Transcript Processing of Natural Signals

Processing of Natural Signals
Natural Signals are analog and may be small
• Range: microvolts to hundreds of millivolts
Digital Communications
• E.g. Multi-level signaling rather than binary
Disk Drive Electronics
• Magnetic information converted to electric
signal with high noise needs filtered.
Wireless Receivers
• Antenna signals are 1GHz+ and microvolt
levels plus lots of added noise.
Optical Receivers
Need electrical-to-optical (E2O) and optical
to-electrical (O2E) converters.
Receiver must work fast at low voltage level.
Sensors
Mechanical, electrical, optical sensors are
everywhere and in noisy environments.
E.g. airbag sensors have <1% variation in C.
Microprocessors and Memory
• Modern systems use analog design expertise
for distribution and timing of data and
clocks across a large chip or among chips.
• Memory uses high-speed “sense amplifiers
extensively.
• Nonidealities in signal and power
interconnects on chip and in package.
• High-speed digital design is analog design!
Why is Analog Design Difficult?
Design Trade-Offs
Digital Design Trade-Offs
• Speed
• Power dissipation
Analog Design Trade-Offs
• Speed
• Power dissipation
• Gain
• Precision
• Supply Voltage
• And more!
Why is Analog Design Difficult?
Other Issues
• Analog circuits are more sensitive to noise,
crosstalk, and other interferers than digital.
• Second-order effects in devices affect analog
devices more heavily than digital.
• Analog circuits can rarely be automated requiring
every device to be hand-crafted.
• Modeling and simulation of many effects in
analog circuits are limited. Requires experience
and intuition when analyzing simulation results!
Why is Analog Design Difficult?
The Semiconductor Medium Itself
• Mainstream IC technology is developed and
characterized for digital applications.
• These technologies do not easily lend
themselves to analog design.
• Analog design requires novel circuits and
architectures to achieve a high performance
in this medium.
Why Integrated?
• Conceived in the late 1950s.
• Evolved from a few components to more than one
billion transistors.
• Moore’s Law: Transistor count per chip doubles
every 1.5 years.
• Minimum dimension has dropped from 25um in
1960 to 0.18um in 2000.
• Allows complexity, speed, and precision that would
be impossible with discrete implementations.
Why CMOS for Analog?
• Low cost of fabrication
• Both analog and digital circuits can be on
the same chip.
– Improve overall performance
– Reduce cost of packaging
• Principal reason is device scaling continues
to improve speed of MOSFETs.
Levels of Abstraction
Robust Analog Design
• Device and circuit parameters vary with the
fabrication process, supply voltage, and
ambient temperature (PVT).
• Circuits must be designed for acceptable
performance over a specified range of PVT.
• Difficult since device parameters vary
significantly from wafer to wafer.