CMOS THz Imaging 277
LO Port RF
Antenna
Sample
Under Test
IF Output
Power Supply
Figure 1 2.17: Measurement setup of THz image system.
is comparable to the designs in other receiving topologies [259, 231] when a
0-dBm LO power is applied.
12.3.5 Wide-Band THz Imaging
The THz image system is set up as shown in Figure 12.17 with samples placed
between the horn antenna and receiver chip. The samples under test ar e held
by an X-Y moving stage, controlled by the testing program. The proposed
THz image system shown is applied to study Panadol pills and animal skin
sample in dry and moisturized conditions at 240 GHz and 280 GHz, respec-
tively. The samples are placed between the antenna of the transmitter and
receiver chip. The samples are held by an X-Y moving stage controlled by
the testing program. Figure 12.18 shows two imaging cases for the biomedical
applications. The first case shows tha t one can differentiate b etween a mois-
turized Panadol pill and a dry one because of strong water absorption at THz
frequencies. In the second case, the moisturized area in the animal skin sample
can be clearly identified from the surrounding dry area. Mor eover, different
images are obtained at 240 GHz and 280 GHz with different abso rption ratios.
12.4 280-GHz Reflective I m aging System
Figure 12.19 shows the block diagram of the proposed CMOS-based THz
reflective imaging sys tem. The entire system consists of two identical
transceivers with on-chip antennas tha t each of them can either generate or
detect a THz signal.
When the transc eiver works as a signal source, the signal generated from
278 Design of CMOS Millimeter-Wave and Terahertz Integrated Circuits
240GHz 280GHz
Moisturized Pills
Moisturized Area
Panadol Pills
Animal Skin
Figure 12.18: Captured THz imag e results of Panadol pills and skin
samples un der 240 GHz and 280 GHz radiation.
the CMOS local oscillator (LO) is directly fed to the on-chip antenna by a T-
line network with high characteristic impedance (high-Z
0
). The application of
a high-Z
0
T-line has two advantages compared to the conventional T-line with
50 Ω impedance. Firstly, for the signals traveling in the T-line with the same
power level, hig he r Z
0
generates larger voltage mag nitude, which can drive the
gate of the mixer mo re effectively. Secondly, the signal propagation loss can be
effectively reduced due to a smaller current magnitude in the high-Z
0
T-line.
When the transceiver works as a signal detector, the incoming THz signal
(RF) is firstly received by the on-chip antenna, then it is down-converted by
the single-gate mixers attached to both ends of the T-line networ k. B ecause
both the LO and RF signals travel in opposite directions with 90
◦
phase shift
in the T-line, a differential IF signal results. Note that the LO freque nc y is
usually slightly different from the RF frequency to have an IF output in the
mixer. Compared to the single-ended mixer design introduce d in Sec. 12.3.2,
a differe ntial mixer output in the base-ba nd provide stronger immunity to the
EM interference as well as common noise rejection. After further amplification
by variable ga in amplifier (VGA), both magnitude a nd phase information of
the resulting differential IF signal can be detected and processed for ima ging
applications.
CMOS THz Imaging 279
Figure 12.19: The block diagram of CMOS-based THz reflective imag-
ing system.
In the following se ctions, the design of each building block in the 280-
GHz CMOS transceiver is discussed, including a high-power CON-based signal
source, a high-gain 2D CRLH T-line-based LWA ar ray and a differential down-
conversion mixer with VGA.
12.4.1 Differential Down-Conversion Receiver
12.4.1.1 Diff e rential Down-Conve rsion Mixer with Bidirectional
Hybrid Coupler
Figure 12.20 shows the schematic of the proposed 280 GHz differential down-
conversion mixer. A 90
◦
phase shifter is designed for both LO and RF signals
with high-Z
0
T-lines implemented by coplanar waveguide (CPW) as well as
the parasitics capacitances contributed by M1, M2 and R1 . The siz e of both
M1 and M2 is optimized between the down-conversion efficiency and the ca-
pacitance loadings to the high- Z
0
T-line. For instance, a smaller size of M1
and M2 helps reduce the loaded capacitance to the high-Z
0
T-line, but it also
reduces the output currents of mixing products. Moreover, they are biased in
the subthre shold region (0.4-V V
GS
) by a diode-connected NMOS transistor
280 Design of CMOS Millimeter-Wave and Terahertz Integrated Circuits
Figure 12.20: Schematic of proposed differential down-conversion
mixer with an input network of 90
◦
high-Z
0
T-line.
(M5) to maximize the nonlinearity. Two λ/4 T-line open stubs at 280 GHz
are connected to the drains of M1 and M2 to improve the down-conversion
efficiency by reducing the LO lea kage to the following common-ga te stages
(M3 and M4).
Assuming the frequencies of both RF and LO signals are closed to each
other, both of them have 90
◦
phase shift without any signal loss , and an equal
amount of LO or RF voltag es are applied to the gate of M1 and M2 with a
90
◦
delayed version of each other that can be expresse d as:
V
M1
= V
RF
· cos(ω
RF
t + φ
RF
+ 90
◦
) + V
LO
· cos(ω
LO
t + φ
LO
)
V
M2
= V
RF
· cos(ω
RF
t + φ
RF
) + V
LO
· cos(ω
LO
t + φ
LO
+ 90
◦
)
(12.7)
where [V
RF
, ω
RF
, φ
RF
] and [V
LO
, ω
LO
, φ
LO
] are the [input voltage mag nitude,
frequency, initial phase] of RF and LO signal, r espectively.
Note that the output voltage of a single -gate mixer can be expres sed as
V
IF
(t) = R
0
g
m
(t)V
RF
(t)
g
m
(t) = a
0
+
∞
X
n=1
a
n
cos(nω
LO
t)
(12.8)
where a
n
, n=(0,1,2...) are the Fourier coefficients of g
m
with respect to ω
LO
.
It can be shown that a
0
and a
1
represent the fundamental transconductance
and the first-order mixing product of (ω
RF
- ω
LO
), respectively. As such, by
substituting (12.7) into (12.8), the first-order mixing product at IF outputs
CMOS THz Imaging 281
Figure 12.21: Post-layout simulation results of the high-Z
0
T-line net-
work with 90
◦
phase delay.
become:
V P
IF
= a
1
· V
RF
R
0
cos [ω
IF
t + (φ
RF
− φ
LO
) + 90
◦
]
V N
IF
= a
1
· V
RF
R
0
cos [ω
IF
t + (φ
RF
− φ
LO
) − 90
◦
]
(12.9)
where ω
IF
= ω
RF
− ω
LO
is the IF frequency. A differential IF output is
observed from (12 .9) that V P
IF
and V N
IF
have the same magnitude and
opposite pha se. As a result, the total output voltage of mixer (V
IF
) is
V
IF
= V P
IF
−V N
IF
= 2a
1
·V
RF
R
0
cos [ω
IF
t + (φ
RF
− φ
LO
) + 90
◦
] . (12.10)
Equation (12.10) indicates that both magnitude and phase information of
the RF signal can b e obtained by the proposed down-conversion mixer ,
which are very useful in the refrac tive index measurement by a THz imaging
system.
A post-layout simulation is performed to study the propos ed differential
mixer design. Figure 12.21 shows the 2-port S-parameter analysis of the high-
Z
0
T-line network under 76Ω system impedance. A good input matching is
observed with S11 sma ller than -10dB in 220–340 GHz. The maximum inse r-
tion loss (S21) in 220 ∼ 340 GHz is 1.6dB. A wide -band 90
◦
phase shift is
observed with ±10 degrees bandwidth of 55 GHz centered at 280 GHz. Figure
12.22(a) s hows the post-layout simulation results of the 280-GHz differ ential
mixer from 2 60 GHz to 320 GHz under the following conditions: the power of
LO signal is fixed at 0 dBm; the frequency of the RF signal is 1GHz above
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.