616 IEEE MICROWAVE AND WIRELESS COMPONENTS LETTERS, VOL. 17, NO.

8, AUGUST 2007

A Compact Wideband CMOS Low-Noise Amplifier

Using Shunt Resistive-Feedback and Series
Inductive-Peaking Techniques
Hsien-Ku Chen, Da-Chiang Chang, Ying-Zong Juang, and Shey-Shi Lu, Senior Member, IEEE

Abstract—A wideband low-noise amplifier (LNA) with shunt re- (NF) degradation for on-chip implementation. An efficient way
sistive-feedback and series inductive-peaking is proposed for wide- to achieve a broadband matching with small power dissipation
band input matching, broadband power gain and flat noise figure and chip area is the common-gate input topology. However, it
(NF) response. The proposed wideband LNA is implemented in
0.18- m CMOS technology. Measured results show that power
has been reported that a common-gate LNA exhibits a lower
gain is greater than 10 dB and input return loss is below 10 dB gain and higher NF than a common-source amplifier [4].
from 2 to 11.5 GHz. The IIP3 is about +3 dBm, and the NF ranges In this letter, a wideband LNA with wide bandwidth, low
from 3.1 to 4.1 dB over the band of interest. An excellent agree- NF, low power consumption, broadband impedance matching
ment between the simulated and measured results is found and at- and small chip-area is reported. Both the shunt resistive-feed-
tributed to less number of passive components needed in this cir- back and series inductive-peaking techniques are adopted to
cuit compared with previous designs. Besides, the ratio of figure-of-
merit to chip size is as high as 190 (mW 1 mm2 ) which is the best
design the 2 11.5 GHz UWB LNA. These techniques have
results among all previous reported CMOS-based wideband LNA. the merits of achieving wideband frequency response, sufficient
power gain and small NF without using many passive compo-
Index Terms—CMOS, feedback, low cost, low-noise amplifier
(LNA), series peaking, small chip size, wideband.
nents that require large die area.

I. INTRODUCTION II. WIDEBAND LNA DESIGN

The conventional cascode-type LNA with inductive source-
W IDEBAND systems have recently gained much attention
due to their capability of high data rate transmission. The
so called ultrawideband (UWB) technology provides attractive
degeneration is one of the most popular LNA topologies due
to its merits of low power consumption, high gain, and high
high speed wireless signaling schemes, which use the frequency reverse isolation. For wideband application, this topology be-
bands in 3.1 10.6 GHz. To interface with the antenna and comes unsuitable because the bandwidth of the input series res-
pre-select filter, the low-noise amplifier (LNA) input impedance onance circuit is limited by its quality factor. Hence, a 3–5 GHz
should be close to 50 across the band from 3.1 to 10.6 GHz. cascode-type UWB LNA with resistive feedback [5] has been
Up to now, many wideband techniques have been proposed proposed to reduce the quality factor of the input series reso-
for the UWB LNA. The distributed amplifier (DA) is widely nance circuit. But the bandwidth of that design is still limited
used for wideband application due to its intrinsic broadband by the quality factor of the input series resonance circuit and
frequency response going all the way down to dc along with capacitive effect, namely, , , , and of M1 and
good input and output impedance matching. Yet, so far, high M2, and thus the input return loss and gain began to roll-off at
power consumption and large die area have hampered its wide- 4.2 GHz. Fig. 1 shows the proposed UWB LNA, which con-
spread applications [1], [2]. Recently, a cascode CMOS LNA tains the conventional cascode-type LNA enclosed in the dash
with a bandpass response at the input for wideband impedance lines of the figure. A smaller feedback resistor R2 (300 ) com-
matching has been reported in [3]. The bandpass-filter (BPF) pared to that (1 k ) of the conventional design [5] is used to
based topology incorporates the input impedance of the cas- reduce the quality factor of the input series resonance circuit
code amplifier as a part of the filter. However, the adoption of further so that a wider input matching can be achieved. How-
the filter at the input requires a number of reactive elements, ever, from the simulation, the feedback resistor R2 degrades the
which inevitably results in a larger chip area and noise figure NF of LNA by approximately 1.2 dB. Moreover, the power gain
of this amplifier at higher frequencies is also degraded due to the
capacitive effect of large size M1 and M2 chosen. To counter-
Manuscript received January 15, 2007; revised April 23, 2007. This work balance this gain degradation, a series-peaking inductor L2 as
was supported by the National Science Council of Taiwan, R.O.C., under Grant shown in Fig. 1 is added to make the gain response flatter. The
NSC-96-2218E002-042. series-peaking inductor L2 can also resonate with the parasitic
H.-K. Chen and S.-S. Lu are with the Graduate Institute of Electronics
Engineering, National Taiwan University, Taipei 100, Taiwan R.O.C. (e-mail: shunt capacitor Cp at the drain of M2. The value of L2 is chosen
[email protected]). to be 1.13 nH for peaking at the frequency of 10 GHz, and C2
D.-C. Chang and Y.-Z. Juang are with the Chip Implementation Center, Na- is used for dc decoupling. The degeneration inductor TL1 was
tional Applied Research Laboratories, Hsinchu 300, Taiwan, R.O.C.
Color versions of one or more of the figures in this letter are available online
chosen to be small (0.12 nH) for matching at higher band and
at http://ieeexplore.ieee.org. implemented by a microstrip (MS) line with a line width of 6 m
Digital Object Identifier 10.1109/LMWC.2007.901797 and a line-length of 162 m. Compared to the spiral inductor,
1531-1309/$25.00 © 2007 IEEE
CHEN et al.: COMPACT WIDEBAND CMOS LNA 617

Fig. 3. Simulated and measured results for input return loss and power gain of
wideband LNA versus frequency.

Fig. 1. Wideband LNA schematic.

Fig. 4. Simulated and measured results for NF of wideband LNA versus fre-
quency.

because the LNA is designed to be integrated with a mixer be-

hind it and the intentional peaking at 10 GHz can pre-compen-
Fig. 2. Die photograph of the wideband LNA.
sate the usual overdamped response of the next stage. Besides,
it can also be seen that the power gain begins to roll-off from
3 GHz if L2 is removed from circuit, which evidences that the
the MS inductor is less susceptible to process variation when the series inductive-peaking technique is useful for extending the
inductance is very small. bandwidth of the resistive feedback cascode-LNA.
Fig. 4 shows the simulated and measured results of the NF
III. EXPERIMENTAL RESULTS of this wideband LNA. Based on Fig. 4, the measured min-
imum and maximum NFs are 3.1 dB (at 9 GHz) and 4.1 dB
The wideband LNA was implemented in a 1P6M 0.18 m (at 11.5 GHz), respectively, over the band of 2–11.5 GHz. From
CMOS technology. The die photograph is shown in Fig. 2. The the simulation it is clear that the shunt feedback resistor R2 for
die area including testing pads is only 0.48 0.69 mm , which the purpose of wideband matching degrades the NF of LNA by
is very area-efficient and consequently cost effective due to the 1.2 dB. Nevertheless, the achieved NFs are still better than the
techniques adopted. The circuit was tested on wafer for RF char- previously measured values of 3.1 10.6 GHz CMOS LNAs
acterization. The LNA draws 11.2 mA from bias-T and the drain [1]–[9]. In addition, this circuit also has the flattest NF across the
voltages of M2 is 1.2 V. band of all UWB LNAs, which is explained as follows. Based
Fig. 3 shows the simulated and measured results of the input on [10] and neglecting gate noise, we can show that the NF of an
return loss and power gain of the wideband LNA. The power LNA using a source degenerative inductor can be approximated
gain is above 10 dB with a 3-dB frequency band up to 12.5 GHz by
while the return losses are below 10 dB from 2 to 11.5 GHz.
Simulation shows that the bandwidth of power matching shrinks
NF (1)
from 9.5 GHz to only 2.5 GHz if R2 is removed from the cir-
cuit. These results unequivocally prove the effectiveness of the
broadband matching realized by using the small R2 shunt resis- where and
tive-feedback technique in conjunction with the source degener- are the series parasitic resistance, and quality
ative inductior-TL1. Fig. 3 also shows the simulated frequency factor of the input – – circuit, respectively,
responses of the LNA with different values of L2. It is found 1 , 50 and the channel
that L2 0.619 nH corresponds to the maximally flat condi- noise parameter. (1) is verified by comparison with simulation
tion. In our work, L2 is chosen to be 1.13 nH for peaking at results (see Fig. 4). According to (1), the frequency responses
10 GHz. An L2 smaller than 1.13 nH (but lager than 0.619 nH) (under-, over-, and critical-damping) of NF are controlled
can result in a flatter gain response (and even a wider band- by and in this work a as close to the maximally flat
width). However, in this work, a larger L2 value corresponding condition as possible is chosen, which explains the excellent
to under-damped rather than maximally flat condition is chosen NF flatness of our LNA.
618 IEEE MICROWAVE AND WIRELESS COMPONENTS LETTERS, VOL. 17, NO. 8, AUGUST 2007

TABLE I
RECENTLY REPORTED PERFORMANCES OF WIDEBAND LNAs

[4] is the simulated results

IV. CONCLUSION
A compact wideband LNA for 2 11.5 GHz application
has been proposed. The shunt resistive-feedback in conjunc-
tion with series inductive-peaking provides a method for
achieving a high performance wideband LNA with low cost.
Because less component are used with this method, the pro-
posed wideband LNA has shown a good agreement between
simulation and measurement results. Besides, this wideband
LNA reveals the best FoM of 62.8 (mW ) and moreover the
Fig. 5. Measured results for IIP3 under the two tones input of 5.2 and 5.8 GHz. highest ratio (190 mW mm ) of FoM to chip-size among all
CMOS-based wideband LNAs.

REFERENCES
From Figs. 3 and 4, it is found that the simulation and mea-
[1] R. Liu, C. Lin, K. Deng, and H. Wang, “A 0.5–14-GHz 10.6-dB CMOS
surement results agree very well, which is attributed to careful cascode distributed amplifier,” in VLSI Circuits Symp. Dig., Jun. 2003,
layout and less number of passive components needed in the vol. 17, pp. 139–140.
proposed LNA compared with previous designs. [2] F. Zhang and P. R. Kinget, “Low-power programmable gain CMOS
distributed LNA,” IEEE J. Solid-State Circuits, vol. 41, no. 6, pp.
To characterize the nonlinear behavior, two-tone signals with 1333–1343, Jun. 2006.
equal power levels at 5.2 and 5.8 GHz, which are commonly [3] A. Bevilacqua and A. M. Niknejad, “An ultrawideband CMOS low-
used frequencies in wireless local area network applications, noise amplifier for 3.1–10.6-GHz wireless receivers,” IEEE J. Solid-
State Circuits, vol. 39, no. 12, pp. 2259–2268, Dec. 2004.
were applied to the proposed UWB LNA. The measured results [4] Y. Lu, K. S. Yeo, A. Cabuk, and J. Ma, “A novel CMOS low-noise am-
in Fig. 5 indicate that this LNA has an of 3 dBm and a plifier design for 3.1- to 10.6-GHz ultra-wide-band wireless receivers,”
of 8.5 dBm. These results demonstrate that the proposed IEEE Trans. Circuits Syst. I, vol. 53, no. 8, pp. 1683–1692, Aug. 2006.
wideband LNA achieves good linearity, too. [5] 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
To evaluate the performance of UWB LNAs for their max- system,” IEEE J. Solid-State Circuits, vol. 40, no. 2, pp. 544–547,
imum power gain, 3 dB-bandwidth, excess NF, power dissipa- Feb. 2005.
tion, and cut-off frequency (representing technology cost), the [6] H. J. Lee, D. S. Ha, and S. S. Choi, “A 3 to 5 GHz CMOS UWB LNA
with input matching using Miller effect,” in Proc. IEEE Solid-State
figure of merit (FoM) is defined as [11] Circuits Conf., Feb. 2006, pp. 731–740.
[7] C. F. Liao and S. I. Liu, “A broadband noise-canceling CMOS LNA
for 3.1–10.6-GHz UWB receiver,” in Proc. IEEE CICC’05, Sep. 2005,
FoM pp. 161–164.
[8] A. Bevilacqu, C. Sandner, A. Gerosa, and A. Neviani, “A fully inte-
grated differential CMOS LNA for 3–5-GHz ultrawideband wireless
This FoM includes the most relevant parameters in order to receivers,” IEEE Microw. Wireless Compon. Lett., vol. 16, no. 3, pp.
evaluate a UWB LNA for low-cost and low-power applications. 134–136, Mar. 2006.
A summary of the state-of-the-art results is given in Table I. 
[9] C. T. Fu and C. N. Kuo, “3 11-GHz CMOS UWB LNA using dual
feedback for broadband matching,” in Proc. IEEE Rad. Freq. Integr.
Our proposed LNA achieves the highest FoM of 62.8 (mW ) Circuits (RFIC) Symp., Jun. 2006, pp. 11–13.
among all CMOS-based wideband LNAs. Besides, the chip [10] H. W. Chiu, S. S. Lu, and Y. S. Lin, “A 2.17 dB NF, 5 GHz band
size of our LNA including testing pads is only 0.33 mm , monolithic CMOS LNA with 10 mW dc power consumption,” IEEE
and thus the ratio of FoM to chip size can be even as high as Trans. Microw. Theory Tech., vol. 53, no. 3, pp. 813–824, Mar. 2005.
[11] D. Barras, F. Ellinger, H. Jäckel, and W. Hirt, “A low supply voltage
190 (mW mm ), which is at least a factor of 2 better than SiGe LNA for ultra-wideband frontends,” IEEE Microw. Wireless
any of the previously published results. Compon. Lett., vol. 14, no. 10, pp. 69–71, Oct. 2004.