[SOUND] Hello, and good to see you back. Let me make a small recap about the previous lecture. In the previous lecture, we went more technical, and gave a general definition of a processor, and an FPGA. We also made a comparison between these two types of processing elements. I hope you remember the main difference between them, which is sequential nature of a processor, versus the parallel nature of an FPGA. I hope now you have some basic intuition on what a processor and FPGAs are. In this lecture, more details on an embedded processor will be given. And we will also discuss the main features of an embedded processor. First of all, a processor is an electronic device that executes instructions. The processor does not understand anything but numbers, and can work with and manipulate only numbers. So instructions are not an exception. They are numbers. However, unfortunately, humans are not as fast with numbers as processors. So instructions are described with words to make them comprehensible for people. All the instructions are stored in memory. And they are fetched from the memory, one by one, to be executed. Obviously, there is a wide range of different applications that can be run on a processor. And it is clear that one type of processor cannot deal with this range of applications in an equally efficient way. So it makes sense to have several types of embedded processors, which are dedicated to certain types of applications. To make an example, let's think about signal processing techniques, which are used for image processing. Signal processing has different, basic linear algebra routines in it's core. DSP Processors are designed specifically for signal processing, which have dedicated units that deal with those routines efficiently. During our course we will mostly talk about Microcontrollers, or MC use. And here I want to mention that Microcontrollers are meant for general purpose applications, and most of them are resource constrained devices, which are not meant for computationally intensive tasks. There are also GPUs with thousands of cores within, which are meant for extremely computationally demanding tasks like graphics rendering. Of course it is sometimes challenging to choose in a specific type of embedded processor. Therefore, we have several metrics to make such choice. Some of them, performance, energy consumption, versatility, and cost. As it was said a bit earlier, it is not possible to cover all the types of embedded processors in this course. However, we are going to elaborate on microcontrollers. I am going to shortly explain each of the main characteristics of a microcontroller. A microcontroller has in its architecture a CPU, which is in charge of fetching, decoding, and executing instructions. The arithmetic logic unit is the unit that actually performs the computations. Data is loaded into the CPU's registers from the data memory using special instructions. In the same way, the result of computations can be stored in memory. The program is stored in a program memory as a sequence of instructions. In order to keep track of where the execution is in this sequence, there is a program counter. There is also RAM memory available in microcontrollers and some kinds of non-volatile memory, such as EEPROM. Microcontrollers include different kinds of peripherals such as communication units, timers, ADCs, DACs, etcetera. If we think about the instruction's execution, the question that arises is, when is it time to actually execute an instruction? The answer is that the execution is clocked, so it happens according to a clock frequency. Each instruction takes on one or several clock cycles to complete. From this, you can conclude that the execution time depends on the clock frequency. The clock frequency of microcontrollers can derive from several megahertz to several hundreds of megahertz. There's a special dedicated unit called Clock Control unit that it's responsible for managing clocking. For example, this unit can scale down an input reference frequency or multiply it to obtain a higher club frequency. A microcontroller may have several sources of clock frequency. One of them is an internal RC oscillator. However, the stability of this kind of oscillators is low, that's why external clock sources such as an external quartz resonator, might be used. It is important to introduce the notion of an interrupt, since it is a powerful tool to handle real-time events. An interrupt is a hardware-generated signal. Indeed, interrupts are meant to break the execution flow of a program when some external event happens. For example, assume we have some communication unit, and we want to check if there is new data in our receive buffer. One way to do it would be to write a piece of code that checks the input buffer once in a while. However, this approach does not allow to know precisely the time when the data arrive. So there is an alternative way to manage this task. We can adjust the microcontroller in a way so the execution of the main program is interrupted precisely when the data comes into the buffer. Moreover, when a certain interrupt occurs, the program counter jumps to a specific location in a program memory assigned to this interrupt. An interrupt service routine, written by a programmer, can reside in this memory location. Once the interrupt service routine is finished, the program counter jumps back to the main program. Each interrupt can have a priority. If two interrupts happen at the same time, the one with the higher priority will be processed first. A standalone CPU is useless if it is not supplied with data. Peripherals are exactly for this purpose, to give a CPU data to work with. Peripherals can be classified into several groups. First, communication units such as SPI, I2C, UART, Ethernet, and others. All these units work based on specific protocols. Next, there are analog units which deal with analog signals. There are ADCs for converting a voltage level to a number, and DACs for converting a number to a voltage level. Timers constitute another group. They are powerful tools when it comes to dealing with clocking and timing issues. With the help of timers, PWM can be generator, or they can be used to measure time intervals. There are units to manage memory. One of them is DMA, which it stands for Direct Memory Access. This unit can transfer data from one location of memory to another without the CPU's intervention. Energy efficiency is an important issue nowadays, and microcontrollers offer flexible alternatives to address it. Modern MCUs have multiple power modes that provide developers with fine grain control over the power consumption of their applications. This way it is possible to power off the CPU, the peripherals, or both of them. Microcontrollers have different programming and debugging interfaces. One of the most widely used ones is JTAG. JTAG programmer is a convenient tool, especially when it comes to debugging. However, a debugger is not cheap to buy. A bootloader can be a solution to this. A bootloader is a program which resides inside of a microcontroller. The bootloader receives a program byte by byte via any communication protocol, and writes it into a program memory of a microcontroller. Even though a bootloader does not allow to debug code, it is a free of charge way to program a microcontroller. This is all I have for this video. I hope now you have a more in-depth knowledge about embedded processors and microcontrollers. In the next video we will cover MCU platforms, and we will also describe some real industrial projects, powered by MCUs. [SOUND]
