digital signal controller

Document Sample
digital signal controller Powered By Docstoc

Digital Signal Controller:                 d) The Solution to the Coprocessor
Abstract:                                  e) Digital Signal Controller (DSC)

  The Digital Signal Controller (DSC) is   3. Requirements of DSP and MCU
the technology that has emerged to            to be incorporated in DSC
handle many of the engineering
challenges that previously required both   4. DSC Construction (SOC)
Digital Signal Processors (DSP) and
microcontrollers (MCU). The DSC is a       5. Features of DSC
hybrid of DSP’s and MCU’s and              a) Enhanced CPU Functionality
coexists alongside these devices. The      b) Flexible Interrupt Structure
DSC architecture is an innovative hybrid   c) High Resolution Analog to
“system-on-a-chip” (SOC) that                 Digital Converter(ADC)
combines the control oriented attributes   d) High Speed Input Capture
of the Microcontroller and the fast        e) Communication Peripherals
calculation capability of the DSP.            (UART,SPI,12C)
Traditionally, a DSP is used to perform
math-incentive tasks and thus are          6. Digital Signal Controller
“number crunching devices” typically          Application
suitable for high end applications like    a) Engine Knock Detection
telecommunications and audio/video         b) Temperature Sensor
processing which require a lot of          c) PIR Detection
processing in a short amount of time.      d) Vibration Sensors
However they do not have the control       e) Data Loggers
capabilities of Microcontrollers.
Whereas , the Microcontrollers are         7. Advantages of DSC
required to control systems by executing
control algorithms but these too have      8. Future and Present status of
limited functionality. Therefore the          DSC
merging of the DSP and MCU as in DSC
that too in an ‘SOC’ architecture leads    9. Conclusion
to a less cost and powerful technology
development where the disadvantages of
each the DSP and MCU separately are
nullified and also the power
consumption is considerably reduced.

  1. Introduction

   2.   Evolution of the DSC
   a)   Microcontroller
   b)   Digital Signal Processing(DSP)
   c)   Coprocessor Approach
                                             DSP is used to perform such math
                                             incentive tasks. However a DSP by itself
                                             is not very suitable for using sensor
Introduction:                                interface. Therefore merging both the
                                             MCU and the DSP together as in the
         In an increasingly competitive      DSC on a single chip will provide the
market driven by time-to market              solution to the problems faced on the
requirements, designers have been using      individual utilization of the component.
low cost Microcontrollers in their           Thus the DSC is developed as an
embedded systems. Engineers design           innovative hybrid ‘system-on-a-
these processors to control the entire       chip’(SOC) architecture that seamlessly
system including all the supervisory         combine the control features of a 16-bit
tasks , in addition to handling user         MCU with extensive DSP functionality.
interfaces. Consider the data obtained       Stages begin Therefore the advent of
from a sensor which is directly sent to a    the DSP is a breakthrough in areas of
centralizing unit. This might perform        requirement of the DSP and the MCUs
some post-processing or display of the       collectively.
sensor data using discrete hardware
components or digital logic. With the        2.Evolution of the DSC
advent of embedded processor solutions
such as 8-bit Microcontroller (MCU) it              The evolution of the DSC went
gradually became highly advantages (in       through the following stages. The stages
terms of cost and flexibility) to replace    begin with the basic individual
the fixed centralized hardware with          components and then their assembling
MCU that could be programmed to              techniques which are as follows
perform the specific task required for the
application. But as sensor-based             a) Mmicrocontroller (MCU)
applications rapidly increase in
complexity, it becomes imperative to                Microcontroller is a functional
embed a greater degree of intelligence to    computer system on a chip. It contains a
the sensor interface thereby requiring       processor core, memory and
greater computational power, memory          programmable input peripherals. MCUs
and peripheral capability that, in turn      are used in automatically controlled
require 16-bit MCUs.                         products and devices, such as
                 Often, substantial post-    automobile engine control systems,
processing operations may need to be         remote controls, office machines etc. by
performed on the data obtained from the      reducing the size, cost and power
sensor. Such competitions may require a      consumption compared to a design using
high degree of high speed mathematical       a separate micro processor, memory,
functionality in the processer used in the   and input devices, microcontrollers
sensor interface and the average 16-bit      make it economical to electrically
MCU architecture would not suffice.          control many ,more processes.
Thus, the MCUs having limited
processing power could execute only
simple control algorithms thereby,
limiting their functionality. Generally a
                                            1.DSPs do not have flexible interrupt
                                            2.DSPs are not very efficient at
                                            manipulating bits. Ex: individual i/o pin
                                            3.DSPs rely largely on off chip memory
                                            and peripherals.
                                            4.DSPs are rarely available in very low
        Consider , data obtained from a     pin counts.
sensor is directly sent to a centralized    5.They do not have the control
unit which might perform some post-         capabilities of the microcontrolles.
processing operations such as Finite
Impulse Response (FIR), Infinite Impulse    To attain the functionalities and
Response (IIR), filtering of data to        capabilities of both the MCU and the
remove noise, computation of fast           DSP the first approach which was
Fourier Transform to analyze the data       sought for is the coprocessor approach.
so that frequency domain data can be
used in further processing stages. Such     c) Coprocessor Approach
computations require a high degree of                Mathematical functions on array
high speed mathematical functionality in    based data can be accomplished using
the processor used in the sensor            coprocessor. In order to reduce the
interface which MCU architecture would      impact on performance a coprocessor
not suffice. A DSP is used to perform       might be used to off load some of the
math-incentive tasks.                       processing from the microprocessor.
                                            Therefore using the DSP as a
b) Digital Signal Processing (DSP)          coprocessor becomes a natural
                                            progressive step when designing
        DSP processing is becoming          embedded systems requiring signal
prevalent in the microcomputer industry.    processing capabilities. It strictly
The development of multimedia               implemented the control algorithm while
applications which require audio and        the microcontroller managed all other
video processing places greater             tasks. But this too some disadvantages
information processing demands on           1. It has fixed functionality whether
microcomputer systems. DSP algorithms       located to external or incorporated in
have found uses in communication            the microprocessor itself.
control audio and image processing. It is   2. It is not scalable, in other words it
also increasingly finding uses in           cannot process more then one math
embedded applications such as               operation at a time.
telecommunication industry. A DSP is        3. coprocessor typically requires the
primarily concerned with performing         microprocessor to perform all data
mathematical functions on array based       transfer can be a burden on the
data. However a DSP by itself is not        microprocessor .
very suitable for use in a sensor           4. This process is time consuming and
interface for 4 main reasons:               costly.
d) The Solution to the Coprocessor          sensor processing applications.
Problem                                     Currently DSCs are being marketed as
                                            “Green” technologies for their potential
         A single chip that incorporates    to reduce power consumption in electric
the advantages of microcontrollers and      motors and power supplies.
DSPs clearly provide the preferred
solution A DSC can execute both
efficient digital signal processing and a
variety of controller operation in a
single chip. By combining the processing
power of DSP with the programming
simplicity of a microcontroller, a DSC
can provide high speed, low cost and
easy to design solution. As a result
Digital Controllers have taken away         3. Requirement of DSP and MCU to be
some of the microcontroller market          incorporated in DSC
shares because they provide the best of              DSC like microcontrollers and
both worlds.                                DSPs require software support. There is
                                            a growing number of software packages
                                            that offer the features required by both
                                            DSP applications and a microcontroller
                                            applications with a broad set of
                                            requirements software solutions are
                                            more rare. They require development
                                            tools, DSP libraries, and optimization
                                            for DSP processor, fast interrupt
                                            handling, multithreading and a tiny foot
       DSC=MCU+DSP                          print.
                                            1.1st basic requirements of a DSP is a
e) Digital signal controller (DSC)          specific hardware to perform the MAC
        DSC can be thought of as hybrid     operation as well as guard bits in the
of microcontrollers and DSP have            accumulator(result) register that
processors. Like microcontrollers DSCs      prevents the result of a MAC instructions
have fast interrupt responses , offer       from overflowing.
control oriented peripherals like PWMs      2.2nd feature requirement of a DSP is the
and watchdog timers, and are usually        ability to access multiple memory
programmed using the c programming          location simultaneously allowing both
language, although can be programmed        data and instruction to be fetched
using the devices native assemble           simultaneously.
language. On DSP, side they                 3.3rd feature of DSP is ability to form
incorporate features found on most          address and fetch operands without
DSPs such as single-cycle MACs              creating a bottleneck to execution units.
(Multiply and Accumulate).                  4. 4th important hardware element of
                                            DSPs barrel shifters are specific circuits
       DSCs are used in wide range of       used to manipulate register contents
applications but the majority goes into     quickly in order to scale data fixed point
motor control, power conversion and         DSps need to scale data to generate the
same dynamic range available on                         Considering only the very
floating point processor.                      crucial features of the DSC, the DSCs
5.The feature of MCU which sets it a           can be seen as chips that offers the
part from DSC is its general purpose           following features:
capability restricted in its instruction set        Single-cycle multipliers
and addressing modes. An MCU                        Flash memory
includes many control oriented                      ADCs and PWMs
instruction; bit manipulating instruction,
and the ability to manipulate data easily      5. Features of DSC
using the stack pointer.
6. Another feature of MCU is the                   A typical DSC architecture possesses
inclusion of on chip peripherals.              several CPU and peripheral
Although modern DSPs include certain           characteristics that render it suitable for
Peripheral modules such as memory and          using a number of sensor applications.
serial communications controller, the          The most beneficial among these
MCU has usually been the component on          features are
a pc board that soaks of the peripheral
chips around it.                               a) Enhanced CPU Functionality
 7. A final feature that has been
associated more closely with the MCU           1.A true DSC includes two 40-bit
than DSP is fast interrupt handling            accumulators which can be used to store
capability.                                    the results of two independent 16x16 bit
                                               multiplication operations.
4. DSC construction (SOC)                      2.Special instructions such as multiply-
                                               and-accumulate (MAC) provide ability
        SOC (system-on-a-chip) refers to       to multiply 16-bit numbers add the
integrated all              components of a    result in an accumulator and prefetch a
computer or other integrated system into       pair of data values from RAM, all in a
mixed signal and often radio frequency         single instruction cycle.
functions-all on one chip. It is a typical     3.An accumulator width of 40-bits rather
application in the area of embedded            than 32 enables the data to temporarily
systems.                                       overflow.
         SOC is believed to be more cost       4.Moreover the CPU of a DSC can
effective since it is the field of             optionally keep the value with in the
fabrication because its packaging is           permitted range by a mechanism called
simpler. SOC designs usually consumes          ‘saturation’ and also round or scale the
less power and have lower cost and high        data when it is written back to RAM.
reliability then the multi chip system that
they replace and fewer packages in the
systems, assembly costs are reduced as
b) Flexible interrupt structure
                                              e) Communication peripherals
1.The DSC architecture provides an            (UART,SPI,12C)
extremely high degree of flexibility in its
interrupt structure. Usually a large          1.With sensor-based systems becoming
number of individually selectable and         spatially and computationally
prioritizable interrupt sources and           distributed, controller need to
vectors are separated.                        communicate data to one another over
2.Interrupt latency of DSC incorporates       some kind of communications bus or
the low latency feature of MCUs which         network.
generally have relatively few registers in    2.In some applications controller may
their CPU or they have “shadow                need to send data to a PC or logic
registers” ,a duplicate registers set that    analyzer via standard protocols, in order
is only used by the interrupt software.       to log or graphically display data obtain
                                              from sensors.
c) High resolution analog to digital          3.Occasionally a controller might
converter (ADC)                               choose to use an external modem to send
                                              the data over a phone line. For these
1. Using sensors makes it imperative to       purposes DSCs provide a 16-bit
use an on chip ADC of sufficient speed        addressable universal asynchronous
as resolution so as to be measure both        receiver transmitter (UART) interface.
small and rapid changes in the input          4.Frequently, an external digital to
quantity.                                     analog converter (DCA) may be used to
2. ADC supports a variety of sampling,        drive actuators based on sensor
conversion and triggering options.            measurements, or to smart sensors may
3. DSC’s typically contain multiple           need to synchronously communicate with
analogue input channels, flexible             each other. For all such purposes DSCs
sampling/conversion options such as           include one or more 16-bit serial
automatic switching between multiple          peripheral interfaces(SPI).
inputs for successive samples etc.            5.Some peripheral devices on the other
4. To reduce the overhead of reading the      hand communicate using the inter-
converted samples. DSCs provide large         integrated circuit(12C) protocol. For
memory mapped ADC buffers with                example, using 12C an external serial
configurable data formatting                  EEPROM may be used by the DSC to
(integer/fractional , signed /unsigned)       store and access an html web page to
                                              visually monitor and control the
d) High speed input capture                   operation of the sensor subsystem.
         DSCs have the capability of
going into low power modes where in           6.Digital Signal Controller
only an internal power oscillator may be      Applications
active. It can then be awakened by a
control pulse or event on an input                   The following are some sensor
capture pin , which would trigger the         based systems for which the DSCs
start of data collection or computations.     provide an ideal architecture solution.
Input capture channels may also be used
to obtain time stamps for certain events.
a) Engine Knock Detection                      b) Temperature Sensor

        Engine knock detection is placed       1.Of the various kinds of physical
by ignition of unburned air/fuel               quantities measured by sensors,
mixtures, and is a limiting factor in          temperature measurement is perhaps the
vehicular power and fuel economy.              most ubiquitous.
Knock is detected primarily by                 2.Accurately calculating the temperature
measuring either vibrations or pressure        on the basis of measured resistance
inside the engine. Signal processing           requires logarithmic computations and
algorithms such as FFT are utilized to         divisions, for which would be highly
determine the frequency characteristics        efficient. In some configurations,
of these vibrations, thereby making it         differential ADC inputs are required.
imperative to make use of DSCs dual            3.In the case of p-n junction sensor the
accumulators to perform complex                voltage varies from device, making
arithmetic.                                    accurate measurement difficult. In either
        Alternatively, background noise        case, the ability to store and access
and normal frequencies observed when           calibration constants in on-chip data
knock is absent, is filtered out using a       EEPROM or program flash greatly
digital filtering routine such as infinitely   helps in compensating for such
impulse response. The declaration              undesirable effects.
algorithm and integrates the filter output     4.Similarly, mathematical operations are
and compares it with the precipitated          needed to line arise the highly non-
references to determine whether knock          linear output of infrared (IR)
occurred. Again , dual accumulators            temperature detectors.
are ideal, as one accumulator can be
used to perform infinitely impulse             c)PIR Detection
response filtering and the other can be
used to integrate the infinitely impulse       1.PIR detectors are used to detect
response output. Since references              motion in a given area, by sensing
conditions vary very widely from one           differences in thermal radiation between
engine to another across different speeds      a background and a person or object in
an temperatures, and even between              motion. They are widely employed in
different cylinders within the same            security installations.
engine, look up tables are widely utilized     2.Typically, a pair of sensors in a
by the detection algorithm. Additionally,      different ADC input configuration is
a minimum of two sensor(ADC) inputs            used to compensate for interferences,
are often required, as are two timing          thereby preventing false triggering of
signals(via input capture) for                 alarms.
synchronization and providing RPM.
                                               d) Vibration sensors

                                                       Vibration sensors, such as
                                               piezoelectric accelerometers, are
                                               primarily employed to detect wear and
                                               tear, and mechanical faults in machinery
                                               or engines. This is accomplished by
measuring certain frequency
components in the output of the             8. Future and Present Status of DSC
transducer. The technique is susceptible           The emergence of a DSC
to the effect of undesirable noise          category of devices was inevitable, given
frequency as well as harmonics and          the growth in popularity of DSP
therefore requires digital filtering        techniques. The main benefit that this
algorithms such as FIR or IIR, which        hybrid solution offers is the ability to
can be performed very efficiently using a   function as both a signal processing
DSC.                                        engine and a control unit. The
                                            functionality would have previously
e)Data Loggers                              required both an MCU and DSP luxury
                                            for which few designers can budget. In
         The powerful features available    addition to the reduced chip set cost,
in the ADC module of a typical DSC          having only programmable device for
lend themselves ideally to data             which to write software and to design is
acquisition systems. For example,           another significant benefit.
several sensor channels can be sampled
and converted alternately at high speeds,       The industry expects the availability
with the channel-switching and data         of such products will further popularize
buffering occurring automatically. DSP      the signal processing algorithm in
functionality enables post-processing,      embedded control applications and will
filtering and analysis of the data even     drive down the cost of products that
before it is logged. The variety of         have previously required expensive
communication channels available in a       components and lengthy development
DSC can be creatively used to               efforts. Further advanced automobile
communicate with a PC or other data         systems that are likely to implement
loggers, possibly using advanced            such technology in the near future are
protocols such as TCP/IP, email and         collision advanced systems, drowsiness
HTTP.                                       detection systems and integrated chassis
                                            control systems that electronically
7. Advantages of DSC                        integrate steering, breaking suspension
                                            systems. In order to market share, the
1.Architecture requires only one            top DSC vendors are Texas instructions,
software program algorithm that             Free scale and microchip technology,
includes the orientation for both the       according to market research firm
worlds.                                     forward concepts. These three
2.This approach greatly simplifies the      companies dominate the DSC market.
development task and reduces additional
overhead, such as the requirement for       9.Conclusion:
MCU and DSP development tools,
learning curve etc.                              Therefore, the DSC is a highly
3.Reduce chipset size.                      efficient technology which merges both
4.Reduce overall component cost.            MCU and DSP features. The DSP is a
5.Low power consumption.                    single chip combination of both the
6.Reduced logistical controls.              MCU and DSP processor. It provides
7.Increased reliability.                    best of both the technologies and
handles the requirement of both the
MCU and DSP sufficiently. It also
provides tremendous features and
benefits in real time applications. Hence
it is a low cost single substitute to both
its components—Digital Signal
Processor and Microcontroller.


1.Electronics for You-“DSC is an
innovative SOC architecture”.
2.Abstract by ROSSBANNYTINE,
Motorola Semiconductors, Austin.

Shared By: