Download Relevant RF Semiconductor Developments

Relevant RF Semiconductor Developments
The 2009 International Solid-State Circuits Conference included several papers on new
RF transceivers. Papers at this conference generally describe new chips a year or so
before true production availability and often give leading indicators of technology trends.
All the transceiver papers relating to multimode radios referenced the need to reduce the
number of filters. In the past few years, receiver front –end performance has progressed
to eliminate the need for interstage SAW bandpass filters, and the transmitter output
noise-reduction bandpass filters. However, for WCDMA and LTE systems, there is still a
need for the duplex filter to protect the receiver from the transmitter output.
Below are a few block diagrams from papers given at that conference and comments.
Figure A. Block diagram from Paper 6.3 “Single-Chip Multiband
WCDMA/HSDPA/HSUPA/EGPRS Transceiver with Diversity Receiver and 3G DigRF
Interface Without SAW Filters in Transmitter / 3G Receiver Paths” by Skyworks
Solutions
This transceiver appears quite good, minimizing external components, and offers a
DigRF interface. Adding additional bands seems mostly to involve adding external
duplexers and switches.
Figure B. Block diagram from Paper 6.2 “A SAW-Less Multiband WEDGE Receiver” by
authors from ST-NXP and Ericsson Mobile Platforms
This transceiver differs from the previous chip by eliminating diversity and using a
proprietary interface to the baseband section instead of DigRF.
Figure C. Transceiver block diagram from paper 6.4 “Single-Chip RF CMOS
UMTS/EGSM Transceiver with Integrated Receive Diversity and GPS” by Qualcomm
This device falls short on receiver performance, requiring an interstage SAW filter on the
WCDMA receiver path. However, it offers GPS and diversity. It was implemented in a
low-coast 0.18u CMOS technology.
Paper 6.5, “A 45nm Low-Power SAW-less WCDMA Transmit Modulator Using Direct
Quadrature Voltage Modulation”, by an author from NXP Semiconductors, was not a
complete transceiver, but rather an experimental transmit modulator using a voltagemode direct modulator rather than a conventional Gilbert-cell mixer. The goal, apparently
met, was to develop an architecture that would reduce the out-of-band noise and
eliminate the transmit-path SAW filters. This type of circuit appears to be a few years
from commercial deployment, since it was done in a 45nm CMOS process, and the rest
of the transceiver circuitry may not yet be available in such a process.
Figure D. From Paper 22.6, “A CMOS Adaptive Antenna-Impedance-Tuning IC
Operating in the 850MHz-to-2GHz Band”
This paper reported on developments at Arizona State University on adaptive antenna
tuning for cellular terminals. This specific paper described a circuit that would produce
the required control for a Reconfigurable Matching Network. It did not describe the
matching network itself, but noted that control of such networks can consume
considerable power as presently implemented.
The final paper of interest shows the growing focus on solving the “duplexer problem”.
Paper 22.7, “A Tunable Integrated Duplexer with 50dB Isolation in 40nm CMOS”, coauthored by engineers from Broadcom and CMOS RF pioneer Dr. Asad Abidi of the
University of California at Los Angeles, is an attempt to use CMOS technology to
replicate the duplexer function in a scalable technology. The results of the tunable
autotransformer-based technology are impressive as a first step, but as the authors note:
“In the RX band, a rejection of about 25dB is measured, and thus to attenuate
the TX noise further some kind of notch filtering in the TX chain is required.
Alternatively the duplexer notch could be tuned somewhat closer to the RX band,
which still ensures more than 40dB of attenuation in the TX band, as well as
35dB in the RX band as a compromise.”
.
Papers of this nature generally represent technologies that are in the 5-year horizon for
production, if ever, since there are considerable risks involved. However, it is a
technology worth watching closely. In the same way that the shift in receiver technology
to direct-conversion eliminated multiple IF SAW filters and provided a quantum step in
cost reduction, integrated duplex filters could provide the next step forward, especially
for multi-band, multi-mode radios
Comments from Earl
In the last paper on "duplexers", all they really achieved (so far) is
a notching of the TX power at the LNA input. This is enough to
severely reduce (maybe eliminate?!) the RX band filter part of a
duplexer, since the job of that filter is solely to protect the LNA
from the PA. The PA band pass is still needed to suppress the wideband
PA noise from the RX band.
The authors took a fair amount of heat following their presentation
from claiming (in the title) that they have a duplexer, when they don't
really.
One half of a duplexer is not a duplexer and several in the audience
made sure they understood that.