for a linear input stage.

FIG. 4 is a schematic diagram of an aspect of an amplifier.

FIG. 5 is a flow diagram of an aspect of an operation for generating a linear input stage.

FIG. 6 is a flow diagram of an aspect of an operation for biasing a linear input stage.

Like reference symbols in the various drawings indicate like elements.

DETAILED DESCRIPTION

FIG. 1 shows an aspect of a wireless transceiver 10 for communicating information. The wireless transceiver 10 may include a Low Noise Amplifier (LNA) 12 for amplifying an input signal. An input signal 14 to the LNA 12 may be amplified by a linear input stage 18 constructed in accordance with the principles of the invention. A bias circuit 16 may supply bias signals to the linear input stage 18 in accordance with the principles of the invention. The LNA 12 preferably includes both the bias circuit 16 and the linear input stage 18. However, the LNA 12 may include the bias circuit 16 combined with a conventional linear input stage, or the linear input stage 18 combined with a conventional bias circuit. An output stage 20 may provide further amplification of the input signal.

A mixer 22 may combine the amplified input signal with a Radio Frequency (RF) LO signal 24. A filter 26 and amplifier 23 may filter and amplify the combined signal, and mix the generated signal with an Intermediate Frequency (IF) LO signal. An analog-to-digital converter (ADC) 28 may convert the mixed signal to a digital signal for further processing.

A digital-to-analog converter 27 may convert a digital signal to an analog signal for transmission by a transmitter 25.

FIG. 2A shows an aspect of an LNA input stage 200 for amplifying an input signal, v_in. The LNA input stage 200 may be constructed using any CMOS process including NMOS and PMOS. The input signal, v_in, to the LNA input stage 200 modulates the resistance of a first device 202 that is connected to a second device 204 having a low input impedance. Due to the low input impedance of the second device 204, the voltage v_x at the junction of the first and second devices 202 and 204 may remain relatively constant. The input stage 200 is configured so that changes in the input signal cause linearly proportional changes in conductance of the first device 202. In the case where v_x is relatively constant and the conductance of the first device 202 changes in linear proportion to changes in the input signal, iout is about linearly proportional to V_in.

FIG. 2B shows an aspect of an NMOS implementation 210 of the LNA input stage 200. The resistance of a first device 212 is modulated in response to an input signal v_in. An NMOS transistor 214 in combination with an amplifier 216 provides a low impedance at the junction of the NMOS transistor 214 and the first device 212.

FIG. 2C shows an aspect of another NMOS implementation 220 of the LNA input stage 210. Here, the resistance of a first NMOS transistor 222 is modulated in response to an input signal v_in. The first NMOS transistor is biased into the triode region. A second NMOS transistor 224 in combination with an amplifier 226 provide a low impedance at the junction of the first and second NMOS transistors 222 and 224.

FIG. 2D shows an aspect of an amplifier 30 for amplifying an input signal in accordance with the principles of the LNA input stage 200. Here, a linear input stage 32 may include an upper MOSFET, MB, 34 and a lower MOSFET, MA, 36 connected in a cascode configuration. The input impedance of the upper MOSFET 34 at the junction of the upper and lower MOSFETs 34 and 36, may be made low relative to the lower MOSFET 36 by controlling the relative sizes of the upper and lower MOSFETs 34 and 36. The linear input stage 32 is preferably constructed as an integrated circuit using Complementary Metal Oxide Semiconductor (CMOS) technology, but other circuit technologies may also be used including discrete MOSFETs. Both NMOS and PMOS devices may be used. An input signal is AC coupled through a capacitor 40 to the gate of the lower MOSFET 36. A bias circuit 38 biases the upper MOSFET 34 into the saturation region and the lower MOSFET 36 into the triode region. Here, the lower MOSFET 36 acts as a variable resistor changing conductance in linear proportion to changes in the input signal. The impedance of the junction of MOSFETs 34 and 36 may be made lower by selecting the transconductance, g_m, of the upper MOSFET 34 to be larger than both g_ds and g_m of the lower MOSFET 36 so that Vds of the lower MOSFET 36 remains relatively constant over changes in the input signal. For example, an input switch ratio defined as the ratio of the size of the upper MOSFET 34 to the size of the lower MOSFET 36 may be selected to be at least four, so that g_m of the upper MOSFET 34 is greater than both the g_ds and g_m of the lower MOSFET. One aspect of the invention recognizes that if the Vds of the lower MOSFET 36 is maintained relatively constant and the lower MOSFET 36 is biased into the triode region, then the output current of the lower MOSFET 36 will be linearly proportional to the input signal. The following derivation illustrates that for a device in deep triode region:

$I_d=\mu C\frac{W}{L}\left[\left(V_{\mathrm{gs}}-V_{\mathrm{th}}\right)V_{\mathrm{ds}}-\frac{1}{2}{V}_{\mathrm{ds}}^2\right],g_{\mathrm{ds}}=\frac{\partial I_d}{\partial V_{\mathrm{ds}}}=\mu C\frac{W}{L}\left[\left(V_{\mathrm{gs}}-V_{\mathrm{th}}\right)-V_{\mathrm{ds}}\right]\approx \beta \left(V_{\mathrm{gs}}-V_t\right),\mathrm{where}\beta =\mu C\frac{W}{L},$ the output AC current is as follows:i_out=v_dsg_ds=β(V_gs−V₁)v_ds Which shows that i_out may be a linear function of the input signal, leading to an increase in linearity. The amount of linearity achieved may be controlled by adjusting the ratio of the upper MOSFET size to the lower MOSFET size. A load resistor 39 may be connected to the upper MOSFET 34. Another way of looking at it is to view the lower MOSFET 36 as a normal MOSFET which has its own transconductance g_m. The following derivation illustrates that the linearity of g_m may be dependent on Vds for a MOSFET operated in the triode region.

$I_d=\mu C\frac{W}{L}\left[\left(V_{\mathrm{gs}}-V_{\mathrm{th}}\right)V_{\mathrm{ds}}-\frac{1}{2}{V}_{\mathrm{ds}}^2\right],$ thus the transconductance of the device is,

$g_m=\mu C\frac{W}{L}{V}_{\mathrm{ds}}=\beta V_{\mathrm{ds}}$

The sensitivity of gm to variations in the input signal may be reduced by reducing the sensitivity of Vds to variations in the input signal, thereby increasing the linearity of the amplification.

However, since β is function of process and temperature variation, the gain of the amplifier may vary too. One way to reduce that sensitivity is to bias the input stage so that βVds is less sensitive to environmental variations.

FIG. 3 shows an aspect of a bias circuit 50 for a linear input stage. The bias circuit 50 may control the variation of the linear input stage transconductance to reduce sensitivity to process, environmental effects such as temperature, and power. The bias circuit 50 includes an upper MOSFET, M2, 52 connected to a lower MOSFET M1, 54. The upper MOSFET 52 is operated in the saturation region and the lower MOSFET is operated in the triode region. A third MOSFET, M3, 56 operates to bias the lower MOSFET 52 into the triode region. To set the bias to the lower MOSFET 52, the magnitude of the current, I3, flowing through M356 may be controlled as well as controlling the physical characteristics of M356 such as size. For example, if I3 is selected to equal I1 (the current flowing through M1), then a bias switch ratio defined as the ratio of the size of M154 to the size of M356 should be selected to be at least greater than one, and preferably greater than 1.4. A resistor 58 connected from the gate of M156 decouples the input signal from the bias circuit 50.

FIG. 4 shows an aspect of an amplifier 59 including a bias circuit 60 connected to a linear input stage 82. The bias circuit 60 is similar in function to bias circuit 50 with corresponding elements in the range of 62 to 68. The linear input stage 82 is similar in function to linear input stage 32 with corresponding elements in the range of 84 to 86. The amplifier 59 advantageously combines the benefits of both the linear input stage 82 and the bias circuit 60. An input signal may be AC coupled through a capacitor 60 to the gate of the lower MOSFET 86. A load resistor 88 may be connected to the upper MOSFET 84 to obtain an output from the drain of the upper MOSFET 84.

The following derivation may be used to select the devices for the linear input stage 82 and the bias circuit 60 of a preferred embodiment, and demonstrate how the gm of the input stage is controlled to be less sensitive to environmental variations. The linear input stage transconductance, g_mA may be as follows:g_mA=βV_ds,A=β(V_b−V_dsat,B−V_th,B)where V_b is the voltage from the gate of MB to ground.

For discussion purpose, let's assume

${\left(\frac{W}{L}\right)}_2={\left(\frac{W}{L}\right)}_B,{\left(\frac{W}{L}\right)}_1={\left(\frac{W}{L}\right)}_A,$ and I1=I2, then V_dsa,2=V_dsat,B and V_th,2=V_th,B, the transconductance of MA becomes; g_mA=β(V_b−V_dsat,2−V_th,2)=βV_ds,1. I1 is not limited to any specific ratio of I2 as long as the ratio of (W/L)₁ to (W/L)_A and (W/L)₂ to (W/L)_B are properly scaled so that the current densities are about the same for those devices. The ratio of the size of M2 to the size of M1 should be approximately equal to the ratio of the size of MB to the size of MA.

For the same reason, let's assume I3=I1, and

${\left(\frac{W}{L}\right)}_1=X·{\left(\frac{W}{L}\right)}_3,$ where X>1.0 and preferably 1.4. Then M1 is also in the triode region, and if M1 in deep triode region, Vgs−Vth>>Vds/2, then I₁≈β(V_gs,1−V_th,1)V_ds,1

$g_{mA}=\beta V_{\mathrm{ds},1}=\beta \frac{I_t}{\beta \left(V_{\mathrm{gs},1}-V_{\mathrm{th},1}\right)}=\frac{I_3}{\left(V_{\mathrm{gs},1}-V_{\mathrm{th},3}\right)}=g_{m,3}/2.$ If current I3 is a constant gm bias current which is;

$\mathrm{Ids}=\frac{A}{\beta },$ where A can be chosen to only depend on an external resistor value and ratio of two transistors[1], then,g_mA=g_m,3/2=√{square root over (2*I₃*β)}/2=√{square root over (A/2)} which is a constant.

Here, I3 does not have to equal I1, instead “X”, the ratio of the size of M3 to the size of M1, can be set to a predetermined value and the ratio of I3 to I1 varied. Also, the ratios

${\left(\frac{W}{L}\right)}_1\mathrm{and}{\left(\frac{W}{L}\right)}_3$ may be varied to bias M1 into the triode region.

FIG. 5 shows an aspect of an operation for generating a linear input stage. Starting at block 100, a semiconductor die is provided. At block 102, a first MOSFET having a predetermined size is formed. At block 104, a second MOSFET having a size greater than the first MOSFET is formed. At block 106, the second MOSFET is connected in cascode with the first MOSFET. At block 108, the first MOSFET is biased into the triode region. At block 110, the second MOSFET is biased into the saturation region. At block 112, an input signal is applied to the gate of the first MOSFET causing a change in I_d of the MOSFETs that is approximately a linear function of the AC voltage applied to the first MOSFET gate.

FIG. 6 shows an aspect of an operation for biasing a linear input stage. Starting at blocks 120 and 122, first and second MOS devices are provided. At block 124, the ratio of the first MOS device size to the second MOS device is selected to be a predetermined value, Rb. At block 126, the second MOS device is connected in cascode with the first MOS device. At block 128, the first MOS device is biased into the triode region. At block 130, the second MOS device is biased into the saturation region such as by connecting the gate and drain of the second MOS device together. At block 132, a linear input stage having “A” and “B” MOS devices is provided. At block 134, the ratio of the “A” MOS size to the “B” MOS size is selected to be about Rb. At block 136, the first and second MOS devices are connected to the linear input stage.

A number of embodiments of the invention have been described. Nevertheless, it will be understood that various modifications may be made without departing from the spirit and scope of the invention. Accordingly, other embodiments are within the scope of the following claims.

Claims

1. An analog signal processing circuit, comprising: a bias circuit;a first circuit including: a control input that communicates with said bias circuit;a first terminal that generates an output current; anda second terminal; anda device that communicates with said second terminal of said first circuit, that includes a variable resistor, and that has a resistance that is modulated in response to an input signal to said analog signal processing circuit.

2. The analog signal processing circuit of claim 1 wherein said bias circuit includes an amplifier.

3. The analog signal processing circuit of claim 1 wherein said first circuit includes a transistor.

4. A wireless transceiver comprising the analog signal processing circuit of claim 1.

5. The analog signal processing circuit of claim 1 wherein said resistance of said device changes as a function of said input signal.

6. The analog signal processing circuit of claim 5 wherein said resistance of said device is proportional to said input signal.

7. An analog signal processing circuit, comprising: bias means for biasing;current generating means including: a control input that communicates with said bias means;a first terminal that generates an output current; anda second terminal; andvariable resistance means for providing a variable resistance and that comprises a variable resistor, for communicating with said second terminal of said current generating means, and for providing a resistance that is modulated in response to an input signal to said analog signal processing circuit.

8. The analog signal processing circuit of claim 7 wherein said bias means includes an amplifier.

9. The analog signal processing circuit of claim 7 wherein said current generating means includes a transistor.

10. A wireless transceiver comprising the analog signal processing circuit of claim 7.

11. The analog signal processing circuit of claim 7 wherein said resistance of said variable resistance means changes as a function of said input signal.

12. The analog signal processing circuit of claim 11 wherein said resistance of said variable resistor means is proportional to said input signal.