This article has been accepted for inclusion in a future issue of this journal.

Content is final as presented, with the exception of pagination.

IEEE JOURNAL OF SOLID-STATE CIRCUITS 1

A 20-GHz 1.9-mW LNA Using gm -Boost and

Current-Reuse Techniques in 65-nm CMOS for
Satellite Communications
Jiajun Zhang , Student Member, IEEE, Dixian Zhao , Member, IEEE, and Xiaohu You , Fellow, IEEE

Abstract— A 20-GHz low-power low-noise amplifier (LNA)

in 65-nm CMOS is presented. The LNA is cascaded with
a single-ended gm -boosted common-gate (CG) stage and a
differential neutralized common-source (CS) stage. Current-
reuse technique is employed to save the power consumption
with little deterioration in gain and noise figure (NF). The
transformer-based gm -boost technique in the CG stage and
neutralization technique in CS stage further enhances the
RF performances. Inter-stage magnetically coupled resonator
(MCR) extends the bandwidth. An elaborate analysis of the
current-reused CG–CS LNA using a transformer-based gm -boost
technique and transformer-based MCR is proposed. Fabricated
in 65-nm CMOS technology, the LNA achieves a measured
power gain of 14.9 dB at 21 GHz with a −3-dB bandwidth of
4.8 GHz. The lowest NF is 3.3 dB at 19.5 GHz and is below
4 dB from 17 to 21 GHz. The LNA consumes 1.9 mW from a
1-V supply, with a chip area of 600 μm × 700 μm.
Index Terms— CMOS, current-reuse, gm -boost, K -band, low- Fig. 1. Diagram of the CMOS phased-array beamforming receiver.
noise amplifier (LNA), low-power, millimeter-wave, satellite com-
munication, transformer. high gain and low NF to guarantee the communication quality,
the power consumption of the LNA in a large-scale phased
array is also of significant importance. Hence, low-power
I. I NTRODUCTION
LNA is essential in millimeter-wave (mm-Wave) phased-array

T HE next-generation high-throughput satellites (HTS) and

low Earth orbit (LEO) satellites open new opportunities
for high-quality satellite services with reasonable cost [1]. The
satellite terminals.
Among LNA topologies, common source (CS) with source
degeneration is the most widely used one because it can
K -band (17.7–20.2 GHz) and Ka-band (27.5–30 GHz) are achieve simultaneous noise and input matchings [4]–[6]. How-
allocated as downlink and uplink spectra, respectively, for high ever, source-degenerated LNA does not allow the circuit
data rate [2], [3]. The requirements of electrical scannability, to work at a low-power level. To cope with that, power-
far-distance coverage, and low cost urge the applications of a constrained LNA topology is introduced in [6]. It lowers
large-scale phased-array antenna (i.e., more than 1000 antenna the strict requirements of transistor size and degeneration
elements in total for HTS ground terminals) based on low-cost inductance by adding a capacitor between the gate and the
silicon ICs. As shown in Fig. 1, the low-noise amplifier (LNA) source of the CS stage. However, both techniques are initially
is the first module after antenna in every channel of the phased- proposed for narrowband communications [6]. A high-order
array receiver. The noise figure (NF) of the LNA dominates input matching network must be utilized to broaden the
the overall NF of the RF front ends. In addition to the required bandwidth, at the cost of relatively high insertion loss and
large chip area [7]–[11]. In contrast, the common-gate (CG)
Manuscript received January 30, 2020; revised April 9, 2020; accepted stage has an input impedance of about 1/gm , which makes
May 4, 2020. This article was approved by Guest Editor Po-Hung Chen.
This work was supported in part by the National Key Research and it attractive despite relatively low gain and high NF [12],
Development Program of China under Grant 2019YFB1803000 and Grant [13]. To solve the problem, several techniques are proposed
2018YFB1802100, in part by the National Nature Science Foundation of to enhance the RF performances of the CG stage. Using the
China under Grant 61674035, and in part by the Fundamental Research Fund
for the Central Universities. This article is an expanded version from the IEEE gm -boost technique [13]–[17], the effective transconductance
A-SSCC, Macau SAR, China, November 4-6, 2019. (Corresponding authors: of the CG LNA is enhanced by applying an inverting voltage
Dixian Zhao; Xiaohu You.) gain to the gate and, hence, increasing the voltage difference
The authors are with the National Mobile Communication Research Labo-
ratory, School of Information Science and Engineering, Southeast University, between the source and the gate. This technique can increase
Nanjing 210096, China, and also with the Purple Mountain Laboratories, the gain and reduce the NF while drawing less dc current than
Nanjing 211100, China (e-mail: [email protected]; [email protected]). conventional CG LNA. Besides, the current-reuse technique
Color versions of one or more of the figures in this article are available
online at http://ieeexplore.ieee.org. has been reported in low-power LNA designs [16], [18]–[22]
Digital Object Identifier 10.1109/JSSC.2020.2995307 to realize high gain without additional power.
0018-9200 © 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: University of Exeter. Downloaded on June 06,2020 at 05:52:41 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

2 IEEE JOURNAL OF SOLID-STATE CIRCUITS

In this article, we propose an mm-Wave current-reuse LNA

using the transformer-based gm -boost technique to achieve
high gain and low noise under low power consumption. To
reveal the design detail, an elaborate analysis of the current-
reuse and gm -boosted LNA is carried out. A well-rounded
design methodology on the gm -boost transformer is proposed.
Apart from the turn ratio and coupling coefficient, we further
examine the impact of the self-inductance and the loss of the
gm -boost transformer. Besides, a transformer-based magnet-
ically coupled resonator (MCR) is employed to extend the
bandwidth. A novel design methodology to achieve equal
peak amplitude and low ripple in MCR is proposed, which
is effective to design broadband transformer-based matching
networks with a high-impedance conversion ratio. To verify
the theory, an LNA implemented in 65-nm CMOS technology
is demonstrated. Compared with the state of the art, it achieves
competitive power gain, NF, and bandwidth with the lowest
power consumption, leading to a figure of merit (FoM) as high
Fig. 2. (a) Conventional cascading CG–CS LNA. (b) Current-reuse LNA with
as 12.3. single-ended CG and single-ended CS. (c) Current-reuse LNA with single-
This article is organized as follows. Section II presents a ended CG and differential CS.
detailed analysis of the current-reuse and gm -boost techniques.
The gain, NF, and the input impedance of the transformer-
based gm -boosted LNA are theoretically derived. The design
methodology on the self-inductance and the coupling coef-
ficient of the gm -boost transformer is performed. In Sec-
tion III, the transformer-based MCR for inter-stage matching
is discussed, and a novel design strategy to achieve low-loss
low-ripple MCR is proposed. Then, circuit implementation
and measurement results of the K -band LNA are shown in
Sections IV and V, respectively. Finally, the conclusion is
drawn in Section VI.

II. C URRENT-R EUSE AND gm -B OOST T ECHNIQUE

Fig. 3. Simulated (a) MAG and (b) NFmin of a cascading LNA with single-
A. Current-Reuse Technique ended CG and differential CS under different biasing conditions.

The CG stage can achieve wideband input matching with

low power consumption, but it shows moderate power gain. utilized to build an MCR for wideband inter-stage matching;
To provide enough power gain, a CS stage with high gain can 3) neutralization technique can be easily realized to increase
be cascaded to form a two-stage CG–CS LNA, as shown in the gain and stability; 4) the differential circuit shows even-
Fig. 2(a). This CG–CS configuration combines the advantages order harmonic suppression; and 5) the LNA output can be
of CG and CS stages. The first CG stage provides easy easily connected to the following differential mixer circuits.
input matching, and the second CS stage provides high gain. To evaluate to the effectiveness of the current-reuse tech-
However, the cascading CS stage will consume additional dc nique used in Fig. 2(c), Fig. 3 shows the simulated maximum
power. To minimize the power consumption, the cascading available gain (MAG) and minimum noise figure (NFmin ) of
CG–CS stage can be stacked and share the current from a the LNA in Fig. 2(c) versus different biasing conditions (i.e.,
single 1-V supply [16]. The current-reuse CG–CS LNA is drain-to-source voltage VDS and drain current ID ). Conven-
shown in Fig. 2(b). The dc bias current flows from the CS tionally, the CG and CS transistors are both biased with 1-V
stage to the CG stage through an on-chip RF choke. The large VDS and a normal current ID (marker α). Using the current-
bypass capacitor provides ac ground. However, the design in reuse technique, the two transistors are stacking, and therefore,
[16] is narrowband, and additional isolation components are each transistor is biased with 0.5-V VDS and ID (marker β). In
demanded. In [22], a stagger tuning technique is adopted to addition, marked with γ , each transistor is biased with 1-V VDS
extend the bandwidth of power gain, but the bandwidth of NF and 1/2ID . Among these three typical biasing schemes, α is of
is still limited. the highest MAG and the lowest NFmin , but it consumes double
As shown in Fig. 2(c), we modify the second CS stage dc power. In comparison, the biasing points on the red dashed
from single-ended to differential topology. This topology has line, including β and γ , consume only half the dc power of α.
several advantages: 1) the common-mode rejection property Moreover, among β, γ , and other biasing points with the same
of the differential CS stage performs ideal dc conduction and power consumption, β has the highest MAG and the low-
ac isolation between the two stages; 2) transformer can be est NFmin . Comparing with the conventional biasing scheme

Authorized licensed use limited to: University of Exeter. Downloaded on June 06,2020 at 05:52:41 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

ZHANG et al.: 20-GHz 1.9-mW LNA USING gm -BOOST AND CURRENT-REUSE TECHNIQUES 3

Fig. 5. Equivalent small-signal model of the gm -boost CG-LNA.

Fig. 4. Different gm -boost techniques. (a) Active device. (b) Capacitive

coupling. (c) and (d) Magnetically coupling.

(i.e., α), the current-reuse CG–CS stage (i.e., β) saves half

power consumption but only degrades the MAG from 27
to 25 dB and the NFmin from 1.73 to 1.93 dB. Hence, the
current-reuse CG–CS stage outperforms conventional cascad-
ing CG–CS stage.

B. gm -Boost Technique
Fig. 6. Simulated MAG and NFmin of a conventional CG-LNA and a gm -
To further enhance the gain and noise performances, the gm - boosted CG-LNA under the same power consumption.
boost technique is proposed in CG LNA. The basic topology
of the gm -boost technique is shown in Fig. 4(a). An auxil-
iary amplifier is inserted between the source and the gate The inductors L P and L S represent the self-inductance of the
of the CG transistor to apply negative voltage gain to the transformer. Two current-controlled current sources represent
gate. However, the auxiliary amplifier will cause additional the magnetically coupling between the two coils, where k is
dc power and noise, especially at mm-Wave frequencies. the coupling coefficient and n is the turn ratio. Similar to [14],
Alternatively, passive devices (e.g., capacitors or transformers) if we neglect the channel length modulation and body effect,
can be employed. As shown in Fig. 4(b), the capacitive the input admittance can be derived as
coupling technique utilizes a pair of cross-coupled capacitors  2 
s L P + n + 2kn + 1 sCgs + (1 + kn)gm
1
CC [15]. It is easy to implement (as long as CC  Cgs ), YIN =   (1)
and it shows almost no additional noise. However, capacitive 1 − 1 − k 2 ω2 Cgs L S
coupling is only available in differential topology, and it can and the effective transconductance can be derived as
only achieve unity inverting gain. For higher voltage gain, (1 + kn)
transformer-based gm -boost technique is reported [14], [16]. Gm =   gm . (2)
1 − 1 − k 2 ω2 Cgs L S
The magnetically coupling between the transformer coils
greatly boosts the effective transconductance of the transistor. With proper selection of L P and Cgs , the imaginary part
Unlike capacitive coupling, magnetically coupling can be real- of (1) can be eliminated. Because the input impedance of
ized in both differential and single-ended topologies [14], as the CG LNA shows relatively low Q-factor, broadband input
shown in Fig. 4(c) and (d). In the differential topology, a three- impedance matching can be realized with the relatively simple
coil transformer should be used for gm -boost function and passive network [12]. Then, the input impedance of the gm -
input matching with the single-ended antenna. It increases the boosted CG LNA has an only real part, and it is reduced by
design complexity. In contrast, the single-ended configuration a factor of (1 + kn) due to the gm -boost technique. It means
is easier to implement, and it shows a relatively low intrinsic that 50- input matching can be realized with a smaller gm ,
loss. Furthermore, single-ended LNA consumes much less and hence, the power consumption is saved. Besides, G m is
power and shows lower input impedance than the differential increased by a factor of (1 + kn), by which the power gain is
one. Previous works [14], [16] mainly focus on the turn ratio effectively increased. The gm -boost technique can also reduce
and coupling coefficient of the transformer. In this article, a the NF of the LNA. The NF of the circuit in Fig. 4(d) is [14]
well-rounded design methodology on the gm -boost transformer    2
γ 1 δα ω 2 n 2 + 2kn + 1
is proposed. Apart from the turn ratio and coupling coefficient, F = 1+ + (3)
the impact of the self-inductance of the gm -boost transformer α 1 + kn 5 ωT (1 + kn)3
is further examined. where γ , α, and δ are the process-dependent parameters.
The small-signal model of the single-ended magneti- The simulated results of a conventional CG LNA and a
cally coupling gm -boosted CG LNA is shown in Fig. 5. gm -boosted CG LNA versus frequencies are plotted in Fig. 6

Authorized licensed use limited to: University of Exeter. Downloaded on June 06,2020 at 05:52:41 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

4 IEEE JOURNAL OF SOLID-STATE CIRCUITS

Fig. 7. Simulated MAG and NFmin of a gm -boosted CG-LNA for a different

turn ratio n under the condition k = 0.7. Fig. 8. Simulated MAG and NFmin of a gm -boosted CG-LNA for a different
coupling coefficient k under the condition n = 1.

assuming a transformer with k = 1 and n = 1. The two LNAs

are kept with the same power consumption. It can be seen that
the gm -boosted CG stage shows a higher MAG and a lower
NFmin , which is consistent with the analysis.
The design of the gm -boost transformer merits considera-
tions, as the improvement of the gm -boosted CG LNA mainly
depends on the transformer parameters: L P , L S , n, and k.
A high n can enhance the voltage amplification and then lead
to high gm and low NF. Fig. 7 shows the simulated MAG
and NFmin of a gm -boosted CG LNA for a different turn ratio
n at 20 GHz, and the dashed lines represent the MAG and
Fig. 9. Calculated and simulated optimum k for transformers with a different
NFmin of the conventional CG LNA. In the simulation, the L S and n.
primary coil L P at the input is swept, and the secondary coil
L S keeps unchanged (i.e., the turn ratio changes from 0.5 to 4). Assuming that d = ω2 L S Cgs , (2) can be rewritten as
The gm -boosted CG LNA can achieve 2.6-dB improvement in
MAG and 0.5-dB improvement in NFmin when n = 1 and (1 + kn)
Gm =   gm . (5)
further increasing the value of n can only bring marginal 1 − 1 − k2 d
improvement in NFmin . The two main noise sources of the CG Equation (5) reaches its maximum when
transistor are channel noise and gate-induced noise [23]. The   
NF contribution of the channel noise decreases as n increases, −d + d n 2 − 2nd + d
but, meanwhile, the contribution of the gate-induced noise k= . (6)
increases. It is because a high n (i.e., a high gm ) will reduce nd
the input-referred noise of the channel noise. At the same To evaluate (5) and (6), we compare the calculated and
time, the gate-induced noise current is amplified and present simulated optimum k for transformers with a different L S and
at the primary coil side (i.e., the input of the LNA) due to the n. As shown in Fig. 9, the two groups of curves are very close
current amplification of the transformer. Hence, a moderate n to each other. It suggests that (6) is an accurate estimation of
is preferred. the optimum k to achieve the highest MAG. Hence, a high k
Moreover, it should be noticed that a high n will potentially close to unity may not be the best option, especially when n
make the LNA unstable. The denominator in (1) is 1 − (1 − is relatively small and L S is relatively large.
k 2 )ω2 L S Cgs . If (1 − k 2 )ω2 L S Cgs > 1, the CG-LNA shows The loss of the gm -boost transformer will affect the NF
negative input resistance, which suggests that the value of L S directly. Therefore, the Q-factors of the transformer coils are
must satisfy utilized to model the loss. The Q-factors of L P and L S , are
Q P and Q S , respectively. The simulated MAG and NFmin of
1 gm -boosted LNA with a different Q P and Q S at 20 GHz are
LS <   . (4)
1 − k 2 ω2 Cgs shown in Fig. 10, assuming that n = 1 and k = 0.7. To
analyze the noise contributions from each coil, we perform
Since n 2 = L S /L P , (4) indicates the upper bound of n. the simulation under three conditions: 1) considering only the
Although a smaller L P can also lead to a higher n, it will loss of L S and sweeping Q S ; 2) considering only the loss of
affect the Q-factor of the input impedance of the CG LNA L P and sweeping Q P ; and 3) considering both the loss of L P
and deteriorate the input matching bandwidth. and L S and sweeping Q P and Q S . As shown in Fig. 10, the
Fig. 8 shows the simulated MAG and NFmin of a gm -boosted loss of the transformer has little impact on MAG. However,
CG LNA for different k at 20 GHz. It can be seen that with the deterioration of NFmin is obvious as Q P decreases. Hence,
the increase of k, NFmin decreases. In contrast to the theory in the loss of the primary coil L P has a larger impact on NFmin
[14], the MAG also decreases, which can be explained by (2). than the loss of L S .

Authorized licensed use limited to: University of Exeter. Downloaded on June 06,2020 at 05:52:41 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

ZHANG et al.: 20-GHz 1.9-mW LNA USING gm -BOOST AND CURRENT-REUSE TECHNIQUES 5

Fig. 13. Simulated MAG and NFmin of a gm -boosted CG-LNA with planar
and overlay transformers.
Fig. 10. Simulated (a) MAG and (b) NFmin of a gm -boosted CG-LNA for
transformers with a different Q under the conditions n = 1 and k = 0.7.

Fig. 14. Transformer-based MCR.

Q-factor and reduce the loss. Although it is difficult to obtain

Fig. 11. 3-D view of (a) planar and (b) overlay transformers. strong coupling for planar transformer, k of about 0.7 is a
proper value weighing MAG and NFmin . As depicted earlier,
when n = 1–2, a moderate k helps to improve the gain,
despite a slight deterioration in NF. Fig. 13 shows the plots
of the simulated MAG and NFmin when the two transformers
in Fig. 12 are applied to the gm -boosted CG LNA. The LNA
with planar transformer shows 0.6 dB higher MAG and 0.2 dB
less NFmin than the overlay transformer at 20 GHz. Hence, a
planar transformer is preferred for the gm -boosted CG LNA
to increase the gain and reduce the NF.

III. MCR FOR I NTER -S TAGE M ATCHING

The transformer is widely adopted for inter-stage matching
at mm-Wave frequencies [25]–[29]. It is because the trans-
former can achieve a compact layout, easy supply or bias
feedings, and single-ended-to-differential signal conversion.
Transformers can also build high-order MCR to extend the
bandwidth. The properties of different high-order coupled
resonator have been analyzed in [25], and it proves the
effectiveness of MCR to enhance the gain–bandwidth product.
In this work, a novel MCR design methodology that is useful
Fig. 12. EM-simulated performances of planar and overlay transformers. for the design of high-impedance-conversion-ratio-matching
(a) Self-inductance. (b) Quality factor. (c) MAG. (d) Coupling coefficient.
networks is proposed. It can achieve low-loss, low-ripple, and
easy-to-implement transformer-based MCR.
In transformer implementation, transformers with n > 2 are The schematic of the MCR is shown in Fig. 14. The
less attractive due to high resistive and capacitive parasitics, transformer can be extracted as self-inductances of the primary
in addition to the low self-resonate frequency [23]. Hence, and secondary windings L 1 and L 2 with a turn ratio n. The
it is appropriate to select n = 1–2. There are mainly two magnetic coupling between the two windings is represented
transformer structures, planar, and overlay (see Fig. 11) [24]. by the coupling coefficient k. The resistors R1 and R2 at
Planar and overlay transformers are designed as examples, each side of the transformer represents the input and output
and the electromagnetic (EM)-simulated results are shown resistances of the transistors. The parallel capacitors at each
in Fig. 12. Overlay transformer has a higher k, but it uses side (including the transistor parasitic capacitance) are C1 and
two metal layers. The lower thin metal with poor Q-factor C2 , respectively. The transformer self-inductances L 1 and L 2
will make NF worse, and the high parasitic capacitance will absorb the capacitances C1 and C2 , form RLC tanks at both
deteriorate the bandwidth. For the planar transformer, both sides, and resonate at ω1 and ω2 , respectively. The resonating
coils can be designed with the top thick metal to improve frequencies ω1 and ω2 are not necessary to be the same.

Authorized licensed use limited to: University of Exeter. Downloaded on June 06,2020 at 05:52:41 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

6 IEEE JOURNAL OF SOLID-STATE CIRCUITS

Fig. 15. Simulated Z 21 of the MCR with different resonating frequencies Fig. 16. Inter-stage matching MCR in this design.
(R1 = R2 = 500 , k = 0.3). (a) C1 = C2 = 70 fF, L 1 = 1 nH, and sweeping
L 2 for 1 nH, 500 pH, and 250 pH. (b) C1 = 70 fF, L 1 = L 2 = 1 nH, and
sweeping C2 for 70, 35, and 17.5 fF.

Fig. 15 show the plots of the simulated Z 21 of MCR with

unequal resonating frequencies (i.e., ω1 = ω2 ), where L 2 is
swept in Fig. 15(a) and C2 is swept in Fig. 15(b). As shown in
Fig. 15, the peaks of Z 21 move toward high frequencies, while
the peak amplitudes decrease. Besides, there are two peaks in
the frequency response of the MCR. It can be seen that the
two peak amplitudes in Fig. 15(a) are equal, while they are Fig. 17. Simulated Z 21 of the MCR with the same turn ratio (R1 = 500 ,
R2 = 5 k, k = 0.3, sweeping L 1 /L 2 for 1/10, 0.8/8, 0.5/5, and 0.2/2 nH).
unequal in Fig. 15(b). The amplitude difference  of the two
peaks of MCR can be derived as
√ of the working band, so we have ω1 ≈ ω2 . With ω1 ≈ ω2 and
k R1 R2 L 1 L 2 (L 1 C1 − L 2 C2 )(R1 C1 − R2 C2 ) (9), we can build the relationship between L 1 , L 2 and R1 , R2
= (7)
M
L1 R1
where ≈ . (12)
    L2 R2
M = R1 R2 L 21 C12 + L 22 C22 + k 2 L 1 L 2 R12 C12 + R22 C22
  Notice that (12) is the most widely used design method for
− 2 1 − k 2 (L 1 L 2 C1 C2 R1 R2 ). (8)
an MCR [26], [28]. Therefore, the turn ratio n is determined
Flat frequency response of the MCR is preferred, and the peak by the ratio between R1 and R2 . This condition permits the
amplitudes should be equal. Let  = 0, and we have two maximum gain but does not optimize the ripple. Practically,
different solutions the values of L 1 and L 2 are limited due to the parasitics and the
ripple will become considerable. Hence, a novel MCR design
R1 C 1 = R2 C 2 (9) method is proposed to mitigate the ripple.
L 1 C1 = L 2 C2 . (10) It is difficult to satisfy (12) for a matching network with
a high-impedance-conversion ratio. For example, the inter-
The first solution is (9). If (9) is satisfied, both peaks in the
stage MCR of the LNA is shown in Fig. 16. The input
Z 21 are of equal amplitude regardless of the self-inductance
resistance of the differential CS stage is around 5 k, almost
of the transformer (i.e., L 1 and L 2 ). As shown in Fig. 15(a),
ten times as large as the output resistance of the single-ended
sweeping L 2 does not change the equality. However, the values
CG stage. Assuming a matching network with unequal load
of L 1 and L 2 cannot be selected arbitrarily, as they can
(R1 = 500  and R2 = 5 k), Fig. 17 shows the plots
change the resonating frequencies ω1 and ω2 of the MCR and,
of the simulated Z 21 with different inductances L 1 and L 2
therefore, the peak frequencies. The frequency spacing of the
satisfying (12). When L 1 = 1 nH and L 2 = 10 nH, the
two peak amplitudes is
 MCR is of the highest and flattest amplitude. However, such
2   a transformer is unrealizable at mm-Wave as the parasitic
ω12 + ω22 − 4 1 − k 2 ω12 ω22
ωS = (11) resistance and capacitance associated with the coils severely
1 − k2 degrade Q-factor and self-resonant frequency [23]. While
which is determined by ω1 and ω2 . If (9) is satisfied, adjusting MCR with L 1 = 0.2 nH and L 2 = 2 nH is easier to implement,
L 1 and L 2 has no impact on the amplitude equality of the peak, it suffers from unacceptable ripple as high as 9 dB. Therefore,
while the situation of the spacing ωS is quite the opposite. As the solution (12) is not suitable for this design.
shown in Fig. 15, the larger the difference between ω1 and ω2 , In this work, the optimization of the MCR is only based on
the wider the spacing between the two peaks, with the penalty the solution (10). When L 1 C1 = L 2 C2 , both peaks in the Z 21
of increased loss and ripple. Generally, moderate bandwidth are of equal amplitude regardless of R1 and R2 . This property
is often required. In this design, the required satellite band is breaks the relationship between L 1 , L 2 and R1 , R2 . The turn
between 17.7 and 20.2 GHz. In order to maximize the power ratio n is no longer consistent with the impedance-conversion
gain, ω1 and ω2 should be located around the central frequency ratio.

Authorized licensed use limited to: University of Exeter. Downloaded on June 06,2020 at 05:52:41 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

ZHANG et al.: 20-GHz 1.9-mW LNA USING gm -BOOST AND CURRENT-REUSE TECHNIQUES 7

Fig. 18. Simulated Z 21 of the MCR with a different turn ratio (R1 = 500 ,
R2 = 5 k, and k = 0.3). (a) Sweeping L 2 /C2 for 5 nH/14 fF, 2 nH/35 fF,
1 nH/70 fF, and 500 pH/140 fF. (b) Sweeping L 1 /C1 for 200 pH/350 fF,
500 pH/140 fF, 1 nH/70 fF, and 2 nH/35 fF.
Fig. 19. Simulated Z 21 (a) peak and (b) ripple of the MCR with different
If (10) is satisfied, the peak and the ripple amplitude of Z 21 L 1 and L 2 (R1 = 500 , R2 = 5 k, and k = 0.3).
of the MCR can be derived as
 
R1 R2 R1 R2
Peak = √ + (13)
L1 L2 L1 L2
   
R1 R2 R1 R2
Ripple = 1 − k 2 + k 2 k 1 − k2 + .
ωL 1 L 2 L1 L2
(14)
Once the transistor size at each side is determined, R1 and
R2 of the MCR are given. Then, the peak and the ripple are
only determined by k, L 1 and L 2 . Increasing k will enlarge
the peak spacing and will not change the peak amplitude at
the cost of a large ripple. However, increasing L 1 and L 2
will increase the peak and decrease ripple. Considering the Fig. 20. Schematic of the proposed CG–CS LNA.
transformer implementation, the values of L 1 and L 2 cannot
be made arbitrarily high. If we treat L 1 L 2 product as a
variable, it can be concluded that both the peak and ripple achieves the highest peak Z 21 when (12) is satisfied. For
amplitude reach its maximum value when (12) holds. In order instance, MCR at point A (L 1 = 1 nH and L 2 = 10 nH) is on
to satisfy (12), the values of L 1 and L 2 must be decreased the red line and satisfies (12). It achieves the highest peak, but
to a certain low level for easy implementation, and thus, the the value of L 2 is too large to realize. Also satisfying (12),
ripple becomes unacceptable. For the circuit in this design, we point B (L 1 = 0.2 nH and L 2 = 2 nH) is easy to implement,
break the relationship in (12). From (13) and (14), although but it suffers from severe ripple up to 9 dB. A better alternative
the peak decreases slightly, the ripple is mitigated. Hence, L 1 is at around point C (L 1 = 1 nH and L 2 = 2 nH). It
and L 2 can be set independently to reasonable values for ripple satisfies (10) instead of (12) and shows a 55-dB peak and
reduction. Besides, as long as (10) is satisfied, the two peaks only 2-dB ripple. The MCR at C not only achieves high-peak
of Z 21 are of equal amplitude. and low-ripple Z 21 but also it is easy to implement. Hence,
As shown in Fig. 18(a), with decreased L 2 and increased C2 , the design method based on (10) is useful for the design of
the two peak amplitudes keep equal. The amplitudes decrease, high-impedance-conversion-ratio-matching networks.
and it can be explained by (13). Besides, L 1 can be increased
to compensate for the ripple. Fig. 18(b) shows the plots of the IV. C IRCUIT I MPLEMENTATION
Z 21 for different L 1 C1 tanks. With increased L 1 and decreased Based on the design considerations, a systematic design
C1 , the ripple is mitigated. Hence, L 1 and L 2 can be adjusted methodology has been developed for the gm -boosted current-
to ease the implementation of the transformer-based MCR with reused CG-CS LNA. Fig. 20 shows the schematic of the
small ripple, while the peak equality is ensured. proposed LNA. The LNA consists of a single-ended CG input
To have an insight into the design of the inter-stage match- stage and differential CS second stage.
ing transformer, Fig. 19 shows the simulated Z 21 for different At the single-ended CG stage, transformer XF1 is adopted
L 1 and L 2 , while (10) is satisfied. Fig. 19(a) shows the for the gm -boost function. As mentioned in Section II, increas-
peak Z 21 of the MCR, and Fig. 19(b) shows the peak-to- ing n can improve MAG and NFmin , but the improvement is
valley ripple of the MCR. The performances of the MCR with marginal when n > 2. Moreover, a large n will reduce the
constant L 1 L 2 product are plotted with gray dot curves as a transistor size of the CG stage for input matching, and then,
reference. The MCR satisfying (12) (i.e., L 1 :L 2 = R1 :R2 ) is the dc current is reduced to a low level. However, the current is
plotted with the dashed red line. For a certain gay curve, the reused by the second differential CS stage. To ensure sufficient
highest peak amplitude appears at its intersection of the red power gain of the second differential stage, a moderate dc
line. In other words, for a certain L 1 L 2 product, the MCR current is needed. Therefore, n = 1 is chosen to provide

Authorized licensed use limited to: University of Exeter. Downloaded on June 06,2020 at 05:52:41 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

8 IEEE JOURNAL OF SOLID-STATE CIRCUITS

Fig. 21. EM-simulated performances of XF2 . Fig. 24. Measured NF of the LNA.

Fig. 25. Measured input P1 dB of the LNA.

Fig. 22. Chip photograph of the proposed CG–CS LNA.
results of XF2 are shown in Fig. 21. It is also a balun
for single-to-differential signal conversion and current-reuse
function. Transformer XF3 is a strongly coupled transformer to
fulfill low-loss output matching. The 1-V dc supply VDD is fed
through the center tap of the output matching transformer XF3
and shared by the two stages. A wide, thick metal connects
the source of the CS stage and the drain of the CG stage as
the dc current path, and a large bypass capacitor provides the
ac ground.

V. M EASUREMENT R ESULTS
The proposed LNA is fabricated in 65-nm CMOS technol-
ogy. As shown in Fig. 22, the LNA occupies a chip area of
Fig. 23. Measured S21 and S11 of the LNA.
600 μm × 700 μm, including G–S–G pads, dc pads, and
decoupling capacitors. The overall circuit consumes 1.9 mW
enough dc current, and then, L P is set to 600 pH to eliminate from a 1-V supply. S-parameters and NF are measured on a
the imaginary part of the input impedance. According to (8), high-frequency probe station from 1 to 26.5 GHz. The LNA
the optimal k to obtain the highest MAG is 0.3. However, is unconditional stable.
it is not the optimal value for NFmin because NFmin always The measured S21 and S11 are shown in Fig. 23. The LNA
improves with increased k. As plotted in Fig. 8, the MAG achieve a peak gain of 14.9 dB at 21 GHz with a −3-dB
decreases, but the NFmin improves with increased k, where bandwidth of 4.8 GHz from 17.2 to 22.0 GHz. The −1.5-dB
the improvement in NFmin is obvious, and the deterioration in bandwidth is 3.9 GHz from 17.7 to 21.6 GHz, covering the
MAG is marginal. Finally, XF1 is designed in planar structure satellite band (i.e., 17.7–20.2 GHz). The measured S11 is
with n = 1 : 1 and k = 0.7, and the EM-simulated results of better than −10 dB from 16.0 to 21.6 GHz, also covering
the planar transformer are shown in Fig. 12. the satellite band.
The second stage employs differential CS topology. Trans- The measured NF is shown in Fig. 24. The lowest NF of
former XF2 is placed between the two stages to form the 3.3 dB is achieved at 19.5 GHz, and the NF is below 4 dB
MCR for inter-stage matching. XF2 is a weakly coupled from 17 to 21 GHz. At the satellite band (i.e., 17.7–20.2 GHz),
transformer, of which k equals 0.3 in order to cover the the NF is less than 3.6 dB. Fig. 25 shows that the measured
working frequencies. The self-inductance is carefully selected input P1 dB is −24 dBm at 21 GHz. The linearity of the LNA
to achieve a high peak and low ripple. The EM-simulated is sufficient for satellite applications.

Authorized licensed use limited to: University of Exeter. Downloaded on June 06,2020 at 05:52:41 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

ZHANG et al.: 20-GHz 1.9-mW LNA USING gm -BOOST AND CURRENT-REUSE TECHNIQUES 9

TABLE I
C OMPARISON OF S TATE - OF - THE -A RT K -BAND CMOS LNA S

Table I summarizes the measured performances of the pro- [4] T. Yao et al., “Algorithmic design of CMOS LNAs and PAs for 60-GHz
posed LNA and compares it with the state-of-the-art K -band radio,” IEEE J. Solid-State Circuits, vol. 42, no. 5, pp. 1044–1057,
May 2007.
CMOS LNAs. This LNA achieves the lowest power consump- [5] D. K. Shaeffer and T. H. Lee, “A 1.5-V, 1.5-GHz CMOS low noise
tion with competitive gain, NF, and bandwidth performances, amplifier,” IEEE J. Solid-State Circuits, vol. 32, no. 5, pp. 745–759,
leading to an FoM as high as 12.3. May 1997.
[6] T.-K. Nguyen, C.-H. Kim, G.-J. Ihm, M.-S. Yang, and S.-G. Lee,
“CMOS low-noise amplifier design optimization techniques,” IEEE
VI. C ONCLUSION Trans. Microw. Theory Techn., vol. 52, no. 5, pp. 1433–1442, May 2004.
In this article, an ultra-low-power LNA in 65-nm CMOS [7] H. Hashemi and A. Hajimiri, “Concurrent multiband low-noise
is proposed. The LNA is cascaded with a single-ended CG amplifiers-theory, design, and applications,” IEEE Trans. Microw. Theory
Techn., vol. 50, no. 1, pp. 288–301, Jan. 2002.
stage and a differential CS stage. The power consumption is [8] A. Ismail and A. A. Abidi, “A 3-10-GHz low-noise amplifier with
minimized, while RF performances are maintained due to the wideband LC-ladder matching network,” IEEE J. Solid-State Circuits,
current-reuse technique. Besides, the transformer-based gm - vol. 39, no. 12, pp. 2269–2277, Dec. 2004.
[9] A. Bevilacqua and A. M. Niknejad, “An ultra-wideband CMOS LNA for
boost technique in the CG stage helps to increase the gain, 3.1 to 10.6 GHz wireless receivers,” in IEEE ISSCC Dig. Tech. Papers,
reduce the noise, and further save power. The neutralization Feb. 2004, pp. 382–533.
technique in differential CS stage improves power gain and [10] M. T. Reiha and J. R. Long, “A 1.2 V reactive-feedback 3.1–10.6 GHz
stability. The inter-stage transformer builds an MCR and low-noise amplifier in 0.13 μm CMOS,” IEEE J. Solid-State Circuits,
vol. 42, no. 5, pp. 1023–1033, May 2007.
extends the bandwidth. Detailed analysis of the designs of [11] L. Belostotski and J. W. Haslett, “Sub-0.2 dB noise figure
the gm -boost transformer and inter-stage matching transformer wideband room-temperature CMOS LNA with non-50  signal-
is presented. Systematic design methodologies on the current- source impedance,” IEEE J. Solid-State Circuits, vol. 42, no. 11,
pp. 2492–2502, Nov. 2007.
reused CG–CS LNA using a transformer-based gm -boost tech- [12] H. Zhang, X. Fan, and E. S. Sinencio, “A low-power, linearized, ultra-
nique and transformer-based MCR are proposed. Fabricated wideband LNA design technique,” IEEE J. Solid-State Circuits, vol. 44,
in 65-nm CMOS technology, the K -band LNA achieves the no. 2, pp. 320–330, Feb. 2009.
lowest power consumption with competitive gain, NF, and [13] D. J. Allstot, X. Li, and S. Shekhar, “Design considerations for CMOS
low-noise amplifiers,” in IEE Radio Freq. Integr. Circuits (RFIC) Sys-
bandwidth performances. At the satellite band (i.e., 17.7– tems. Dig. Papers, Fort Worth, TX, USA, Jun. 2004, pp. 97–100.
20.2 GHz), the LNA achieves 14.9-dB gain with 1.5-dB ripple [14] X. Li, S. Shekhar, and D. J. Allstot, “Gm -boosted common-gate LNA
and 3.6-dB NF and only consumes 1.9 mW from the 1-V and differential colpitts VCO/QVCO in 0.18-μm CMOS,” IEEE J. Solid-
State Circuits, vol. 40, no. 12, pp. 2609–2619, Dec. 2005.
supply. [15] W. Zhuo et al., “A capacitor cross-coupled common-gate low-noise
amplifier,” IEEE Trans. Circuits Syst. II, Exp. Briefs, vol. 52, no. 12,
R EFERENCES pp. 875–879, Dec. 2005.
[1] Z. Zhang et al., “6G wireless networks: Vision, requirements, architec- [16] S. Shekhar, J. S. Walling, S. Aniruddhan, and D. J. Allstot, “CMOS
ture, and key technologies,” IEEE Veh. Technol. Mag., vol. 14, no. 3, VCO and LNA using tuned-input tuned-output circuits,” IEEE J. Solid-
pp. 28–41, Sep. 2019. State Circuits, vol. 43, no. 5, pp. 1177–1186, May 2008.
[2] G. W. de Jong, D. M. W. Leenaerts, and E. van der Heijden, “A fully [17] M. Khurram and S. M. R. Hasan, “A 3–5 GHz current-reuse gm -boosted
integrated ka-band VSAT down-converter,” IEEE J. Solid-State Circuits, CG LNA for ultrawideband in 130 nm CMOS,” IEEE Trans. Very Large
vol. 48, no. 7, pp. 1651–1658, Jul. 2013. Scale Integr. (VLSI) Syst., vol. 20, no. 3, pp. 400–409, Mar. 2012.
[3] C. Friesicke, P. Feuerschutz, R. Quay, O. Ambacher, and A. F. Jacob, [18] M. Parvizi, K. Allidina, and M. N. El-Gamal, “A sub-mW, Ultra-Low-
“A 40 dBm AlGaN/GaN HEMT power amplifier MMIC for SatCom voltage, wideband low-noise amplifier design technique,” IEEE Trans.
applications at K-band,” in IEEE MTT-S Int. Microw. Symp. Dig., San Very Large Scale Integr. (VLSI) Syst., vol. 23, no. 6, pp. 1111–1122,
Francisco, CA, USA, May 2016, pp. 1–4. Jun. 2015.

Authorized licensed use limited to: University of Exeter. Downloaded on June 06,2020 at 05:52:41 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

10 IEEE JOURNAL OF SOLID-STATE CIRCUITS

[19] G. Sapone and G. Palmisano, “A 3–10-GHz low-power CMOS low-noise Jiajun Zhang (Student Member, IEEE) received the
amplifier for ultra-wideband communication,” IEEE Trans. Microw. B.Sc. degree in information science and engineering
Theory Techn., vol. 59, no. 3, pp. 678–686, Mar. 2011. from Southeast University, Nanjing, China, in 2017,
[20] Y. Lin et al., “Analysis and design of a CMOS UWB LNA with dual- where he is currently pursuing the Ph.D. degree.
R LC-branch wideband input matching network,” IEEE Trans. Microw. His current research interests include RF and
Theory Techn., vol. 58, no. 2, pp. 287–296, Feb. 2010. millimeter-wave integrated circuits for wireless com-
[21] Y. Lin et al., “A 3.1–10.6 GHz ultra-wideband CMOS low noise ampli- munications and phased-array systems.
fier with current-reused technique,” IEEE Microw. Wireless Compon.
Lett., vol. 17, no. 3, pp. 232–234, Mar. 2007.
[22] R.-M. Weng, C.-Y. Liu, and P.-C. Lin, “A low-power full-band low-noise
amplifier for ultra-wideband receivers,” IEEE Trans. Microw. Theory
Techn., vol. 58, no. 8, pp. 2077–2083, Aug. 2010.
[23] B. Razavi, RF Microelectronics, 2nd ed. Upper Saddle River, NJ, USA: Dixian Zhao (Member, IEEE) received the B.Sc.
Prentice-Hall, 2011. degree in microelectronics from Fudan University,
[24] J. R. Long, “Monolithic transformers for silicon RF IC design,” IEEE Shanghai, China, in 2006, the M.Sc. degree in micro-
J. Solid-State Circuits, vol. 35, no. 9, pp. 1368–1382, Sep. 2000. electronics from the Delft University of Technology
[25] M. Vigilante and P. Reynaert, “On the design of wideband transformer- (TU Delft), Delft, The Netherlands, in 2009, and the
based fourth order matching networks for E-band receivers in 28-nm Ph.D. degree in electrical engineering from the Uni-
CMOS,” IEEE J. Solid-State Circuits, vol. 52, no. 8, pp. 2071–2082, versity of Leuven (KU Leuven), Leuven, Belgium,
Aug. 2017. in 2015.
[26] H. Jia, C. C. Prawoto, B. Chi, Z. Wang, and C. P. Yue, “A full ka- From 2005 to 2007, he was with Auto-ID Lab,
band power amplifier with 32.9% PAE and 15.3-dBm power in 65- Shanghai, where he developed the non-volatile mem-
nm CMOS,” IEEE Trans. Circuits Syst. I, Reg. Papers, vol. 65, no. 9, ory for passive RFID tags. From 2008 to 2009, he
pp. 2657–2668, Sep. 2018. was with Philips Research, Eindhoven, The Netherlands, where he designed
[27] D. Zhao and P. Reynaert, “A 60-GHz dual-mode class AB power the 60-GHz beamforming transmitter for presence detection radar. From
amplifier in 40-nm CMOS,” IEEE J. Solid-State Circuits, vol. 48, no. 10, 2009 to 2010, he was a Research Assistant with TU Delft, working on
pp. 2323–2337, Oct. 2013. the 94-GHz wideband receiver for imaging radar. From 2010 to 2015, he
[28] C.-H. Li, C.-N. Kuo, and M.-C. Kuo, “A 1.2-V 5.2-mW 20–30-GHz was a Research Associate with KU Leuven, where he developed several
wideband receiver front-end in 0.18-μm CMOS,” IEEE Trans. Microw. world-class 60-GHz and E-band transmitters and power amplifiers. In April
Theory Techn., vol. 60, no. 11, pp. 3502–3512, Nov. 2012. 2015, he joined Southeast University, Nanjing, China, and is currently a Full
[29] H. Wang, C. Sideris, and A. Hajimiri, “A CMOS broadband power Professor. He has authored or coauthored more than 50 peer-reviewed journal
amplifier with a transformer-based high-order output matching network,” and conference articles, one book, and two book chapters. His current research
IEEE J. Solid-State Circuits, vol. 45, no. 12, pp. 2709–2722, Dec. 2010. interests include millimeter-wave integrated circuits, transceivers and phased-
[30] W.-C. Wang et al., “A 1 V 23 GHz low-noise amplifier in 45 nm planar array systems for 5G, satellite, radar, and wireless power transfer applications.
bulk-CMOS technology with high-Q above-IC inductors,” IEEE Microw. Prof. Zhao was a recipient of the Innovative and Entrepreneurial Talent of
Wireless Compon. Lett., vol. 19, no. 5, pp. 326–328, May 2009. Jiangsu Province in 2016, the IEEE Solid-State Circuits Society Predoctoral
[31] H.-Y. Chang et al., “65-nm CMOS dual-gate device for ka-band Achievement Award in 2014, the Chinese Government Award for Outstanding
broadband low-noise amplifier and high-accuracy quadrature voltage- Students Abroad in 2013, and the Top-Talent Scholarship from TU Delft in
controlled oscillator,” IEEE Trans. Microw. Theory Techn., vol. 61, no. 6, 2007 and 2008. He serves as a Technical Program Committee (TPC) Member
pp. 2402–2413, Jun. 2013. or the Sub-Committee Chair of several conferences, including the IEEE
[32] M.-H. Tsai et al., “ESD-protected K-Band low-noise amplifier using European Solid-State Circuits Conference (ESSCIRC), the IEEE Asian Solid-
RF junction varactors in 65-nm CMOS,” IEEE Trans. Microw. Theory State Circuits Conference (A-SSCC), and the IEEE International Conference
Techn., vol. 59, no. 12, pp. 3455–3462, Dec. 2011. on Integrated Circuits, Technologies and Applications (ICTA).
[33] M.-H. Tsai et al., “A 17.5-26 GHz low-noise amplifier with over 8 kV
ESD protection in 65 nm CMOS,” IEEE Microw. Wireless Compon.
Lett., vol. 22, no. 9, pp. 483–485, Sep. 2012. Xiaohu You (Fellow, IEEE) received the B.S., M.S.,
[34] P. Qin et al., “Compact wideband LNA with gain and input matching and Ph.D. degrees in electrical engineering from the
bandwidth extensions by transformer,” IEEE Microw. Wireless Compon. Nanjing Institute of Technology, Nanjing, China, in
Lett., vol. 22, no. 9, pp. 483–485, Sep. 2012. 1982, 1985, and 1989, respectively.
[35] Y. Ding, S. Vehring, and G. Boeck, “Design and implementation of an From 1987 to 1989, he was a Lecturer with the
ultracompact LNA with 23.5-dB gain and 3.3-dB noise figure,” IEEE Nanjing Institute of Technology. Since 1990, he has
Microw. Wireless Compon. Lett., vol. 29, no. 6, pp. 406–408, Jun. 2019. been with Southeast University, Nanjing, first as an
[36] T.-P. Wang, “A low-voltage low-power K-band CMOS LNA using DC- Associate Professor and then as a Professor. He
current-path split technology,” IEEE Microw. Wireless Compon. Lett., contributed more than 40 IEEE journal articles and
vol. 20, no. 9, pp. 519–521, Sep. 2010. two books in the areas of adaptive signal process-
[37] J.-F. Yeh, C.-Y. Yang, H.-C. Kuo, and H.-R. Chuang, “A 24-GHz ing and neural networks and their applications to
transformer-based single-in differential-out CMOS low-noise ampli- communication systems. He was the Premier Foundation Investigator of the
fier,” in Proc. IEEE Radio Freq. Integr. Circuits Symp., Jun. 2009, China National Science Foundation. From 1999 to 2002, he was the Principal
pp. 299–302. Expert of the C3G Project, responsible for organizing China’s 3G mobile
[38] A. Sayag et al., “A 25 GHz 3.3 dB NF low noise amplifier based upon communications research and development activities. From 2001 to 2006, he
slow wave transmission lines and the 0.18 μm CMOS technology,” in was the Principal Expert of the National 863 Future Project. His research
Proc. IEEE Radio Freq. Integr. Circuits Symp. (RFIC), Atlanta, GA, interests include mobile communications, adaptive signal processing, and
USA, Jun. 2008, pp. 373–376. artificial neural networks with applications to communications and biomedical
[39] C. Geha, C. Nguyen, and J. Silva-Martinez, “A wideband low-power- engineering.
consumption 22–32.5-GHz 0.18-μm BiCMOS active balun-LNA with Dr. You received the Excellent Paper Award from the China Institute of
IM2 cancellation using a transformer-coupled cascode-cascade topol- Communications in 1987 and the Elite Outstanding Young Teacher Award
ogy,” IEEE Trans. Microw. Theory Techn., vol. 65, no. 2, pp. 536–547, from Southeast University in 1990, 1991, and 1993, respectively. He is
Feb. 2017. currently the Chairman of the IEEE Nanjing Section.

Authorized licensed use limited to: University of Exeter. Downloaded on June 06,2020 at 05:52:41 UTC from IEEE Xplore. Restrictions apply.