Design of digital signal processing applications to choose dsp, what factors should be considered

Texas Instruments is well known, the United States Texas Instruments (Texas Instruments, TI) is the world's most well-known DSP chip manufacturers, the application of its products is also the most widely used, TI's production of the ding MS320 series of DSP chips are widely used in various fields. TI in 1982 successfully launched its first generation of DSP chips TMS32010, which is a milestone in the history of DSP applications, since then, DSP chips began to get a real wide range of applications. Due to the TMS320 series DSP chip has a low price, easy to use and powerful features, so gradually become the most influential and most successful DSP series processor. At present, TI has three major series of products on the market: (1) for digital control, motion control TMS320C2000 series, including TMS320C24x/F24x, TMS320LC240x/LF240x, TMS320C24xA/LF240xA, TMS320C28xx and so on. (2) TMS320C5000 series for low-power, handheld devices, wireless terminal applications, including TMS320C54x, TMS320C54xx, TMS320C55x and so on. (3) for high-performance, multi-functional, complex applications TMS320C6000 series, mainly including TMS320C62xx, TMS320C64xx, TMS320C67xx and so on. U.S. analog device company ADI also has a certain share of the DSP chip market, has launched a series of DSP chips with their own characteristics, its fixed-point DSP chip ADSP2101/2103/2105, ADSP2111/2115, ADSP2126/2162/2164, ADSP2127/2181, ADSP-BF532 and the Blackfin series, floating-point DSP chips are ADSP21000/21020, ADSP21060/21062, and Tiger Shark TS101, TS201S. Motorola Motorola launched a relatively late DSP chip. 1986 the company introduced a fixed-point DSP In 1986, the company introduced a fixed-point DSP processor MC56001; in 1990, it introduced a floating-point DSP chip MC96002 compatible with the IEEE floating-point format, as well as the DSP53611, the 16-bit DSP56800, the 24-bit DSP563XX, and the MSC8101, among other products. Jiel's SC-1000 and SC2000 two series of embedded DSP core, mainly for telecommunications infrastructure, mobile communications, multimedia servers and other emerging applications.DSP chip selection parameters according to the different applications and design goals, the focus of the choice of the DSP chip is also different, and its main parameters include the following aspects: (1) arithmetic velocity First of all, we need to determine the algorithm of digital signal processing, algorithm to determine the amount of operations and completion time will be largely determined, according to the amount of operations and time requirements can be estimated DSP. And its time requirements can be estimated DSP chip computing speed of the lower limit. In the choice of DSP chip, the main measures of the chip's computing speed: MIPS (Millions of InstructionsPer Second), millions of instructions / second, the general DSP for 20-100M IPS, the use of ultra-long instruction word TMS320B2XX for 2400M IPS. It must be noted that this is a fixed-point DSP chip computing speed measurement indicators, it should be noted that manufacturers provide the indicators generally refers to the peak indicators, therefore, the system design should be left with a certain margin. ?MOPS (Millions of OperationsPer Second), performing millions of operations per second. The question of this indicator is what is an operation, usually the operation includes CPU operations in addition to address calculation, DMA access data transfer, I/0 operations and so on. Generally speaking, the higher the MOPS means the faster the multiply-accumulate-sum operation is. MOPS can provide a comprehensive description of the performance of the DSP chip. MFLOPS (Million Floating PointOperations Per Second), million floating point operations / second, which is a measure of floating point DSP chip. Key indicators. For example, TMS320C31 in the main frequency of 40MHz, the processing capacity of 40MFLOPS, TMS320C6701 in the instruction cycle of 6ns, single-precision operations up to 1GFLOPS . Floating-point operations include floating-point multiplication, addition, subtraction, storage and other operations. It should be noted that the manufacturers provide this indicator generally refers to the peak indicator, therefore, the system design should be careful to leave a certain margin. MBPS (Million Bit Per Second), which is a measure of the bus and I / 0 port data throughput rate, that is, the bandwidth of a bus or I / 0. For example, for the TMS320C6XXX, 200MHz clock, 32bit bus, bus data throughput rate of 800Mbyte/s or 6400MBPS. ACS (Multiply-AccumulatesPer Second), such as the TMS320C6XXX multiply acceleration up to 300MMACS-600MMACS. 600MMACS. Instruction Cycle, the time required to execute an instruction, usually in ns (nanoseconds), such as the TMS320LC549-80 in the main frequency of 80MHz is the instruction cycle of 12.5ns. ?MAC time, the time it takes to perform a multiplication and addition operation: most DSP chips can complete a MAC operation in one instruction cycle. ?FFT/FIR execution time, the operation time to run an N-point FFT or N-point FIR program. Since FFT operation/FIR operation is a typical algorithm for digital signal processing, this metric can be used as a comprehensive indicator of chip performance. Table 1 is a comparison of some DSP chips based on some of the above parameters. (2) Arithmetic precision: Generally, the arithmetic precision of floating-point DSP chips is higher than that of fixed-point DSP chips, but the power consumption and price also rise. General fixed-point DSP chip word length of 16-bit, 24-bit or 32-bit, floating-point chip word length of 32-bit. Accumulators are generally 32-bit or 40-bit. Fixed-point DSPs are characterized by high main frequency, high speed, low cost and low power consumption, and are mainly used in the fields of control, communication, voice/image, consumer electronics, etc., where the computational complexity is not high. Usually problems that can be solved with fixed-point devices, try to use fixed-point devices, because it is economical, fast, low cost, and low power consumption. However, when programming, we should pay attention to the dynamic range of the signal, and add the fixed-point operation to limit the dynamic range of the signal in the code. Although we can improve the operation accuracy by improving the algorithm, this will increase the complexity of the program and the amount of operations accordingly. Floating-point DSP speed is generally lower than fixed-point DSP processing speed, its cost and power consumption are higher than fixed-point DSP, but because of its floating-point data format, and thus the processing accuracy, dynamic range are much higher than fixed-point DSP, suitable for high arithmetic complexity, high precision requirements of the application; even if it is a general application, in the programming of floating-point DSP, do not have to consider the overflow of data and the accuracy of the problem, and thus programming than fixed-point DSP. Even for general applications, when programming a floating-point DSP, there is no need to consider data overflow and lack of precision, thus programming is more convenient and easier than fixed-point DSP. Therefore, arithmetic precision requirements is a compromise, need to be based on experience to determine the optimal combination of a point. (3) Selection of word length: Generally, floating-point DSP chips use 32-bit data words, and most fixed-point DSP chips are 16-bit data words. And Motorola fixed-point chips use 24-bit data words in order to achieve a compromise between fixed-point and floating-point accuracy. Word length size is an important factor affecting the cost, which affects the size of the chip, the number of pins and the size of the memory, the design in order to meet the performance indicators under the conditions of the smallest possible data word. (4) memory and other on-chip hardware resource arrangements: including the size of the memory, the number of on-chip memory, bus addressing space. The size of the chip memory determines the chip operating speed and cost, for example, TI company with a series of DSP chips, different kinds of chip memory configuration and other hardware resources vary. Through a careful analysis of the algorithmic program and application goals can be determined by the general requirements of the DSP chip on-chip resources. Several important considerations are the number of on-chip RAM and ROM, whether or not to expand the memory, the bus interface / interrupt / serial port is sufficient, whether or not with A/D conversion. (5) development and debugging tools: perfect, convenient development tools and related support software is the development of large-scale, complex DSP system requirements, to shorten the product development cycle has a very important role. Development tools include software and hardware two parts. Software development tools mainly include: C compiler, assembler, linker, program libraries, software simulators, etc., after determining the DSP algorithm, the program code written through the software simulator for simulation to determine the necessary performance indicators. Hardware development tools include in-circuit hardware emulators and system development boards. The online hardware emulator is usually a JTAG peripheral scanning interface board, which allows online debugging of the designed hardware; before the completion of the hardware system, different functions of the development board to run in real time on the designed DSP software, which can improve the development efficiency. Even in some small number of products, the development board directly as the final product. (6) power consumption and power management: generally speaking, personal digital products, portable devices and outdoor equipment, etc. have special requirements for power consumption, so this is also a problem that should be considered. It usually includes the choice of supply voltage and power management functions. The power supply voltage is usually low, the implementation of the chip's low voltage power supply, usually 3.3V, 25V, I8V, 0.9V, etc., in the same clock frequency, their power consumption will be far lower than the 5V power supply voltage of the chip. After strengthening the management of the power supply, usually with hibernation, wait mode, etc. to save power consumption. TI, for example, provides detailed, functionality with the type of instruction and processor configuration of the application notes. (7) price and manufacturers of after-sales service factors: the price includes the price of the DSP chip and development. The price of tools. If the use of expensive DSP chips, even if the performance is high, the scope of its application is also certainly subject to certain limitations. However, the low-priced chip is bound to be less functional, less on-chip memory, poorer performance, which brings certain difficulties in programming. Therefore, according to the application of the actual system, to determine a moderately priced DSP chip. Also give full consideration to the manufacturer to provide after-sales service and other factors, good after-sales technical support is also an important resource in the development process (8) Other factors: including the DSP chip package form, environmental requirements, supply cycle, life cycle, etc. DSP application selection type example for digital control, motion control DSP system development of DSP chip selection for digital control, motion control main. To have disk drive control, engine control, laser printer control, airbrush control, motor control, power system control, robot control, high-precision servo system control, CNC machine tools and so on. Of course, these are mainly for the application of digital motion control system design, in the control of these systems, not only require peripheral circuits dedicated to digital control systems, but also require the chip to have the general characteristics of a digital signal processor. For example, in the DSP control system for controlling a DC brushless motor, the DC brushless motor operation process is subject to two kinds of control, one is speed control, that is, control of the current supplied to the stator coil; the other is phase change control, in which the rotor reaches a specified position to change the stator on-phase, realizing a change in the stator magnetic field, and this kind of control actually realizes the mechanism of the physical brushes. Therefore, this kind of motor needs to have position feedback mechanism, such as Hall element, photoelectric code disk, or using trapezoidal reverse electromotive force characteristics of reverse electromotive force over-zero detection, and so on. Motor speed control is also based on the position feedback signal, calculated rotor speed, and then use PI or PID and other control methods, real-time adjustment of PWM duty cycle, etc. to achieve stator current regulation. Therefore, the control chip to carry out more computational processes. Of course, there are also specialized DC brushless motor control chip; but in general, in most applications, in addition to motor control, always need to do some other control and communication and other things, so the choice of a corpse with WIVI, but also has a strong mathematical function of the chip is also a good choice. Motorola's Digital Signal Processor DSP568xx series integrates a general-purpose digital signal processor fast computing functions and the stator current regulation. Motorola's digital signal processor DSP568xx series integrates the general-purpose digital signal processor fast arithmetic functions and microcontroller peripheral rich features, making the series particularly suitable for those who require strong data processing capabilities, but also to have more control functions of the application, the control of the DC brushless motor is a series of DSP is one of the typical applications. In addition, the field of digital motion control and TI's Ding M S320C24x series, TMS320Lx240xx series, in particular, TMS320LF2407A in the control of a very wide range of applications, TMS320LF2407A as a digital control system optimized specifically for the optimization of general-purpose programmable microprocessors, not only has a low-power and Code confidentiality features, and it integrates a very strong digital signal processing capabilities, but also integrates the digital control system necessary inputs, outputs, A/D conversion, event capture and other peripherals, its clock frequency of 40MHz, the instruction cycle is less than 50ns, the use of improved Harvard structure and pipeline technology, in an instruction cycle can be executed several instructions. From the point of view of running speed, accuracy of the above two chips are similar, but the TMS320LF2407A debugging and development environment (CCCS) is more mature, can be referred to a wealth of information, which will undoubtedly reduce the development cycle. For low-power, handheld devices, wireless terminal applications, DSP chip selection C54X, C54XX, C55X compared to a series of other main features is low-power, so the most suitable for personal and portable Internet access as well as wireless communications applications, such as cell phones, PDA, GPS and other applications. The processing speed is between 80---400M IPS. C54XX and C55XX generally only have McBSP synchronous serial port, HPI parallel interface, timer, DMA and other peripherals. It is worth noting that the C55XX provides an EMIF external memory expansion interface, which allows direct use of SDRAM, while the C54XX does not.