IEEE JOURNAL OF SOLID-STATE CIRCUITS, VOL. 41, NO. 12, DECEMBER 2006 2807
A 77-GHz Phased-Array Transceiver With
On-Chip Antennas in Silicon: Transmitter
and Local LO-Path Phase Shifting
Arun Natarajan, Student Member, IEEE, Abbas Komijani, Student Member, IEEE, Xiang Guan, Member, IEEE,
Aydin Babakhani, Student Member, IEEE, and Ali Hajimiri, Member, IEEE
Abstract—Integration of mm-wave multiple-antenna systems on
silicon-based processes enables complex, low-cost systems for high-
frequency communication and sensing applications. In this paper,
the transmitter and LO-path phase-shifting sections of the first
fully integrated 77-GHz phased-array transceiver are presented.
The SiGe transceiver utilizes a local LO-path phase-shifting archi-
tecture to achieve beam steering and includes four transmit and
receive elements, along with the LO frequency generation and dis-
tribution circuitry. The local LO-path phase-shifting scheme en-
ables a robust distribution network that scales well with increasing
frequency and/or number of elements while providing high-res-
olution phase shifts. Each element of the heterodyne transmitter
generates
+
12.5 dBm of output power at 77 GHz with a band-
width of 2.5 GHz leading to a 4-element effective isotropic radi-
ated power (EIRP) of 24.5 dBm. Each on-chip PA has a maximum
saturated power of
+
17.5 dBm at 77 GHz. The phased-array per-
formance is measured using an internal test option and achieves
12-dB peak-to-null ratio with two transmit and receive elements
active.
Index Terms—Integrated circuits, LO-path, mm-wave, multiple
antenna, phase interpolation, phase rotator, phase shifters, phased
array, power amplifier, radar, SiGe, transceiver, transmitter.
I. INTRODUCTION
I
NTEGRATED mm-wave systems are the next step in the
continuous effort to extend the advantages of silicon-based
integration to yet higher frequencies. This move towards
mm-wave frequencies is spurred by two forces—first, the
need to lower system cost and improve system performance
for potentially widespread sensing applications like 24-GHz
and 77-GHz vehicular radar [1]–[5] and second, the desire to
achieve high data rates by leveraging the larger bandwidths
available at higher frequencies such as 24 GHz and 60 GHz [6],
[7], [9]. Silicon integration at these frequencies brings several
benefits with it such as minimal incremental cost of devices,
and short on-chip interconnects which enable the realization of
complex architectures that are tailored for particular mm-wave
sensing and communication applications. As expected, a
major challenge with such integration is efficient high-power
generation in silicon at mm-wave frequencies as the device
scaling that makes devices faster also leads to a reduction in
breakdown voltages [10]. One possible method of addressing
Manuscript received May 1, 2006; revised September 1, 2006.
The authors are with the California Institute of Technology, Pasadena, CA
91125 USA (e-mail: arun@caltech.edu).
Digital Object Identifier 10.1109/JSSC.2006.884817
this challenge is to combine output power from several devices
through various combining methods [11], [12]. In the case of
wireless applications, one of the more efficient and cost-effec-
tive ways to achieve this power combining is to perform it in
air, i.e, through spatial beamforming methods using multiple
antennas. However, in order for this spatial power combining to
be flexible enough for radar and communication applications,
the beam needs to be steerable in space.
Phased arrays are a special class of multiple antenna systems
that provide a well-known solution to the requirement of elec-
tronic beam steering. In addition to providing beamforming and
beam steering capabilities, phased-arrays provide higher effec-
tive isotropic radiated power (EIRP) in the transmitter and lower
noise figure in the receiver [7], [13]. While phased arrays in-
crease transmit EIRP and improve system SNR, they present a
new challenge, namely, the need for integrated electronic phase
shifting to achieve coherent signal combining in the desired
direction.
The above-mentioned benefits of phased arrays have led to
increasing interest in integrated phase shifters [14]–[16], as well
as demonstrations of integrated phased-array transmitters and
receivers [8], [17], [18]. Though LO-path phase-shifting was
presented as a viable candidate for integrated phased-arrays in
[8], a direct attempt to extend the centralized architecture em-
ployed in [8] to a system that integrates an entire transceiver at
a higher frequency becomes extremely difficult due to the need
for extensive buffering and large silicon area. Therefore, in this
work, we utilize a local LO-path phase-shifting scheme to re-
alize the first fully integrated 77-GHz phased-array transceiver
implemented in a SiGe process. In this paper, we focus on the
transmitter and the phase-shift architecture while the receiver
and the on-chip dipole antennas are discussed in greater depth
in the companion paper [33]. To the authors’ best knowledge,
the complete integration of four transmit and receive elements,
along with the frequency generation and phase-shifting cir-
cuitry, in this transceiver represents the highest levels of silicon
integration achieved at mm-wave frequencies [19], [20], [33].
In the following sections, the design and architecture of
the transmitter and LO-path phase-shifting sections of the
phased-array transceiver will be discussed in greater detail.
Section II provides a brief overview of the spectrum around
77 GHz. Section III focuses on the transmitter and the local
LO-path phase-shifting architecture, while the circuits in the
signal and LO-path are discussed in Section IV. The measure-
ment results are presented in Section V.
0018-9200/$20.00 © 2006 IEEE
2808 IEEE JOURNAL OF SOLID-STATE CIRCUITS, VOL. 41, NO. 12, DECEMBER 2006
Fig. 1. Phased-array transmitter. (a) Operating principle of phased-array transmitter. (b) Improvement in EIRP in phased-array transmitter.
II. THE SPECTRUM AT 77 GHz
Collision-avoidance radar systems on vehicles are expected
to play a major role in reducing accidents and improving auto-
motive safety. The uses of mm-wave sensors are, however, not
restricted to automobiles as there are numerous military and
commercial sensing applications that require high-frequency
sensors [22].
In the case of automotive radar, efforts are underway to
utilize these systems for blindspot detection, adaptive cruise
control and collision-warning applications [21]. Though vehic-
ular radar systems in the near future are permitted to operate
at 24 GHz, these systems will have to transition to 77 GHz
due to concern over frequency bands near 24 GHz that are
used for sensitive measurements in astronomy [23]–[25]. While
adaptive cruise control (ACC) systems require beams to have 3
beamwidths and a scanning range of 8 in the azimuth, larger
scanning ranges would be necessary for collision-avoidance
applications [26], [27]. A beamwidth of 3
in the azimuth and
elevation planes calls for 36-dBi system directivity. Splitting
this directivity evenly between the transmitter and receiver leads
to a 18-dBi directivity requirement in the transmitter. A single
transmitter with a highly directive antenna limits the scanning
range whereas in a phased array the required directivity can be
partitioned between the array gain and the directivity of the an-
tennas, thereby permitting beam scanning. For a phased-array
with
elements, the directivity is dBi, which trans-
lates to 12-dBi directivity for a 16-element array. Such a phased
array, coupled with planar antennas with directivity 6 dB, leads
to sufficient system directivity. If each element in this system
Fig. 2. Centralized LO-path phase-shifting architecture.
can transmit 12 dBm, the EIRP for the entire array with the
antenna is 42 dBm. The EIRP can be further increased by
increasing the antenna gain in the elevation plane, as the gain in
the elevation plane does not affect the scanning in the azimuthal
plane.
III. S
YSTEM
ARCHITECTURE
This section provides a brief introduction to phased-array
transmitters followed by a description of the transmitter and
the local LO-path phase-shifting architecture employed in the
transceiver.
A. Phased-Array Transmitter Overview
A phased array is a multiple-antenna system in which beam-
forming is achieved by varying the relative phase shifts in each
element. As shown in Fig. 1(a), for a certain phase shift setting
in each element of an
-element phased-array transmitter, the
NATARAJAN et al.: PHASED-ARRAY TRANSCEIVER WITH ON-CHIP ANTENNAS IN SILICON: TRANSMITTER AND LOCAL LO-PATH PHASE SHIFTING 2809
Fig. 3. Local LO-path phase-shifting architecture.
signals from all elements add up coherently in one direction and
incoherently in other directions leading to formation of a beam.
Electronic phase-shifting enables beam steering, eliminating the
need for any moving mechanical components. An important ad-
vantage arising from the coherent combining of signals in an
-element phased-array transmitter is the improvement in EIRP
by
dB, as shown in the example in Fig. 1(b).
B. Local LO-Path Phase Shifting Architecture
The phase-shifting capability required in each element of a
phased-array transmitter can be implemented in myriad ways.
The tradeoffs of implementing the phase-shift at different points
in the transmit or receive chain have been discussed in [8].
Phase-shifting in the LO-path is considered advantageous since
the circuits in the LO-path operate in saturation and therefore it
is relatively simple to ensure that the gain of each element does
not vary with the phase shift setting. Additionally, the require-
ments on phase-shifter linearity, noise figure and bandwidth are
substantially reduced when LO-path phase shifting is adopted.
Earlier integrated phased-array receiver and transmitter de-
signs [7], [17] introduced a centralized LO-path phase-shifting
scheme in which an
-phase voltage-controlled oscillator
(VCO) generates multiple phases of the LO, as shown in Fig. 2.
These multiple phases are then distributed to the phase selector
in each element which selects the appropriate phase of the
LO for the desired beam direction. One limitation with this
approach stems from the fact that the phase resolution is limited
by the number of phases generated by the VCO. Another lim-
itation, more important at mm-wave frequencies, arises from
the necessity to distribute all the LO phases to each element
since the distribution of a large number of LO phases precludes
a power-matched, buffered LO-phase distribution network
with transmission-line (t-line) interconnects and matched LO
buffers. As a result, the centralized scheme is unsuitable for
an array operating at high frequencies and/or having a large
number of elements as such arrays would require a larger
distribution network with intermediate buffering.
The above-mentioned limitations of the centralized
phase-shifting scheme dictated the move to the local LO-path
phase-shifting architecture adopted in this system. In this
Fig. 4. Simulated closed-loop phase noise of 50-GHz frequency synthesizer.
architecture (shown in Fig. 3), the output of a single-phase
VCO is distributed to the phase rotator in each element
through a buffered binary-tree distribution network. The use
of power-matched buffers ensures an LO signal with sufficient
amplitude at the phase rotator input. The phase rotator in each
element generates the LO quadrature phase locally and then
interpolates between the in-phase (I) and quadrature-phase (Q)
LO signals to generate the desired phase shift in each element.
From the detailed description of the phase rotator circuitry,
presented in Section IV, it can be seen the phase shift resolution
in this approach depends primarily upon the resolution of inter-
polator weights which can be generated with high-resolution by
DACs. This increased resolution can also be used to improve
phase matching between different elements through calibration
procedures.
In addition to the resolution of the weights, the resolution
of an LO-path phase-shifting architecture is also limited by the
phase noise of the LO signal as the phase setting in each element
is affected by LO phase noise which translates to jitter in the
beam direction. An estimate of this degradation can be obtained
from a sample 50-GHz synthesizer phase noise plot shown in
Fig. 4. The output of the phase rotator is a weighted combina-
tion of the LO signal and a delayed version of the LO signal.
2810 IEEE JOURNAL OF SOLID-STATE CIRCUITS, VOL. 41, NO. 12, DECEMBER 2006
Fig. 5. 77-GHz phased-array transmitter architecture.
Ignoring the effect on phase noise of the correlation induced by
this weighted combination, the rms jitter in the phase setting is
given in radians by
(1)
where L(f) is the phase noise of the closed loop synthesizer in
dBc/Hz. For the sample synthesizer plot in Fig. 4, with
Hz, the rms jitter in the phase setting in each element is 2.1 .
C. Transmitter Architecture
The 77-GHz phased-array transmitter has four elements and
is a part of a fully integrated 77-GHz four-element phased-array
transceiver. The LO frequency generation circuits are shared be-
tween the receiver (RX) and the transmitter (TX). It is important
to note that each transmit and receive element includes an inde-
pendent phase rotator that applies the desired phase of the LO
to the upconversion or downconversion mixer in that element.
The architecture of the transmitter is shown in Fig. 5. The
transmitter utilizes a two-step upconversion scheme with an IF
frequency of 26 GHz. The on-chip VCO generates the 52-GHz
LO signal necessary for the second upconversion in the TX
(and for the first downconversion in RX) while the quadrature
26-GHz signal required for the first upconversion (and second
downconversion in RX) is provided by a quadrature injection-
locked divide-by-two following the VCO.
In the transmit signal path, the baseband signals are upcon-
verted to 26 GHz by a pair of quadrature upconversion mixers.
The signal distribution to all the elements is done at IF through
a network of distribution amplifiers. The RF mixer in each ele-
ment upconverts the 26-GHz input signal to 77 GHz, providing
the input for a driver that feeds on-chip 77-GHz power ampli-
fiers. The adopted frequency plan leads to the undesired product
of the second upconversion falling at 26 GHz while the RF is
at 77 GHz. The tuned mixer, driver, and power amplifier (PA)
stages provide sufficient attenuation at 26 GHz, therefore the
second upconversion does not employ quadrature upconversion.
In the LO-path, the output of the differential, crosscoupled
52-GHz VCO is distributed to the phase rotators in each ele-
ment through a symmetric network of distribution buffers that
ensures that the phase of the LO signal is the same at the input
of the phase rotator in all transmit elements. A cascade of di-
vide-by-two frequency divider blocks following the VCO gen-
erate the 50-MHz signal that is used by an off-chip PFD to lock
the VCO.
IV. C
IRCUITS IN TRANSMIT AND LO PAT HS
A. IF Stage
The baseband to IF upconversion is achieved using
Gilbert-type quadrature upconversion mixers with a shorted
t-line as load (Fig. 6). As the mixers drive a pair of IF dis-
tribution buffers that are input-matched to 100-
differential,
the output impedance of the mixers is designed to be 50-
differential to provide maximum power into the distribution
network. The mixers and the buffers draw 46 mA from a 2.5-V
supply.
The multiple elements present on the same die in an inte-
grated phased array result in on-chip interconnects that are up
to 1.5 mm in length. At the LO frequency of 52 GHz, this rep-
resents 0.52
. Therefore, interconnect modelling is critical, and
adoption of t-line based interconnects that are easily and reli-
ably modeled dramatically simplifies the design effort.The t-line
structure adopted in this system is shown in Fig. 6. The presence
of the ground shield improves the isolation between adjacent
NATARAJAN et al.: PHASED-ARRAY TRANSCEIVER WITH ON-CHIP ANTENNAS IN SILICON: TRANSMITTER AND LOCAL LO-PATH PHASE SHIFTING 2811
Fig. 6. IF stage.
Fig. 7. RF stage.
t-lines by 20 dB, which is important given the number of signal
t-lines in the integrated transceiver.
While t-lines simplify modelling, the design is still heavily
floorplan dependent as the load impedances are a strong func-
tion of interconnect length. This dependency can be eliminated
by conjugate-matching each circuit block at the input and
output to the t-line interconnects, thereby ensuring that the load
impedances are independent of the floorplanning. However, it
must be noted that this separation of design and floorplanning
is achieved at the cost of bandwidth. For a simple shunt-series
t-line matching network, the bandwidth depends upon the
transformation ratio which can be high, as the range of im-
pedances achievable with on-chip t-lines for reasonable layout
parameters and acceptable loss is limited to 65
. While this
reduction in bandwidth can be desirable for some applications,
it can pose a problems for broadband applications. This chal-
lenge can be overcome by reducing the quality factor (
) of the
tuned loads or by using higher order matching networks [28].
In the transmitter, the
of the tuned loads was chosen such
that the system had a bandwidth of 2.5 GHz. Furthermore, the
Fig. 8. Simulated power-match at RF mixer output under small-signal
conditions.
characteristic impedance of t-line interconnects was generally
chosen to be 50
to allow for probe-based measurements at
internal test points.
