SINCE the approval of the ultra-wideband (UWB) radio technology for low power wireless communication application in February, 2002, [1] UWB systems has become an increasingly popular technology which is capable of transmitting data over a wide spectrum of frequency with very low power and high data rate. The IEEE UWB standard (IEEE 802.15.3a [2]) has not been completely defined, two major proposed solutions, MB-OFDM and DS-UWB, are all allowed to transmit in a band between 3.1 and 10.6 GHz. The band definition of MB-OFDM is illustrated in Fig. 1 which extended form 3168 MHz to 10560 MHz and the band definition of DS-UWB from 3100 MHz to 4900 MHz and from 6000 MHz to 9600 MHz The bandwidth of MB-OFDM containing 14 bands; there are two bands of DS-UWB which are Low-Band and High-Band.

All receiver architectures of UWB is usually used a cascade system. The overall noise figure of a cascade of systems depends on both the individual noise figures as well as their gains. The dependency on the gain results from the fact that, once the signal has been amplified, the noise of subsequent stages is less important. To develop an equation for the system noise figure, consider the block of Fig. 2, [2], where each *F*_{n} is a noise factor and each *G*_{n} is power gain. The total noise factor is the sum of individual constructions, and is there for given by
TeX Source
$$F = F_1 + {F_2-1 \over G_1}+{F_3-1 \over G_1G_2}+\cdots + {F_N-1\over \prod^{N-1}_{n=1}G_n}\eqno{\hbox{(1)}}$$As a result, system noise figure tend to be dominated by the noise performance of the first couple of stage in a receiver. The LNA is the first stage of a receiver typically. Therefore, its noise figure is critical since it directly adds to the total noise figure of whole system. This paper is focused on the design and implementation of high gain and wideband low-noise amplifier for ultra-wideband receiver in a 0.18 um Standard RF CMOS Process.

SECTION II

## Design of Ultra-Wideband LNA

### A. Ultra-Wideband Amplifier Design

The proposed low-noise amplifier is shown in Fig. 3 which consists of wideband input impedance matching networks, cascode amplifier with shunt-peaking load, inductor-coupling resonated feedback and output buffer. The constituent of wideband matching networks are inductors L_{1},L_{g},L_{s} and the impedance of R_{F}L_{F}C_{F} feedback, capacitor C1 and transistor M_{1}. Cascoded amplifier is transistors M_{1} and M_{2} with shunt-peaking load consists of inductor L_{L} and resistor R_{L}; and the impedance of R_{F}L_{F}C_{F} feedback is also a part of the load of the amplifier. A common-drain amplifier is a good choice of wideband output impedance matching for measurement purpose.

The feedback loop Z_{f} is due to the amplifier A, thus the equivalent circuit could be modified into Z_{1} and Z_{2} by the Miller's Theory which is shown in Fig. 4. The input and output equivalent circuit is modified into Figs. 5 and 6.

One feedback loop Z_{F} could play two roles in this low-noise amplifier, the input impedance Z_{1} is becoming a part of input matching network; the output impedance Z_{2} is becoming a part of the load of the cascode amplifier. Therefore, the shunt-peaking load and Z_{2} coupled by the inductor L_{cp}, the inductor-coupling resonated load is established.

### B. Input Impedance Matching Networks

A small signal hybrid- π model of input matching networks is shown in Fig. 7. The source degeneration input matching will resonate a 50 Ohm input impedance in the resonance frequency f_{r1} in equation (3), and the equivalent impedance is stated in Eqn. (2). The impedance of Z_{1} will contribute a resonated frequency f_{r2} which is expressed in equation (4). And the serial L_{1} and C_{1} will also provide a resonated frequency f_{r3} where is shown in equation (5). Therefore, the input matching networks are resonated in three resonated frequency in 50 Ohm system, and the input wideband matching network is finished as the Chebyshev filter [3].
TeX Source
$$\eqalignno{Z_{in} &= sL_g +{1\over sC_{gs}}+{g_m\over C_{gs}}L_S \approx sL_g +{1\over sC_{gs}}+\omega_TL_S&\hbox{(2)}\cr f_{r1} &= {1\over 2\pi \sqrt{L_S}C_{gs}}&\hbox{(3)}\cr f_{r2} &= {1\over 2\pi \sqrt{L_F}C_{F}}&\hbox{(4)}\cr f_{r3} &= {1\over 2\pi \sqrt{L_1}C_{1}}&\hbox{(5)}}$$

### C. Inductor-Coupling Resonated Low-Noise Amplifier

The low-noise amplifier is using the cascode amplifier with eliminating the Miller's Effect on input transistor to achieve high-frequency performance and have good isolation from mixer stage to antenna; the equivalent circuit is illustrated in Fig. 8. And the equivalent load Z_{out} is shown in Fig. 9, where the output impedance is expressed as Eqn. (6).
TeX Source
$$\eqalignno{Z_{L} &= sL_L +R_L\cr Z_F &= sL_F +{1\over sC_F}+R_F\cr Z_{out} &= {Z_L\cdot A_{\nu} Z_F\over Z_L(A_{\nu}+1)+A_{\nu}Z_F}\cr&= {s^3A_{\nu}(L_{cp}+L_L)L_FC_F + s^2 A_{\nu}[(L_{cp}+L_L)C_F R_F]s A_{\nu}[(L_{cp}+L_L)C_F R_L R_F C_F R_L R_F]+A_{\nu}R_L \over s^2[(A_{\nu}+1)(L_{cp}+L_L)C_F+A_{\nu}L_FC_F]+s[C_FR_L+A_{\nu}C_FR_F]+A_{\nu}}&\hbox{(6)}}$$

The inductor-coupling resonated load retrenches the demand of second stage amplifier that also scants quiescent current and power consumption which produced by transistors of second stage. Power gain is illustrated in Fig. 11; the gain is governed by center frequency f_{L1} and f_{L2}. The value of coupling-inductor should be selected carefully because the inductor is in the character of low pass filter, and the transition frequency f_{12} must between two center frequencies. Frequency f_{12} is expressed in Eqn. (7), L_{cp} is the value of inductor and R_{L} is the inner resistance.
TeX Source
$$F_{12}={2\pi r_L \over L_{cp}}\eqno{\hbox{(7)}}$$

If the value of inductor L_{cp} is too large and the transition frequency is below f_{L1} shown in Fig. 12(a) gain will not be passed before transition frequency f_{12} which is called over coupled shown in Fig. 12(b). The power gain only produced by second stage of resonated load which is the center frequency f_{L2}.

On the contrary, the value of inductorr L_{cp} is too small and the transition frequency surmount f_{L2} shown in Fig. 13(a) the gain will not be flat enough which is called under coupled shown in Fig. 13(b). The power gain decreasing around center frequency f_{L1}.

The transfer function of inductor-coupling resonated load is H(f) expressed in Eqn. (8) which consists of H_{1}(f),H_{2}(f) and H_{3}(f);H_{1}(f)H_{2}(f) are the transfer function of first and second stage of resonated load, H_{3}(f) is the transfer function of the low pass filter, L_{cp}.
TeX Source
$$\eqalignno{H(f) &= H_1(f)\cdot H_2(f)\cdot H_3(f)\cr&=(sL_L + R_L){1\over 1-A_V}{s^2 L_FC_F +sC_FR_F + 1 \over sC_F}{2\pi R_L \over L_{cp}}&\hbox{(8)}}$$

SECTION III

## Measurement Results

The measurement results of the proposed UWB LNA using Agilent Vector Network Analyzer (VNA) 8510C and Noise Figure Analyzer (NFA) N8975A are given in Fig. 14 to 19. In Fig. 14/15 that can be seen that the input/output return loss (S11/S22) are lower than −5.7 dB/−13.7 dB between 3.1 GHz to 10.6 GHz, respectively. The power gain whose peak value is 10.8 dB at 1.2 V which covers the complete band definitions of MB-OFDM and DS-UWB; the bandwidth is shown in Fig. 16. In Fig. 17, it can be seen that the noise figure is below 5.8 dB between 3.1 GHz to 10.6 GHz and the minimum noise figure is 3.9 dB at 5.0 GHz and which at 1.2 V supply voltage. The noise figure is still sufficient because the noise figure requirement is relaxed in UWB radio systems. [9] In Fig. 18, the input-referred 1 dB compression point (IP1 dB) is −17 dBm at 3.1 GHz, −14 dBm at 6.1 GHz and −16 dBm at 10.1 GHz. The IIP3 is −5 dBm at 7128 MHz and 7138 MHz in two tone tests, which is shown in Fig. 19. The power consumption is 6.2 mW operating at 1.2 V supply voltage which including the output buffer.

Table II is the performance at 1.2 V supply voltage. The chip photo of the proposed UWB LNA is presented in Fig. 20. The die area including the pads is 1.2 mm × 1.1 mm. The on-wafer measurements using Ground-Signal-Ground (GSG) and DC probes have been considered during the testing phase.

A CMOS UWB LNA is designed with inductor-coupling resonated load to extend the bandwidth. The results show that the proposed LNA gives 10.8 dB gain and 9.4 GHz 3 dB bandwidth (1.2 GHz–10.6 GHz) while consuming 6.2 mW through a 1.2 V supply. Over the 3.1—10.6 GHz frequency band, a minimum noise figure of 3.9 dB and input return loss lower than −5.7 dB have been achieved. The circuit is implemented in TSMC 0.18 um standard RF CMOS process.

### Acknowledgment

This work was supported by Industrial Technology Research Institute under Project 8301XS2810. The authors would like to thank the chip implementation center (CIC) for chip fabrication and technical support.