What is a DSP chip？

DSP chip, also known as digital signal processor, is a microprocessor with a special structure. The internal structure of the DSP chip adopts a Harvard structure that separates program and data, with dedicated hardware multipliers, widely using pipeline operations, and providing special DSP instructions that can be used to quickly implement various digital signal processing algorithms. According to the requirements of digital signal processing, DSP chips generally have the following main characteristics:
(1) One multiplication and one addition can be completed within one instruction cycle.
(2) The program and data space are separated, allowing for simultaneous access to instructions and data.
(3) The chip has fast RAM, which can usually be accessed simultaneously in two blocks through independent data buses.
(4) Hardware support with low or no overhead loops and jumps.
(5) Fast interrupt handling and hardware I/O support.
(6) Having multiple hardware address generators that operate within a single cycle.
(7) Multiple operations can be executed in parallel.
(8) Support pipeline operations, allowing for overlapping operations such as finger retrieval, decoding, and execution.
Compared with general-purpose microprocessors, DSP chips have relatively weaker other general-purpose functions.
The Development of DSP Chips
The world's first single-chip DSP chip was the S2811 announced by AMI in 1978, and the 2920 commercial programmable chip released by Intel in 1979 was a major milestone for DSP chips. Both of these chips do not have the single cycle chips required for modern DSP chips. In 1980, NEC Corporation of Japan launched the μ PD7720, which was the first commercial DSP chip with a multiplier. The first Japanese company to produce floating-point DSP chips using CMOS technology was Hitachi, which introduced floating-point DSP chips in 1982. In 1983, Fujitsu Corporation of Japan launched the MB8764, which had an instruction cycle of 120ns and dual internal buses, resulting in a significant leap in throughput processing. The first high-performance floating-point DSP chip was the DSP32 launched by AT&T in 1984.
Among so many types of DSP chips, the most successful is a series of products from Texas Instruments (TI) in the United States. TI successfully launched the first generation DSP chip TMS32010 and its series products TMS32011, TMS32C10/C14/C15/C16/C17 in 1982. Subsequently, they successively launched the second generation DSP chips TMS32020, TMSC25/C26/C28, the third generation DSP chips TMS32C30/C31/C32, the fourth generation DSP chips TMS32C40/C44, the fifth generation DSP chips TMS32C50/C52/C53, as well as high-performance DSP chips TMS32C80/C82 that integrate multiple DSPs. Since 1980, DSP chips have made rapid progress and their applications have become increasingly widespread. In terms of computing speed, the MAC (single multiplication and single addition) time has been reduced from 400ns in the early 1980s (such as TMS32010) to 40ns (such as TMS32C40), and the processing power has increased by more than 10 times. The key multiplier components inside the DSP chip have decreased from around 40 in 1980 to below 5, and the on-chip RAM has increased by more than an order of magnitude. From the perspective of manufacturing process, the 4 μ N-channel MOS process was adopted in 1980, while now the sub micron CMOS process is widely used. The number of pins in DSP chips has increased from a maximum of 64 in 1980 to over 200 now, indicating an increase in structural flexibility. In addition, the development of DSP chips has greatly reduced the cost, volume, weight, and power consumption of DSP systems.
Classification of DSP chips
DSP chips can be classified in the following three ways.
1. Classified by basic characteristics
This is classified based on the working clock and instruction type of the DSP chip. If a DSP chip can operate normally at any frequency within a certain clock frequency range without any performance degradation except for changes in computing speed, such DSP chips are generally referred to as static DSP chips.
If there are two or more DSP chips whose instruction sets and corresponding machine code pin structures are compatible with each other, then such DSP chips are called consistent DSP chips.
2. Divided by data format
This is classified based on the data format in which the DSP chip operates. A DSP chip that works in fixed-point format is called a fixed-point DSP chip. A DSP chip that works in floating-point format is called a DSP chip. The floating-point formats used by different floating-point DSP chips are not exactly the same. Some DSP chips use custom floating-point formats, while others use IEEE standard floating-point formats.
3. Classified by purpose
According to the purpose of DSP chips, they can be divided into general-purpose DSP chips and specialized DSP chips. Universal DSP chips are suitable for ordinary DSP applications, such as TI's series of DSP chips. Specialized DSP chips are designed for specific DSP operations and are more suitable for special operations such as digital filtering, convolution, and FFT.
Selection of DSP chips
The selection of DSP chips is a crucial step in designing DSP application systems. Only by selecting a DSP chip can other circuits of the peripheral circuit set system be further designed. In general, the selection of DSP chips should be determined according to the actual application system needs. Generally speaking, when choosing a DSP chip, the following factors should be considered.
1. The computational speed of DSP chips. Computing speed is one of the most important performance indicators for DSP chips and a major factor to consider when choosing a DSP chip. The computing speed of DSP chips can be measured by the following performance indicators:
(1) Instruction cycle. It is the time required to execute an instruction, usually measured in ns. (2) MAC time. That is, the time taken for one multiplication and one addition. (3) FFT execution time. The time required to run an N-point FFT program. （4）MIPS。 Execute millions of instructions per second. （5）MOPS。 Execute millions of operations per second. （6）MFLOPS。 Execute millions of floating-point operations per second. （7）BOPS。 That is, performing one billion operations per second.
2. The price of DSP chips. Determine a DSP chip with an appropriate price based on the actual application situation.
3. Hardware resources of DSP chips.
4. The computational speed of DSP chips.
5. Development tools for DSP chips.
6. Power consumption of DSP chips.
7. Other factors such as packaging form, quality standards, lifecycle, etc.
The computational complexity of DSP application systems is the basis for determining the selection of DSP chips with high processing power. So how to determine the computational complexity of a DSP system to choose a DSP chip?
1. Process according to sample points
Sample processing refers to the DSP algorithm looping through each input sample once. For example; A 256 tap adaptive FIR filter using LMS algorithm, assuming that the calculation of each tap requires 3 MAC cycles, then the calculation of 256 taps requires 256 * 3=768 MAC cycles. If the sampling frequency is 8KHz, that is, the interval between sample points is 125 μ s, and the MAC cycle of the DSP chip is 200 μ s, then 768 cycles require 153.6 μ s of time, which obviously cannot be processed in real time and requires the use of a faster chip.
2. Processing by frame
Some digital signal processing algorithms do not cycle every input sample, but cycle every certain time interval (usually referred to as a frame). So choosing a DSP chip should compare the processing capability of the DSP chip and the computational complexity of the DSP algorithm within a frame. Assuming the instruction cycle of a DSP chip is P (ns) and the time of one frame is Δ τ (ns), the maximum amount of computation provided by the DSP chip in one frame is Δ τ/P instructions.
The basic structure of DSP chips
The basic structure of DSP chips includes:
(1) Harvard structure; (2) Assembly line operation; (3) Specialized hardware multiplier; (4) Special DSP instructions; (5) Fast instruction cycles.
Characteristics of DSP system
The word signal processing system is based on digital signal processing and therefore has all the characteristics of digital processing:
(1) Convenient interface. DSP systems are compatible with other systems or devices based on modern digital technology, and such system interfaces are much easier to implement certain functions than analog systems that interface with these systems.
(2) Programming is convenient. The programmable DSP chip of DSP system enables designers to flexibly and conveniently modify and upgrade software during the development process.
(3) Good stability. The DSP system is based on digital processing and is less affected by environmental temperature and noise, with high reliability.
(4) High precision. The precision achievable by a 16 bit digital system.
(5) Good repeatability. The performance of analog systems is greatly affected by changes in component parameters, while digital systems are basically unaffected. Therefore, digital systems are easy to test, debug, and produce on a large scale.
(6) Easy integration. The digital components in DSP systems have a high degree of standardization, making them easy to integrate on a large scale.
Application of DSP chips
Since the birth of DSP chips, they have experienced rapid development. The rapid development of DSP chips is attributed to both the advancement of integrated circuits and the huge market. In just over a decade, DSP chips have been widely used in many fields such as signal processing, communication, radar, and more. At present, the price of DSP chips is also getting lower and lower, and the performance price ratio is increasing, with enormous potential for application. The main applications of DSP chips are:
(1) Signal processing - such as digital filtering, adaptive filtering, fast Fourier transform, correlation operations, spectral analysis, convolution, etc.
(2) Communication - such as modems, adaptive equalization, data encryption, data compression, slope cancellation, multiplexing, fax, spread spectrum communication, error correction coding, waveform generation, etc.
(3) Speech - such as speech encoding, speech synthesis, speech recognition, speech enhancement, speaker recognition, speaker confirmation, voice mail, voice storage, etc.
(4) Image/Graphics - such as 2D and 3D graphics processing, image compression and transmission, image enhancement, animation, robot vision, etc.
(5) Military - such as secure communication, radar processing, sonar processing, navigation, etc.
(6) Instruments and meters - such as spectrum analysis, function generation, phase-locked loops, seismic processing, etc.
(7) Automatic control - such as engine control, deep space, autonomous driving, robot control, disk control.
(8) Medical care - such as hearing aids, ultrasound equipment, diagnostic tools, patient monitoring, etc.
(9) Household appliances - such as high fidelity speakers, music synthesizers, tone control, toys and games, digital phones/televisions, etc.