18ECE205J - FPGA-based

Embedded System Design

P.Ramani
Assistant Professor(SG)
SRM IST,Ramapuram
 Unit I-Basics of Peripherals
• Embedded System
• An embedded system, typically engages in “data processing” for the purposes of
data logging and/or control of an external process, system, or systems, by
providing various outputs to external devices, as required, based on the available
input data.
• Embedded systems may be described as dedicated, microcomputer-based
systems, operating in real time, running code optimized for execution
speed and size that are usually designed to perform specific tasks related to
control of a process or system, subject to certain predefined constraints and
operating conditions
Embedded System Architecture
Figure. The VERDAN Computer.
• Comparison of Von Neumann and Harvard
architectures.
• Embedded system microcontrollers, and microprocessors :
Von Neumann, or Harvard, memory architectures

1. The Von Neumann memory configuration has only one “zero”

location for both data and memory and therefore data and program
code must reside in the same memory space.
Since data and addresses each have their own bus in the Harvard
configuration, program and data are stored separately in different
regions of memory to allow instructions to be fetched while data is
being processed and stored/accessed. Instruction address zero is
distinct from data address zero, in the Harvard architecture, which
makes it possible for address and data to have different bit sizes,
e.g., 8 bit data versus 32 bit instructions
Block diagram of the Intel 8048
• The 8048 included an arithmetic section consisting of an
Arithmetic Logic Unit (ALU), accumulator, carry flag and
instruction decoder. The arithmetic logic unit held 8-bit
values obtained from either an accumulator latch or a
temporary register, and produced a result defined by the
instruction decoder.

• The stack was implemented using pairs of registers in the

data memory area. The program counter was an
independent counter, but the Program Counter stack used
pairs of registers, also in data memory. Eight registers R0-
R7, an eight level stack, an optional second register bank
and the data store were all accessible by the ALU
Accumulator operations
 Micro controller
• Atmel
• Cypress Semiconductor
• Infineon Technologies
• Maxim (Dallas Semiconductor),
• NXP, ST Microelectronics, Silicon Laboratories,
• Texas Instruments and Winbond.
8051-Microcontroller uses NMOS Technology
Advanced Microcontroller uses CMOS Technology
 Embedded Systems Applications
• Televisions,
• cable boxes,
• satellite boxes,
• cable modems,
• routers
• printers,
• microwave ovens,
• surround sound systems,
•
 Cont..
• computer monitors,
• digital cameras,
• zoom lenses,
• cars and trucks
• stereos,
• dishwashers,
• dryers,
• washing machines,
•
• cell phones,
• digital multimeters,
• calculators,
• air conditioners,
• mp3 players,
• heaters, flight-control systems
• children’s toys, oscilloscopes, ships, industrial
and process control applications,
 Cont..
• railway systems,
• laboratory equipment,
• personal computers,
• data collection/logging
• equipment, numerical processing
applications,“smart” shoes, robotics,
fire/security alarms
• Biometric systems, proximity detectors,
inertial guidance systems, GPS devices, UAVs
3.Discrete Time Mode (DTM)-
• These systems “wake-up” at predetermined intervals,
sample input data, carry out the appropriate responses
and then go back to ”sleep“.
Example
• A system may be required to “wake-up, respond to
some set of input conditions on an event-driven basis
and the go back to “sleep”.
• Some systems employ “watch dog” functions that in the
absence of the system responding within a pre-
determined period of time, automatically reset
themselves as a way of avoiding the system becoming
“locked-up” because of some anomalous situation, or
malfunction, and subsequently failing to function as
described previously.
• Embedded systems may be implemented as
open, or closed, loop systems
 An “open loop” system,sometimes referred to as a
“feedforward” system, as shown in Figure

•It acquires input information and produces appropriate

outputs based on the acquired inputs, without any ability
to determine whether or not, ultimately, the correct
action, or actions have taken place.

• An open loop embedded system gathers

information,reacts to the input parameter values in a
predefined way and produce the appropriate output
signals/commands,
Example :
•A thermostat senses temperature, (TempSensed), compares the
temperature to a preset value, (TempUpperLimit)

TempUpperLimit < TempSensed

closes some switch contacts to turn on a fan.

•However, this simple system does not know whether the fan is
actually operating, or if it is operating at a speed sufficient to return
the temperature to an acceptable value, within a required period of
time.
• Furthermore, if the (TempSense) exceeds (TempUpperLimit) the
system will continue to attempt to providing cooling, but make no
attempt to take further corrective action.
• A motor is controlled by a sensor and an
embedded system consisting of a
microcontroller and a pulse width modulator
(PWM) driving an amplifier with sufficient
output voltage and current to drive motor.
• In this case the speed of the motor is
determined by the embedded system which
controls the duty-cycle of the pulse train
produced by the PWM.
• Therefore the controller is able to provide
proportional control of the fan by controlling
the average amount of power provided to it.
• Some motors have integral tachometers and/or
Hall effect sensors that can be used to produce
an analog signal that can be returned directly
to the summing junction to produce an error
signal to be processed by the controller and to
confirm that the motor is running at the
appropriate speed.
 closed loop system with direct feedback
•In Figure the system returns a signal, e.g., an analog
voltage/current, or digital data,that reflects the output state
of the system, to the input for comparison.
•In some cases, the embedded system is provided with
input parameters that are presumed to represent the state of
the a system and the controller compares these values that
characterize the current state of the system with predefined
state conditions and makes decisions regarding what steps
must be taken, if any, to bring the system into compliance
with these conditions.
•Other embedded systems utilize a sensor, or sensors,
to determine if the parameter values are consistent
with the desired state of a system and additional
sensors that represent the actual state of the process
or system. In both cases, the input and output sensor
signals are provided to a “summing junction”, or
equivalent, to produce an error signal as shown in
Figure
•In the former case, one sensor is used to establish a
comparison between “setpoint”, i.e, “desired input
parameter value” and another sensor is employed to
determine the output
state, or “actual state”,
• so that the desired versus actual state can be
determined to allow the controller to establish what
error, if any, exists and take such as action, or actions,
as may be required to minimize the resulting “error”
signal at the summing junction.
• Thermostats and cruise controls are example of such
systems.
 Embedded System Performance
Criteria
• Two of the most important considerations for
embedded systems are
1) that they perform each tasks correctly
and
2) that all of the reactions/responses by an
embedded system occur in a timely manner.
• Timeliness, or lack thereof, in the present
context can be characterized as Soft,Firm or
Hard.
Types
• 1. A “Hard Real-Time System” (HRTS) is one
in which failure could produce a catastrophic
result, e.g. failure of a fire alarm system, or a
pacemaker.
2.A “Firm Real-Time System” (FRTS) failure might
be an automotive cruise control for which the
latest value of the current speed is not available in
time for the cruise control algorithm to determine
what, if any, corrective action is required.

• In such cases, the algorithm may be able to use

the previously reported speed and still perform the
necessary operations to maintain relatively
constant speed.
3.A “Soft Real-Time System” error(SRTS) is
exemplified by the failure of an ATM which,
while perhaps inconvenient, is hardly a firm,
or hard failure.
• Failures of these three types (HRTS, FRTS and
SRTS) are usually analyzed in terms of
“deadline misses” and their respective
impacts on the system.
• Latency, in the case of an embedded system,
refers to the delay (t1−t0) between the time
(t0)
• when a condition exists requiring a response
and the time that the response occurs (t1).
• Such delays can arise as the result of hardware
delays in sensors, microcontrollers, peripheral
output devices and software delays produced
by program-execution overhead,
• e.g. program execution speed and interrupt
servicing
Embedded systems employ various techniques to
minimize latency:
•Direct Memory Access (DMA) - this technique,
while not normally involving any preprocessing of
data, allows I/O to occur relatively transparently
without requiring significantCPU overhead.
•I/O devices can transfer data to/from the embedded
system by directly accessing the microcontroller’s
memory space.
•This technique addresses both latency and
bandwidth overhead by allowing the data to be
transferred at a
rate most appropriate for the external device(s) and
whenever it is available
Figure. Block diagram of typical
microcontroller/DMA configuration.
 Looping
• Looping - Since an I/O device can itself be in an
active (“busy”) or waiting (‘idle”) state, an
embedded system can remain in a loop waiting
for a flag to be set, before continuing with
program execution.
• This has the advantage that having the
microcontroller remain in an idle state waiting for
data reduces at least some of the latency.
• But this technique has the limitation that the CPU
is unable to accept data from other sources, while
in this mode.
 Polling
• Polling - Alternatively, an embedded system
can“poll” status flags for I/O devices to determine
if a device:
• 1) has data available for transfer to the
microcontroller,
• 2) is available for receipt of data, or
• 3) is busy.
• Polling may be periodic, or aperiodic, depending
on the application.
 FIFOs (First-In-First-Out buffers)
• FIFOs (First-In-First-Out buffers) and other
forms of buffering can be used with external
devices to buffer I/O until the microcontroller’s
resources are available.
• This technique, as in the case of the use of
DMA, can be employed for occasions when
data is being gathered faster than it can be
handled by the microcontroller
 Interrupts
• Interrupts -interrupt schemes can be
employed that interrupt the CPU only when
I/O needs to occur.
• An interrupt is a request initiated by a device
requesting that the CPU be “interrupted” to
service, i.e., process, some particular task.
• If the device initiating the interrupt has a task
of sufficiently high priority, then
• The interrupt request is accepted,
• interrupts of the same, or lower, priority are
blocked,
• the current task is suspended (which requires
that the state of the CPU23 be fully preserved
• to allow the interrupted task to be completed at
a later time),
• the requested task is processed.
If other interrupt requests of higher priority than that of the original task
exist, they will all be processed before the CPU returns to continue
processing of its original task. If the microcontroller is engaged in a task
and a series of increasingly higher priority interrupt requests occur,
before the preceding interrupt has been fully serviced, then the stack
will contain state information about each of the interrupted tasks that
has been suspended by a higher priority task request and the original
task, except for the highest priority interrupt, which will then be serviced
by an interrupt service routine (ISR).

It is important to fully preserve the state of each lower priority

task on the stack, e.g., by storing the contents of the accumulator,
program counter, program status word and any other registers involved
•The microcontroller can then restore the previous task(s)
and continue program execution until the next interrupt
occurs.

•The overall latency of the embedded system is the

time between when the input data is ready (latched) and
the time at which the microcontroller is able to input the
data, process it and produce the required results

For example,
a hard disk, UART, or device which has “busy” states, the
microcontroller must wait until the device is ready to accept
data/commands, i.e., is in a “non-busy state
 Latency
Total maximum latency (Lmax) for a system can be defined as:

Lmax=Lsensors+Lmicrocontroller+Lperipherals+ Lactuators+ . . .

Lmicrocontroller is a function of program execution times, time required

to service interrupts, wake-up time24, boot time25, etc

• Interrupts introduce delays because of the time required to service a given

interrupt and the fact that they are handled in order of priority. In the worst
case, a low order priority interrupt will have to wait until all higher order
interrupts have been serviced before it is serviced.

• In some applications, as long as the embedded system responds within a

predefined period of time, the system is performing satisfactorily.

•In other cases, different response times are required depending on the
state of the processes being monitored/controlled
•Thus for interrupt-driven I/O, an additional latency factor is “priority” which
determines task precedence. While higher priority tasks, whether input or
output, are addressed earlier than lower priority tasks, in some applications,
all tasks may be assigned the same priority, so that no task takes
precedence, over any other.

•Interrupts also make it possible to detect conditions internal to the

microcontroller such as timer/counter overflow, data available in an internal
UART, an internal UART being available for character transmission, that
a multiplication product is available, etc.

• A microcontroller responds to interrupts by first determining if more than

one interrupt has occurred If so, the microcontroller then services the
interrupts on the basis of priority by halting execution of the current task,
storing all the information required to restore that task and then “servicing”
the interrupt request

•Most microcontrollers have a reserved interrupt referred to as an None-

Maskable-Interrupt (NMI) which has priority over all other interrupt

This interrupt is usually reserved for catastrophic events such as hard

disk, or other serious failures.
Embedded system subsystems
•Microcontrollers need a wide range of subsystems if they are to be the basis for
complex embedded system applications, e.g., voltage A/D and current/voltage
D/A converters, mixers, pulse width modulators (PWMs), programmable gain
amplifiers (PGAs), instrumentation amplifiers, etc., as
shown in Table.

•Embedded systems involve multiple analog/digital data input channels. since

data is often provided by a wide variety of sensors, communications channels,
etc. The embedded system employs a icrocontroller/microprocessor for
numerical computation and logic functions that are to be performed on the input
data, e.g., numerical processing of input data and the decision-making based
thereon. Output drivers for a variety of devices, e.g., motors, linear/rotary
actuators, LCD and other types of display devices, communications devices for
I2C, CAN, SPI, RS23228, etc., are also required as part of the embedded
system and interconnect directly with the microcontroller and the actuators,
displays, PCs, networks, etc.
•Typically, an embedded system involves a combination of analog and digital
devices, under the control of a microcontroller, that collectively serve as a feed
back/control system to monitor, and control, a wide variety of
electromechanical, electro-optical systems, chemical process, etc.

•Embedded systems can be as simple as a fan controller, used to control one,

or more, fans, to maintain pre-defined temperature(s) in a server, or consist of a
complex network of embedded systems collecting and sharing data, as well as,
handling various output/control functions
 Fig. simple example of analog signal
processing (ADC and DAC)

converting analog signals to a digital form before processing gave rise to other
problems, e.g., aliasing. digital filtering overhead
Sampling at a rate below the highest frequency component of a given
signal can give rise to a phenomenon known as “aliasing” as shown in
Figure

Example of aliasing

The Nyquist-Shannon Sampling Theorem, also referred to as the Nyquist

or Shannon criteria, requires that under these conditions the sample rate
be equal to, or greater than, twice the highest frequency component of
the signal, or equivalently, that if the frequency component is B Hertz,
that the sample rate be:
fs = 2βB
 Sensors and Sensing

• Sensors and Sensing Sensors are devices that convert one or more
physical parameters into digital or analog signals for processing and
control applications.

•Such sensors, often referred to as transducers, typically convert

physical parameters such as
•temperature, pressure, linear/curvilinear motion of objects (acceleration,
velocity, displacement, etc.), salinity, hydrogen ion concentration (pH)
wind speed,

• ocean currents, vibration, presence of toxic materials, fire, proximity of

objects,
• force (linear, torque), fluid flow (velocity, acceleration, displacement,
etc.),

•radiation measurement (low frequency RF, high frequency RF,

microwave, ultraviolet, visible light, infrared, etc.), heat flux,
stress/strain, chemical signals (e.g., smells),quasi-static electric and
magnetic fields (i.e. no radiative),

•altitude,metal, resistance, capacitance, inductance, electrical power,

mechanical power,

•mass flow, volume flow, etc., into an analog voltage or current, or the
digital equivalent.
Most Commonly used sensors Like :

• Input sensors include microphones, tachometers,

thermistors/thermocouples, sonar or other forms of acoustic sensors,
pressure sensors, infrared position sensors, mechanical switches of various
configurations, distance (altimeters, ranging sensors, etc.), velocity,
acceleration, roll, yaw, pitch, GPSs, and proximity detectors

• Common output devices include speakers, electric motors/positioners,

wireless connections of various types and protocols, liquid crystal (LCD)
displays, and PC connections.

•Inherent Limitations of Sensor

* Bandwidth and dynamic range …
dynamic range can introduce noise

• Dynamic range for digital signals is defined in terms of the bit error ratio
(BER) which is the ratio defined as:
BER = (number of altered bits)/(number of bits transmitted)
 Types of Sensors
1.Optical Sensors
Various types devices are available for measuring of optical parameters and
the presence/absence of radiation in the optical portion of the electromagnetic
spectrum. These devices

• Photomultipliers - these devices have multiples stages of light amplification

that results in a current that is proportional to the intensity of illumination. These
devices use low work function , e.g., alkali metal-based coatings, to convert
photon impacts into electrons and hence currents.

• Pin Diodes/Photodiodes - these devices are semiconductors, e.g., PN

junctions which produce a current which is proportional to the intensity of the
illumination.
2.Capacitive Sensing
Applications of Capacitive Sensing
*Automobiles,
*Mobile phones,
*A wide range of consumer electronics including home
appliances,
* stereos, televisions,
* A wide variety of consumer products as well as a broad range
of military and industrial applications

•Capacitive sensing is based on a very simple relationship between the area of

a capacitor, the distance between two conducting surfaces of a capacitor, d,
the permittivity of free space, 0, the relative dielectric constant, r and its
capacitance, C, as follows
capacitive sensing techniques are also used to function as “sliders”, proximity
detection, LED dimming, volume controls, motor
controls, etc.

* touch pads and touch

* screens.

•Many capacitive sensing applications consist of a conducting surface often

protected by glass or plastic that senses.

•A typical capacitive sensing arrangement involves two separate conducting

surfaces often created by traces on a printed circuit board which represent a
capacitance of 10 to 30 picofarads. Assuming that the traces are protected by
an insulating material perhaps 1 millimeter in thickness, an approaching finger
represents a capacitance in the range of 1-2 picofarads.
3.Magnetic Sensors
There are various forms of magnetic sensors which operate, in some cases,
by closing or opening switch contacts (reed relays), utilizing the Hall effect to
vary current flow, sensing current flow etc.

They are used in in a wide variety of applications including:

• Brushless DC motors
• Pressure sensors
• Rotary encoders
• Tachometers
• Vibration sensors
• Valve position sensors
• Pulse counters
• Position sensors
• Flow meters
• Shaft position sensors
• Limit switches
• Proximity sensors
Magnetic sensors can be classified in terms of the range of magnetic field
strength which they sense as follows:

• Low Fields - magnetic fields whose strength is less than 1 gauss.

• Earth’s Field - magnetic fields in the range from 1 microgauss to 10 gauss.
• Bias Magnetic Fields - magnetic fields of strength greater than 10 g.

since a magnetic field is a vector field, both magnitude and direction of the field,
a magnetic sensor can use use director, strength or direction and strength to
measure a particular parameter :

•a vector magnetic sensor utilizes both magnitude and direction,

• an omnidirectional magnetic sensor uses magnetic field in one direction,
• a bidirectional magnetic sensor measures magnetic field in both directions,
• a scalar magnetic sensor utilizes magnetic field strength only
Magnetic sensors can be further classified as:

• Anisotropic magneto-resistive (AMR) sensors - these are used for measuring

position in terms of angular, linear position and displacement in fields
comparable to that of the Earth. They consist of thin film resistors that are
created by depositing nickel-iron on silicon whose resistance can be varied by
several percent in the presence of a magnetic field.

• Bias magnetic field sensors - these sensors use Hall devices,

Fluxgate sensors - these sensors are often used in navigation systems.

• Hall effect sensors - these sensors sense current in a small plate as a result
of the Lorentz force F = q(vxB) on electrons. This in turn produces a Hall
voltage which is directly proportional to the magnetic field.
•Magneto-inductive sensors - these sensors utilize a single winding coil which has
a ferromagnetic core in the feedback loop of an operational amplifier to form a
relaxation oscillator. Changes in the the ambient magnetic field alter the frequency
of the oscillator by as much as 100%. A shift in frequency can be detected by a
microcontrollers “capture/compare” functionality.

• Search coil sensors - this sensor relies on the fact that a changing magnetic filed
induces a changing electric field in a coil. However, search coil sensors require
that either the magnetic field is varying or the coil is moving.

• Squid sensors - Based on the Josephson junction is the most sensitive and is
capable of sensing fields as low as 10−15 gauss and as high as 9x104 gauss (9
tesla) which is equivalent to fifteen order of magnitude.
4.RF Sensors

RF sensors detect fluid viscosity, fluid contamination, fluid flow, linear and
rotational speeds, displacement and position in automotive and aeronautical
applications.
They normally operate in ranges from DC to 1 Gigahertz, -170◦C to 1000◦C.

This type of sensor can be used with both ferrous and non-ferrous materials as
well as, glass, plastic, liquids and composites.

This type of sensor measures the magnetic susceptability and electric

permittivity within a predefined volume of space.
Embedded System Processing
Embedded systems are capable of providing several different types of functionality

• Data collection/processing/transmission - since many embedded systems carry

out data processing on data that either began as digital data, or was subsequently
translated into the digital equivalent of an analog signal or signals, microcontrollers
must be capable of carrying out a number of different low level computational tasks
such as addition, subtraction, multiplication and division as well as bit
manipulations, shift operations, bit testing.

• Overflow and underflow handling, array manipulations, together with various

loop and nesting functions. Computation of algorithms such as Fast Fourier
Transforms, digital filtering, etc., are facilitated by special functions such as those
provide by a MAC

•The computational tasks are passed to the DSP or other specialized co-
processor and the results of the computation are then made available to the
microcontroller via shared memory, or other means. DSPs are specialized
microprocessors designed to compute algorithms used in digital imaging, radar,
seismic, sensor array, statistical, communications, biomedical signal processing..
DSPs are specialized microprocessors designed to compute algorithms
used in digital imaging, radar, seismic, sensor array, statistical,
communications, biomedical signal processing. Some examples of such
algorithms are shown in the following Table
Table .Algorithms used in embedded systems.
• Field Programmable Gate Arrays (FPGAs) are also used as co-processors as a
result of the availability of extremely fast ADCs which among other things have made it
possible to apply a variety of digital algorithms to radio frequency (RF) applications.
FPGAs are also capable of supporting parallel processing by employing multiple
CPUs(multicore), Multipliers- Accumulators(MACs) and other special function devices
such as graphics processors, DMA controllers, etc. on a single chip. MACs provide
very high speed multiplications and the have the capability of adding the products to
previous products.

•Microcontrollers with integral MACs are very useful for carrying out multiplications for
which the product is “accumulated” a common requirement for many digital signal
processing algorithms such as vector-dot-products (audio, video, images) Finite
Impulse Response (FIR) and Infinite Impulse Response (IIR) filters, Fast Fourier
Transforms (FFTs), discrete cosine transforms (DCTs), convolution algorithms, etc

For vector-dot-product calculations a typical calculation can be represented by:

Convolution :
y[n] = y[n] + x[i] h[n − i]

matrix multiplication,
•Microcontroller manufacturers have begun to revise some of their standard
microcontroller architectures to allow them to perform one, or more, of the
functions formally assigned to co-processors, or other external hardware

•Systems processes and control algorithms, of the type addressed by

embedded systems, can often be expressed in terms of one, or more, systems
of linear, or in some cases partial differential equations that involve resistance,
capacitance, inductance, OpAmps, etc., such as:
mixed signal design
fundamentals
 Mixed-Signal design fundamentals
• A mixed-signal integrated circuit is any integrated
circuit that has both analog circuits and digital circuits on a
single semiconductor die.
• In real-life applications mixed-signal designs are
everywhere, for example, in smart mobile
phones, sensor systems with on-chip standardized digital
interfaces (like I2C, UART, SPI, CAN etc.),voice related
signal processing, aerospace and space
electronics, IoT, unmanned aerial vehicles (UAVs),
automotive and other electrical vehicles.
• Mixed-signal ICs also process both analog and digital
signals together. For example, an analog-to-digital
converter (ADC) is a typical mixed-signal circuit.
• Mixed-signal circuits or systems are typically cost-effective
solutions for building any modern consumer electronics,
industrial, medical, measurement, space etc. applications.
• A mixed signal integrated circuit may also contain on-chip
memory blocks like OTP, which complicates the
manufacturing compared to analog ICs.
• A mixed signal integrated circuit minimizes off-chip
interconnects between digital and analog functionality in
the system, typically size and weight of the realization due
to minimized packaging and reduced system board
(PWB/PCB, module substrate etc.) size and therefore
increase the reliability of the system.
• Mixed signal devices are available as standard
parts, but custom designed application-specific
integrated circuits (ASICs) are designed for new
applications or if new standards are emerging or
new energy source(s) are implemented in the
system and the production volumes are
estimated to be high.
• Availability of ready and tested analog and mixed
signal IP blocks from foundries or dedicated
design houses have lowered the gap to realize
mixed signal ASICs.
• There also exists few mixed
signal FPGAs and microcontrollers, which typically may
include analog-to-digital and digital-to-analog converter(s),
operational amplifiers, even wireless connectivity blocks
etc. on the same chip with digital logic.
• Mixed signal FPGAs are an extension of field-programmable
analog arrays.
• These mixed signal FPGAs and microcontrollers are bridging
the gap between standard mixed signal devices, full custom
ASICs and possibly embedded software world, when
products are under development or the volumes are too
low for an effective full custom ASIC realization.
• However there can be few performance limitations
typically with these type of FPGAs and microcontrollers
like resolution of ADCs, speed of digital to analog
conversion (DAC) and limited number of inputs and
outputs etc.
• Despite of these possible limitations mixed signal
FPGAs and microcontrollers can speed up the system
architecture design, prototyping and even small and
medium scale production.
• They also can be backed-up with development boards,
development community and possibly also software
support.
• An analog-mixed-signal system-on-a-chip (AMS-SoC) can be
a combination of analog circuits, digital circuits, intrinsic
mixed-signal circuits (like ADCs), and in some cases possibly
also embedded software.
• Integrated circuits (ICs) are generally classified as digital
(e.g. a microprocessor) or analog (e.g. an operational
amplifier). Mixed-signal ICs are chips that contain both
digital and analog circuitry on the same chip.
• This category of chip has grown in volumes dramatically
specifically with the increased use of 3G, 4G, 5G etc. cell
phones and other portable technologies as well as increase
of sensors and electronics in automotives.
• Mixed-signal ICs are often used to convert analog
signals to digital signals so that digital devices can
process them.
• For example, mixed-signal ICs are essential
components for FM tuners in digital products such as
media players, which have digital amplifiers.
• Any analog signal (such as an FM radio transmission, a
light wave or a sound) can be digitized using a very
basic analog-to-digital converter, and the smallest and
most energy efficient of these would be in the form of
mixed-signal ICs.
• Mixed-signal ICs are more difficult to design and
manufacture than analog-only or digital-only
integrated circuits.
• For example, an efficient mixed-signal IC would
have its digital and analog components share a
common power supply.
• However, analog and digital components have
very different power needs and consumption
characteristics that make this a non-trivial goal in
chip design.
• Realization of mixed-signal functionality requires from
manufacturing technologies additionally to traditional
active (various transistors) also options and modelling
understanding to realize well performing passive
elements like coils, capacitors and resistors on the
same chip with digital functionality.

• High voltage transistors might be needed in the power

management functions on the same chip with digital
functionality possibly with a low power CMOS
processor system.
• Some advanced mixed signal technologies may
enable even analog sensor elements like pressure
sensors or imaging diodes on the same chip with
analog digital converter.
• Typically mixed-signal ICs are not necessarily
needing fastest digital performance but need
more mature models of active and passive
elements for more accurate simulations and
verification like for testability planning and
reliability estimations.
• Therefore mixed-signal circuits are typically realized
with larger line widths than highest speed and most
dense digital logic, and the implementation
technologies can be 2-4 nodes behind the latest digital
implementation technologies.
• Additionally mixed signal processing may need passive
elements like resistors, capacitors and coils, which may
require specific metal or dielectric layers or similar.
• Because of these specific requirements the
manufacturers of mixed signal devices foundries can be
different from the digital IC manufacturers or
foundries.
 Examples
• Examples of mixed-signal integrated circuits
include data converters using delta-sigma
modulation, analog-to-digital converter/digital-
to-analog converter using error detection and
correction, and digital radio chips.
• Digitally controlled sound chips are also mixed-
signal circuits. With the advent of cellular
technology and network technology this category
now includes cellular telephone, software
radio, LAN and WAN router integrated circuits.
• Because of the use of both digital signal
processing and analog circuitry, mixed-signal ICs
are usually designed for a very specific purpose
and their design requires a high level of expertise
and careful use of computer aided design (CAD)
tools.
• There also exists specific design tools like mixed
signal simulators or description languages
like VHDL-AMS. Automated testing of the finished
chips can also be challenging
• The particular challenges of mixed signal include:
• CMOS technology is usually optimal for digital performance and
scaling while bipolar transistors are usually optimal for analog
performance, yet until the last decade it has been difficult to either
combine these cost-effectively or to design both analog and digital
in a single technology without serious performance compromises.

• The advent of technologies like high performance CMOS, BiCMOS,

CMOS SOI and SiGe have removed many of the compromises that
previously had to be made.

• Testing functional operation of mixed-signal ICs remains complex,

expensive and often a "one-off" implementation task.
• Systematic design methodologies comparable to digital
design methods are far more primitive in the analog and
mixed-signal arena.
• Analog circuit design can not generally be automated to
nearly the extent that digital circuit design can. Combining
the two technologies multiplies this complication.
• Fast-changing digital signals send noise to sensitive analog
inputs. One path for this noise is substrate coupling.
• A variety of techniques are used to attempt to block or
cancel this noise coupling, such as fully differential
amplifiers, P+ guard-rings,differential topology, on-chip
decoupling, and triple-well isolation.
• Examples of mixed signal FPGAs and
microcontrollers:
– Analog Devices CM4xx Mixed-Signal Control
Processors
– Fusion FPGA (from Microsemi now part of Microchip
Technology)
– PSoC PSoC Programmable System on Chip, product
from Infineon Technologies (former Cypress
Semiconductor)
– Texas Instruments' MSP430
– Xilinx mixed signal FPGA
Programmable logic
• Because of the inherent cost in designing ICs and the
sophistication of the equipment and techniques for
manufacturing them, the most economic way of
producing them is in large quantities.
• However, many IC designs are needed in relatively
small quantities and ideally, an IC should be
manufacturable in small quantities if required but
producible in large quantities if needed.
• This has given rise to a family of programmable logic
devices, as shown in Figure , which can be
economically manufactured in large numbers, but can
also be programmed to provide large numbers of
various relatively low-volume configurations.
Hierarchy of programmable logic
devices.
• PLDs typically consist of large numbers of flip-
flops and gates which can be connected under
software control in arbitrarily complex
configurations to provide specific logic
functionality.
• there are three basic types of PLDs,
• simple PLDs (SPLDs),
• complex PLDs (CPLDs) and
• field programmable PLDs typically referred to as
field programmablegate arrays (FPGAs).
• The programmable logic arrays, which are one
type of PLD, employ fuses which can be
permanently rendered in an open state by special
types of hardware/software programmers.
• The generic array logic (GAL) is a PLD similar to
the PAL except that it is reprogrammable and
they are relatively high-speed devices that are
compatible with both 3.3 and 5 volt logic.
• In addition to array logic, both GALs and PALs
include output logic, e.g., tristate controls
and/or gates that allow the combination of
logic arrays and output logic referred to as a
“macrocell” to be implemented.
• PALs and GALS typically have multiple inputs
and outputs which further increases their
versatility and usefulness.
 Boundary Scanning
• Boundary scanning is a technique that allows
programmable devices to be tested externally, i.e.,
without access to the internal logic.
• Internal registers are provided by the device’s
manufacturer that allow testing of the internal logic and
interconnections.
• However, the device is not aware that such scanning is
taking place and therefore the tests can be carried out
while the device under test (DUT) is operating in an
unperturbed state, or states.
• PSoC3/5 include, within their respective architectures, a
test controller that can be used to access the device’s I/O
pins for boundary testing by employing an internal serial
shift register routed across all of their pins and hence the
name “boundary scan”
• The interface used to control the values in the
boundary scan cells is called the Test Access Port
(TAP) and is commonly known as the JTAG
interface. It consists of three signals:
• (1) Test Data In (TDI),
• (2) Test Data Out (TDO), and
• (3) Test Mode Select (TMS).
• Also included is a clock signal (TCK) that clocks
the other signals. TDI, TMS, and TCK are all inputs
to the device, and TDO is output from the device
Macrocells, Logic Arrays and UDBs
• Combining gate arrays and macrocells provides a
significantly higher level of functionality,
• particularly when the macrocells include
registers, ALUs, flip-flops, etc., than that
obtainable through the use of gates alone.
• One of the simplest configurations of such a
combination consists of a sum-of-products (SoP)
combinatorial logic function and a flip-flop.
• An example of a device employing multiple such
macrocells, as first defined cells.
 UDBs
• UDBs consist of a combination of uncommitted
logic similar to programmable logic devices,
structured logic (datapaths) and a flexible routing
scheme .
• UDB blocks support the following:
• – Universal digital block arrays as large as 64
UDBs.
• – Portions of UDBs can be either chained, or
shared, to enable larger functions.
• – Multiple digital functions supported by the
UDBs include timers, counters,
• PWMs (with dead band generator), UART, SPI,
and CRC generation/checking.
• – Each UDB includes:
• an ALU-based, 8-bit datapath
• Two fine-grained PLDs
• A control and status module
• A clock and reset module
 The Datapath
• The datapath, shown in Figure 5.11, contains
an 8-bit single-cycle ALU, with associated
compare and condition generation circuits.
• A datapath may be chained with datapaths in
neighboring UDBs to achieve higher precision
functions
• Each datapath has two comparators, with bit-
masking options, which can be configured to
select a variety of datapath register inputs for
comparison.
• Other detectable conditions include all zeros, all
ones, and overflow.
• These conditions form the primary datapath
output selects to be routed to the digital
peripheral fabric as outputs, or inputs, to other
functions.
 Input/Output FIFOs
• Each datapath contains two 4-byte FIFOs, that
can be individually configured for direction as
• an input buffer (system bus writes to the FIFO,
datapath internals read the FIFO), or
• an output buffer (datapath internals write to
the FIFO, the system bus reads from the FIFO).
• These FIFOs generate status that can be
routed to interact with sequencers, interrupt,
or DMA requests.
 Chaining
• The datapath can be configured to chain conditions and
signals with neighboring datapaths.
• Shift, carry, capture, and other conditional signals can be
chained to form higher precision arithmetic, shift, and
CRC/PRS functions.
• In applications that are oversampled, or do not need the
highest clock rates, the single ALU block in the datapath can
be efficiently shared with two sets of registers and condition
generators.
• ALU and shift outputs are registered and can be used as inputs
in subsequent cycles.
• Usage examples include support for 16-bit functions in one 8-
bit datapath, or interleaving a CRC generation operation with a
data shift operation
 Datapath Inputs and Outputs
• The datapath has three types of inputs:
configuration, control, and serial/parallel data.
• The configuration inputs select the control store
RAM address.
• The control inputs load the data registers from
the FIFOs and capture accumulator outputs into
the FIFOs.
• Serial data inputs include shift in and carry in.
• A parallel data input port allows up to eight bits
of data to be brought in from routing.
 Datapath Working Registers
• Each datapath module has six, 8-bit working
registers all of which are readable and
writable by the CPU and DMA.
• Accumulator (A0,A1)
• Data (D0,D1)
• FIFOs (F0,F1)
 Datapath ALU
• The Datapath block’s ALU consists of three,
independent, 8-bit, programmable functions that
employ an arithmetic/logic, a shifter unit and a
mask unit.
Carry Functions:
• The carry in option is used in arithmetic
operations. The CI SELA and CI SELB configuration
bits determine the carry in for a given cycle. The
carry out option is a selectable datapath output
and is derived from the currently defined MSB
position, which is statically programmable
• A value of ‘01’ sets the carry mode as registered so that
add with carry and subtract with borrow operations
can be implemented. In this mode, the carry flag
represents the result of the previous cycle.
• A value of ‘10’ sets the routed carry mode for cases in
which the carry is generated somewhere else and
routed to the input allowing controllable counters to
be implemented.
• Finally, the value ‘11’ sets the chainable carry mode
allowing the carry to be chained from the previous
datapath and used to implement single-cycle
operations of higher precision involving two or more
datapaths
 ALU Masking Operations
• An 8-bit mask register in the UDB static
configuration register space defines the
masking operation.
• In this operation, the output of the ALU is
masked (ANDed) with the value in the mask
register.
• A typical use for the ALU mask function is to
implement free-running timers and counters
in “powers-of-two” resolutions.
All Zeros and Ones Detection:
• Each accumulator has dedicated all zeros and
all ones detect capability.
• These conditions are statically chainable as
specified in UDB configuration registers.
 Overflow
• Overflow is defined as the XOR of the carry
into the MSB and the carry out of the MSB.
• The computation is done on the currently
defined MSB as specified by the MSB SEL bits.
• Although this condition is not chainable, the
computation is valid when done in the most
significant datapath of a multi-precision
function, as long as the carry is chained
between blocks.
Shift Operations
• Datapath Chaining:
Each datapath block contains an 8-bit ALU, which
is designed to chain carries, shifted data, capture
triggers, and conditional signals to the nearest
neighbor datapaths to create higher precision
arithmetic functions and shifters
• Datapath and CRC/PRS:
The datapath has special connectivity to allow
cyclic redundancy checking (CRC) and pseudo
random sequence (PRS) generation.
 Compares
• There are two compares, one of which has fixed
sources (Compare 0) and the other has
dynamically selectable sources (Compare 1).
• Each compare has an 8-bit, statically programmed
mask register, which enables the compare to
occur in a specified bit field.
• By default, the masking is off (all bits are
compared) and must be enabled.
• Comparator 1 inputs are dynamically
configurable.
 Dynamic Configuration RAM
(DPARAM)
• Each datapath contains a 16 bit-by-8 word
dynamic configuration RAM.
• The purpose of this RAM is to control the
datapath configuration bits on a cycle-by-cycle
basis, based on the clock selected for that
datapath.
• This RAM has synchronous read and write
ports for purposes of loading the
configuration via the system bus.
 Status and Control Mode
• When operating in status and control mode,
this module functions as a
• status register,
• interrupt mask register, and
• control register
Counter Mode
 Boolean Functions
Boolean algebra provides the operations and rules for working with
this set and forms the foundation for development and use of
digital circuits and for VLSI design.
The set of postulates include:
• closure
• the existence of identity elements for AND (1), and, OR (0) but not
for NOT,
(A + 0 = A, A ∙ 1 = A)
• associative: A + (B+C) = (A+B) +C,
• commutative: A + B = B + A and A ∙ B = B ∙ A,
• distributive: A + (B ∙ C) = (A + B) ∙ (A + C) and A ∙ (B + C) = (A ∙ B) + (A ∙
C),
and,
• inverse: A + A0 = 1 and A ∙ A0 = 0.
• Boolean functions can be expressed as a
• sum-of-products (SoP)
• F = (A ∙ B) + (A ∙ B) (5.1)
• or as a product-of-sums (POS)
• F = (A + B) ∙ (A + B
 Karnaugh Maps
• Karnaugh maps can be used to convert truth
tables and logic equations into logic diagrams
and
• as a substitute for both. In addition, Karnaugh
maps make it possible to simplify logic
diagrams.
 Quine-McCluskey
• The Q-M algorithm offers a number of
advantages:
• There is no limitation on the number of input
variables
• It always finds the “prime implicants”38
• The algorithm can be applied in the form of a
computer program
 Combinatorial Circuits
• A combinatorial circuit is defined as any combination of
the basic operations AND, OR and NOT that includes
both inputs and outputs.
• Each of the outputs is related to a unique function.
• A classic example of a combinatorial circuit is the “half-
adder” which is capable of producing a 1-bit sum, and
carry, based on the following functions:
• Sum = A0 ∙ B + A ∙ B0
• and the resulting carry bit, if any, by
• Carry = A ∙ B
 Sequential Logic
• Unlike combinational logic which has no
internal state and whose output depends
solely on the state of the input, at any given
time, sequential logic output is a function of
its internal state at any given time and inputs.
• Sequential logic is “clock-based” and relies on
a combination of combinational logic and one,
or more, flip-flops as shown in Figure .
 Finite State Machines
• The concept of a finite-state machine (FSM) is
an abstraction of a system whose allowed
states are restricted to only one of a finite
number of states at a time and the
“transitions” those states.
• A transition between states occurs only as a
result of inputs, sometimes referred to as
being event-driven.
• FSMs are defined in terms of the:
• allowed states
• input signals,
• output signals,
• next-state function,
• output function,
and,
• an initial state.
and, as a result, FSMs are sequential machines
Hardware Description Languages (HDL)
• The process of converting from the description of a logic
circuit to an implementation defined in terms of gates is
referred to as synthesis.
• VHDL was originally designed to ease the documentation
challenges of ASIC designs but was soon found to be to
facilitate the design of very high-speed integrated circuits
and is referred to as VHSIC.
• VHDL is a VHSIC hardware description language based on
ADA and developed by the Department of Defense
beginning in 1983.
• Verilog is a hardware description language, similar in some
respects to VHDL, that is used to model electronic circuits
primarily at the register level.
 Warp
• Cypress Semiconductor supports a subset of
Verilog known as Warp. However, there are a
number of significant differences between
Verilog and Warp