This file type includes high resolution graphics and schematics when applicable.

Since the United States’ Federal Communications Commission (FCC) earmarked the spectrum from 3.1 to 10.6 GHz for ultrawideband (UWB) communications use, interest has increased in the development of components for this broad bandwidth. The frequency band holds great promise for high-data-rate wireless communications and is geared to co-exist with licensed wireless applications within the more narrowband segments in that frequency range. Of course, producing high-frequency components capable of operating within the UWB frequency range—such as low-noise amplifiers (LNAs)—is critical to the use of practical UWB applications.

**Related Articles**

• Compact LNA Drives 2.5-GHz Base Stations

• LNA Boosts 6 To 20 GHz

• Variable-Gain LNA Reaches 26 GHz

Unfortunately, the LNA is one of the greater challenges in terms of UWB components. But with the proper circuit design and semiconductor process, such a broadband low-noise amplifier can be produced with high gain and low noise figure.

Of course, minimizing the power consumption of an LNA for UWB applications is another important requirement. This is no trivial design task, even with the benefit of excellent modern semiconductor processes. UWB LNAs for frequency bands of 3.0 to 5.0 GHz and 3.1 to 10.6 GHz have been widely reported based on silicon CMOS technology.^{1-6} Another UWB LNA was reported for the frequency band of 2 to 6 GHz, with current reuse topology utilizing a noise-shaping technique.^{7}

The current report highlights a broadband CMOS LNA for UWB applications from 3 to 7 GHz. It employs current-reuse techniques to minimize power consumption and enhance gain at the upper end of the frequency range. A shunt peaking inductor was used to improve flatness gain and to increase the total bandwidth, while a common-drain amplifier helps achieve a wideband output impedance match for measurement purposes.

The proposed LNA is a two-stage amplifier with a current-reuse cascade common source structure, useful both for saving power consumption and maintaining good power gain. The two-stage design, initially targeted for power consumption of 6 mW from a +1-VDC supply, is shown in **Fig. 1**. This results in drain current of approximately 6 mA. Assuming a 6 mA current drawn by transistor M_{1} for the first stage, the calculated size for transistor M_{1} is approximately 150 μm based on Eq. 1^{8}:

I_{D} ≈ 0.5μ_{n}C_{ox}(W/L)(V_{GS1} – V_{t})^{2 }(1)

where:

V_{GS1} = +0.7 VDC;

V_{t} = +0.5 VDC; and

μ_{n}C_{ox} = 369 μm for a typical 0.18-μm silicon CMOS semiconductor process.

The required transconductance, g_{m1}, can be determined by Eq. 2:

g_{m1 }= ∂I_{DD}/∂I_{GS1 }= μ_{n}C_{ox}(W/L) (V_{GS1} – V_{t}) (2)

Rearranging Eq. 1 to:

(V_{GS1} – V_{t} = [2I_{DD}/μ_{n}C_{ox}(W/L)]^{0.5}

and substituting into Eq. 2, the equation for g_{m1} can be simplified to Eq. 3:

g_{m1 }= [2μ_{n}C_{ox}(W/L)I_{D}]^{0.5 }= (2βI_{D})^{0.5} (3)

where β = μ_{n}C_{ox}(W/L) is known as the transconductance parameter. Hence, the calculated g_{m1} is approximately 60.7 mA/V.

By ignoring the Miller effect of gate-drain capacitance (C_{gd1}) on transistor M_{1}, the input impedance of M_{1 }can be given by Eq. 4^{9}:

Z_{in1 }= jω(L_{g1} + L_{s1}) + (1/jωC_{gs1 }+ g_{m1}L_{s1}/C_{gs1 }(4)

where:

g_{m1} = the transconductance of M_{1};

C_{gs1} = the gate-source capacitance of M_{1};

L_{s1} = the source degeneration inductor; and

L_{g1} = the gate input inductor.

The real part of the input impedance in Eq. 4 is given by Eq. 5:

ReZ_{in1 }= g_{m1}L_{s1}/C_{gs1 }(5)

With given values of g_{m1} and C_{gs1}, the desired impedance match to R_{s} (usually 50 Ω) can be obtained by setting L_{s1} accordingly. The imaginary part of the input impedance can then be compensated with an input matching inductance, L_{g1}. The corresponding resonant frequency, ω_{0}, is approximated by Eq. 6:

ω_{0} ≈ [1/(L_{s1} + L_{g1})C_{gs1}]^{0.5 }(6)

The source generation inductor, L_{s1}, is added for linearity and stability improvement, whereas the gate inductor L_{g1} is needed for impedance matching between the source impedance and input of transistor M_{1}. In this design, the value of inductor L_{s1} is chosen to be small enough (approximately 0.5 nH) so that it could be easily replaced by a bondwire inductance in order to reduce the chip area when necessary. Next, the value of gate inductor L_{g1} is calculated from Eq. 6 to be approximately 1.5 nH using a resonant frequency of 5 GHz.

The current-reuse topology employs transistor M_{2} stacked on top of the input transistor M_{1} to save power consumption. The current-reuse function is implemented by L_{d1}, L_{g2} and C_{g2} to boost gain. Inductor L_{g2} must form a series-resonant circuit with capacitance C_{g2} and the input capacitance C_{gs2} of transistor M_{2} to create a low-impedance path to couple the output of M_{1} to the input of M_{2}, while inductor L_{d1} is chosen to be large enough to provide a high-impedance path to block the signal in the desired band.^{10} Parameter C_{bp} is a bypass capacitor. Therefore, the input can be amplified twice under this concurrent scheme.

It should be noted that the series resonant circuit consisting of L_{g2} and C_{g2} exhibits a narrowband characteristic. Consequently, the current-reuse function is maximized around the resonant frequency, which is designed to enhance the gain at the upper end of the desired frequency band.

This file type includes high resolution graphics and schematics when applicable.

## Reducing Power Consumption

This file type includes high resolution graphics and schematics when applicable.

To achieve low power consumption, the width of transistor M_{1} is set to 260 μm, while the width of M_{2} is set to 200 μm. The inductances of L_{d1} and L_{g2} are 5.3 and 1.8 nH, respectively. Resistor R_{b2} is used to provide bias voltage for transistor M_{2}. The inductor L_{d2} (2.3 nH) and resistor R_{d} (54 Ω) are chosen to provide inductive peaking for extending the frequency range and providing peak gain at the center of the passband, achieving nearly flat overall wideband LNA gain. The last stage of the LNA consists of transistor M_{3} and the current source is the 50-Ω buffer for measurement purposes. Capacitors C_{in} and C_{out} are DC blocks.

The current-reuse amplifier functions as a two-stage cascade amplifier. The noise figure of the second stage contributed by both the MOSFET (device M_{2}) and inductors L_{g2} and L_{d1} can be reduced by the gain of the first stage. With this design technique, not only is high gain possible, but low noise figure can be achieved simultaneously.

**Figures 2 through 6** offer simulation results for the proposed UWB LNA for simulations performed with the SpectreRF simulation program from Cadence Design Systems, utilizing a 0.18-μm silicon CMOS semiconductor process from Taiwan Semiconductor Manufacturing Co. (TSMC). **Figure 2** shows the small-signal S-parameters, where it can be seen that the UWB LNA achieves average simulated gain of 10.6 dB from 3 to 7 GHz, with gain flatness of better than 1.5 dB. The maximum gain of 12.3 dB occurs at 5 GHz.

Over the frequency range of interest, the input return loss (S_{11}) is less than 7.7 dB, while the output return loss (S_{22}) is less than 13 dB. The reverse isolation is less than -36 dB from 1 to 10 GHz, as shown in **Fig. 3**. The stability factor (K-factor) was also computed using S-parameter data and was found to be greater than 1 across the frequency range from 1 to 8 GHz **(Fig. 4)**. This indicates that the proposed LNA is unconditionally stable across the frequency range of interest.

As **Fig. 5** shows, the noise figure is better than 3.3 dB between 3 and 7 GHz with minimum noise figure of 2.6 dB at 4.3 GHz. The output power versus input power was measured for a single test tone at 5 GHz **(Fig. 6)**, with a 1-dB compression point of about -10 dBm. The power consumption is only about 7 mW from a +1-VDC supply, which includes the power of the output buffer.

The **table** summarizes the measured performance of the UWB LNA compared to previous works. The layout is shown in **Fig. 7**. The amplifier can achieve a wide bandwidth, high gain, and low noise figure while also operating with minimal power consumption—about 7 mW across a frequency range of 3 to 7 GHz.

This file type includes high resolution graphics and schematics when applicable.

## References

This file type includes high resolution graphics and schematics when applicable.

1. A. Bevilacqua and A.M. Niknejad, “An Ultra wide band CMOS Low-Noise Amplifier for 3.1-10.6 GHz Wireless Receivers,” IEEE Journal of Solid-State Circuits, Vol. 39, No. 12, 2004, pp. 2259-2268.

2. C.W. Kim, M. S. Kang, P.T. Anh, H.-T. Kim, and S.G. Lee, “An Ultra-Wideband CMOS Low Noise Amplifier for 3-5 GHz UWB System,” IEEE Journal of Solid-State Circuits, Vol. 40, No. 2, 2005, pp. 544-547.

3. Y. Lin and S.H. Hsu, “A 3.1-10.6 GHz Ultra-Wideband CMOS Low Noise Amplifier with current reused technique,” IEEE Microwave and Wireless Components Letters, Vol. 17, No. 3, 2007, pp. 232-234.

4. H.L. Kao, A. Chin, K.C.Chang, and S.P. McAlister, “A low-Power Current-Reuse LNA for Ultra-Wideband Wireless Receivers from 3.1 to 10.6 GHz,” 2007 Topical Meeting on Silicon Monolithic Integrated Circuits in RF Systems, 2007, pp. 257-260

5. Z.Y. Huang, C.C. Huang, Y.T. Hung, and M.P. Chen, “A CMOS Current Reused Low-Noise Amplifier for Ultra-Wideband Wireless Receiver,” ICMMT, 2008, pp. 1499-1502.

**Related Articles**

• Compact LNA Drives 2.5-GHz Base Stations

• LNA Boosts 6 To 20 GHz

• Variable-Gain LNA Reaches 26 GHz

6. K.T. Ansari and C. Plett, “A CMOS Low Power Ultra-Wideband LNA Utilizing Feedback Technique,” Microsystems and Nanoelectronics Research Conference, 2009, pp. 5-8.

7. W.H. Hung, K.T. Lin, J.Y. Hsieh, and S.S. Lu, “A 2-6 GHz Broadband CMOS Low-Noise Amplifier with Current Reuse Topology Utilizing a Noise-Shaping Technique,” ISCAS, 2011, pp. 1291-1294.

8. B. Razavi, Design of Analog CMOS Integrated Circuits, McGraw-Hill, New York, 2001.

9. T.H. Lee, The Design of CMOS Radio-Frequency Integrated Circuits, 2nd ed., Cambridge University Press, New York, 2004.

10. S.A.Z. Murad, R.K. Pokharel, R. Sapawi, H. Kanaya, and K. Yoshida, “High Efficiency, Good Linearity, and Excellent Phase Linearity of 3.1-4.8 GHz CMOS UWB PA with a Current-Reused Technique,” IEEE Transactions on Consumer Electronics, Vol. 56, No. 3, 2010, pp. 1241-1246.

This file type includes high resolution graphics and schematics when applicable.