Download the full version of Guide: http://www.armbbs.cn/forum.php?mod=viewthread&tid=94547
Chapter 1 Digital Signal Processing beginner preparations
This issue began to lead them to learn DSP tutorial tutorials, before learning first wants to understand a concept, DSP has two meanings, one is the DSP chip is the Digital Signal Processor, and the other is Digital Signal Processing, that is, we often say that the digital signal processing technology. This tutorial is primarily concerned with the latter.
table of Contents
Chapter 1 Digital Signal Processing beginner preparations
The difference between 1.3 Cortex-M7 core DSP and DSP-professional
1.4 ARM provides CMSIS-DSP library
1.5 TI 32-bit fixed-point DSP provided library IQmath
1.6 ARM DSP software replaces the advantages of analog devices
1.1 Important Beginners
- Questions about learning methods, you can see the accessories section A.
- In recent years more and more powerful performance single-chip, low-end applications of basic DSP chip microcontroller can be used to do.
- The current single-chip AI is a certain outlook, ARM has been vigorously promoting, many software companies and research institutions are constantly trying. By Citie under can understand: SCM AI in the spring really come, ARM DSP library has the latest support NEON, and support for Python http://www.armbbs.cn/forum.php?mod=viewthread&tid=94406 .
1.2 STM32H7 of DSP Features
STM32H7 Cortex-M7 is the use of core and the DSP core carrying function, let understand the core block diagram M7 by:
Key to see the following two design elements:
- DSP
DSP unit incorporates a number of dedicated instruction set (SMID mainly MAC instructions and fast multiply accumulate instructions) can accelerate the speed of the digital signal processing is performed.
- FPU
Cortex-M7 double precision floating point core support, which greatly speed up the processing speed of floating point operations.
The following is Cortex-M3, M4, and M7 is an exploded view of a set of instructions:
By this figure, we can understand the following:
- M4 and M7 have the same series DSP instruction set.
- M7 M4 series to be more compared to the floating-point instruction set.
- At the same time pay attention to a small detail here, floating point instructions are to begin with the characters of V. By this point, we can easily verify that the correct turn on the FPU (MDK or IAR debugger to view the state of floating-point operations corresponding disassembly if there are such instructions).
DSP performance comparison of various kernels M:
- DSP core performance Cortex-M7 strongest.
- Cortex-M3, M4 and M33 are medium-performance, wherein M3 weakest.
- Cortex-M0, M0 + M23 and the weakest performance.
The difference between 1.3 Cortex-M7 core DSP and DSP-professional
M distinction DSP core of DSP processing unit professional:
1.4 ARM provides CMSIS-DSP library
To facilitate implementation of DSP functions, ARM made a special DSP library CMSIS-DSP, primarily digital signal processing algorithms include the following:
- BasicMathFunctions
Provides basic data operations, such as addition, subtraction and other basic operations, as a function of the end of the floating-point operations _f32 to _q8, _q15, _q31, the end of the fixed-point arithmetic function, the following is a partial screenshot API:
- FastMathFunctions
The main provider of SIN, COS, and SQRT square root operation.
- ComplexMathFunctions
Complex mathematical operations, primarily vector modulo arithmetic and the like. Here are some screenshots API:
- FilteringFunctions
Primarily filtering function, such as IIR, FIR, LMS, etc. The following is a partial screenshot API:
- MatrixFunctions
Mainly matrix operations.
- TransformFunctions
Conversion function. It includes plural FFT (the CFFT), a plurality of inverse FFT (CIFFT), real FFT (rfft), inverse real FFT, the following is a partial screenshot API:
- ControllerFunctions
Control functions, PID control function is mainly positive and cosine functions.
- StatisticsFunctions
Statistical functions functions, such as an average, maximum, minimum, power, and so the RMS, the following is a partial screenshot of API.
- SupportFunctions
Support functions, such as data copy, Q format and floating point format conversion.
- CommonTables
arm_common_tables.c bit flip or provide documents related to the parameter table.
1.5 TI 32-bit fixed-point DSP provided library IQmath
The initial use of fixed-point library, I feel more professional in handling the conversion between formats conversion between a variety of Q, Q values and floating point format of some, makes clear.
So this tutorial will also be introduced and IQmath supporting one example.
1.6 ARM DSP software replaces the advantages of analog devices
Many places in our daily lives DSP used to life in the equipment, for example:
Alternatively analog components by ARM DSP software can reduce the cost, PCB area and design time, while increasing the flexibility and adaptability.
- Reduce BOM cost
The most obvious benefit of converting analog circuit software is the bill of materials (BOM) cost reduction.
- Increase design flexibility
Used to continuously adjust the analog filter circuit for obtaining optimum performance, it is not uncommon. Smaller circuit board modifications cause new electrical characteristics abruptly change the parasitic capacitance or inductance, thus achieving the desired results in an analog circuit. Converts analog circuit DSP algorithm can not only eliminate this risk, it can also be adjusted according to need software, and more flexible.
- Reduce product size
BOM cost reduction has the added benefit of also allows developers to reduce the size of their products.
- Shorten design cycle time
Converting the analog circuit design software helps to shorten the cycle. There are several reasons:
-
- First of all, there are many tools available to software designers to simulate and generate DSP algorithms needed to replace analog circuits. This is generally much faster than the time required by the circuit simulation and test adjustment circuit.
- Second, if your needs change, you can make changes in the software, which can be done in minutes, without having to re-adjust circuit boards or hardware modifications.
- Site adaptability
In some products, the designer difficult to predict various situations encountered by users in the field. The use of DSP algorithms, designers and even the user can be adjusted in real time to adapt to site conditions, without extensive hardware modifications.
There are many benefits of using digital signal processing algorithms replace analog circuit. Designers need to weigh the benefits in practical application, choose the most appropriate solution.
1.7 Matlab installation
Matlab is a very important process in learning DSP aids, but also need to master in Chapter 2 to Chapter 5 introductory tutorial were introduced.
1.8 summary
The main issue is to do some tutorial introductory presentation, the next phase will start combat tutorial.
