Computers and their components

A computer system takes in data, processes it, stores it and gives out results. Each of these jobs needs its own hardware.

• Input devices are needed so that data and control signals can get into the computer, either from a person (for example a keyboard or a touchscreen) or automatically (for example a sensor). Without input there is nothing to process.
• Output devices are needed so that the results of processing can be used: shown on a screen, printed, played through speakers, or used to drive a motor.
• Primary memory (main memory: RAM and ROM) is needed because the processor can only work on the data and instructions held in main memory. A program, and the data it is using, must be loaded into RAM before it can be run. ROM holds the instructions that start the computer.
• Secondary storage is needed because RAM is volatile: everything in it is lost when the power is switched off. Secondary storage is non-volatile, so the operating system, programs and files are kept when the computer is off. It also holds far more data than main memory.
• Removable storage is secondary storage that can be taken out of the computer, such as a USB flash drive, a memory card, an external hard disk or an optical disc. It is needed to move data from one computer to another, to keep backup copies away from the computer, and to archive data.

Input, output and storage devices that are not part of the processor are called peripherals. Each peripheral is connected to the computer through a port, for example a USB port or an HDMI port (ports are explained in 4.1).

Input devices are peripherals used to provide data and control signals to a computer system. Data comes in many different forms, and there is a wide range of devices that can be used. The input process might be manual, such as through a keyboard or mouse, or automated, such as through a sensor. Some of the more common input devices, and their main purpose, are listed below.

A touchscreen is a display screen that is also an input device: it detects where it has been touched.

Early touchscreens worked with CRT screens, but could be used just as well with flat screens. Emitters were placed along the sides of the screen with detectors opposite them. The emitters produced infrared light or ultrasonic waves. When a finger touched the screen it blocked some of the light or ultrasound, so some of the detectors measured a weaker signal.

A modern touchscreen has the layers that make the display, with extra touch-detecting layers just beneath the surface of the screen. There are two approaches: the resistive touchscreen and the capacitive touchscreen, explained in the next two sections.

In every type of touchscreen, the processor takes readings from the measuring devices and uses them to calculate the position (the coordinates) of the touch. The processor then compares this position with what is shown on the screen, so it can carry out the action the user selected, for example choosing an item from a menu.

A resistive touchscreen has two layers separated by a thin space beneath the screen surface. The screen is not rigid, so when a finger presses on it, the pressure moves the top layer down until it makes contact with the lower layer. The point of contact creates a voltage divider in the horizontal and vertical directions, and the processor uses these voltages to work out where the screen was pressed.

A capacitive touchscreen does not need a soft screen. Instead, it uses the fact that a finger touching a glass screen causes a change in capacitance in a circuit component just below the screen. The most effective type is projective capacitive touch (PCT) with mutual capacitance: there is a circuit beneath the screen containing an array of capacitors. It can detect the touch of several fingertips at the same time (multi-touch), which allows more advanced applications.

To input sound, a microphone is needed. A microphone has a diaphragm, a flexible material that vibrates when sound reaches it. The diaphragm is connected to circuitry, so the vibration causes a change in an electrical signal. A dynamic microphone uses a coil that moves past a magnet. A condenser microphone uses a change in capacitance instead, and another alternative is a piezoelectric crystal.

The microphone produces an analogue electrical signal. An analogue-to-digital converter (ADC) converts it into a digital signal so that it can be processed inside the computer.

Output devices are peripherals that reproduce or display data and information from a computer system. Information is data that has been processed and formatted so that it is useful to the user. Some common output devices are listed below.

A printer produces a hard copy of information. Some printers print only in black (monochrome) and some print in colour using CMYK (cyan, magenta, yellow and black), which are mixed to make other colours. A printer forms an image from tiny dots, and its resolution is measured in dpi (dots per inch): the more dots per inch, the higher the resolution.

A laser printer uses a powder called toner, supplied in a cartridge. A colour laser printer has 4 toner cartridges, one for each CMYK colour. A laser is used to create static electricity on the areas that make up the image or text being printed. Toner sticks to these areas, is transferred to the paper, and is then fixed to it by a fuser (heated rollers).

A 3D printer creates solid three-dimensional objects. The object must first be drawn as a 3D model, using a computer-aided design (CAD) application, or scanned using a 3D scanner. The model is split into many thin horizontal layers (slices). The printer uses a material such as a polymer, acrylic or plastic, and prints each layer on top of the one before. Building an object up layer by layer like this is called additive manufacturing.

There are other types of 3D printing:

• Direct 3D printing uses inkjet technology. The print head moves from side to side, and it can also move up and down to build the layers of the object.
• Binder 3D printing uses two passes for each layer. The first pass sprays dry powder on to the layer. The second pass sprays a binder (a type of glue) to make a solid layer.
• Newer technology uses lasers and UV light to harden a polymer.

To output sound, a speaker (loudspeaker) is needed. It works in effectively the reverse way to a microphone. Digital data from the computer is converted to analogue by a digital-to-analogue converter (DAC). The analogue signal is fed to the speaker as a varying electric current. In most speakers, the current flows through a coil inside the magnetic field of a permanent magnet. As the size and direction of the current keep changing, the coil moves backwards and forwards. This moves a cone (diaphragm), which creates the sound.

Virtual reality has many uses. For example, it lets an engineer walk around inside a dangerous place, such as a nuclear power plant, without actually being there. Engineers can plan changes or repairs in complete safety, and try out different scenarios before carrying them out. A virtual reality headset gives the user the feeling of being there. It works like this:

• Video is sent from a computer to the headset, either through an HDMI cable or from a smartphone fitted into the headset.
• Two feeds are sent to an LCD or OLED display (sometimes there are two screens, one for the left side of the image and one for the right). Lenses between the eyes and the screen focus and reshape the image for each eye, giving a 3D effect.
• Most headsets have a 110° field of view, which is enough to give a pseudo 360° surround image.
• A frame rate of 60 to 120 images per second is used to give a realistic image.
• As the user moves their head (up and down or left to right), sensors measure the movement. A microprocessor works out the direction of the movement, and the image on the screen changes to match. The sensors are usually gyroscopes or accelerometers; LEDs are used with mini cameras to track head movement further.
• Headsets use binaural (surround) sound, so the speaker output seems to come from behind, from the side or from a distance, giving realistic 3D sound.
• Some headsets use infrared sensors to track eye movement as well as head movement. This makes the depth of field more realistic: for example, objects in the foreground look fuzzy when the user’s eyes show they are looking into the distance.

The RAM (random access memory) inside a computer holds all the data and instructions that are currently being processed. A program must be loaded into RAM from secondary storage before it can be run, because RAM can be read and written much faster than secondary storage. The processor fetches instructions (and any data it needs) directly from RAM, then decodes and executes them.

RAM is split into separate locations, each with its own unique memory address. The processor uses these addresses to access the data in RAM. Any location can be accessed in the same amount of time, unlike storage devices where something (such as a read-write head) has to be moved into place. This is why it is called random access: any location can be reached "at random" with no effect on speed.

RAM is volatile: it needs power to hold its data, and if the power is switched off, all the data in RAM is lost. Any data that must be kept has to be saved to secondary storage before the computer is switched off. If you have ever lost work because you had not saved it when the computer switched off, you have seen the effect of RAM’s volatility.

RAM has a smaller capacity than secondary storage, but it is usually the largest part of main memory.

ROM (read-only memory) usually stores the boot sequence (BIOS) of a computer. Unlike RAM, it is non-volatile: it does not lose its data when the power is switched off. The contents of ROM are set by the manufacturer and, as the name says, the computer usually only reads from it. This is a good thing, because ROM is crucial to the computer, and any change could stop it starting up properly.

The BIOS (Basic Input/Output System) stored in ROM is a short sequence of instructions that checks that the core components of the computer (RAM, the basic input and output devices and secondary storage) are connected and working correctly.

After that, the boot sequence loads the essential parts of the operating system from secondary storage into RAM. From then on, the operating system runs the computer, managing memory, storage and requests for input and output.

RAM and ROM are both types of primary memory (main memory), but they are different in several important ways.

RAM has a much greater capacity than ROM. RAM holds the data and instructions that are being used, such as the operating system and any programs you are running. When you put a disc into a games console, the code on the disc is not run until it has been copied from the disc into RAM, which is why you see a loading screen.

Because ROM is read-only, it stores core instructions that do not need to change, such as the BIOS and the code that loads the operating system into RAM (called bootstrapping).

RAM and ROM are used together in many devices, not only in computers. In an embedded system such as a washing machine, a car or a television:

• ROM stores the program (the firmware) that controls the device, such as the washing cycles, and the start-up instructions. These must not be lost when the power is switched off, and the user must not be able to change them.
• RAM stores the data being used while the device is running, such as the settings the user has chosen, the current temperature, the time left, or the video being recorded.

DRAM stands for dynamic random-access memory. It is commonly used for a computer’s main memory because it is relatively cheap to manufacture.

A DRAM memory cell is usually made of a capacitor and a transistor. The capacitor stores a charge: whether it is charged or discharged represents a single bit. The transistor acts as a switch to set or change the value. The charge on a capacitor leaks away over time, so DRAM needs a frequent memory refresh to restore the capacitors to their original charge. This happens constantly in the background, so the user does not notice it.

SRAM stands for static random-access memory. It uses a different type of circuit (flip-flops), which does not need to be refreshed. SRAM is generally about twice as fast to read or write as DRAM, but it is more expensive to manufacture, so it is used where speed matters most, for example for registers and cache memory.

DRAM and SRAM are both volatile: they need a constant power supply to keep their data.

Most electronic devices contain ROM. As the name says, ROM is not meant to be written to. Traditional ROM circuits, called mask ROM, have their contents written during manufacture, and they cannot be changed afterwards.

PROM (programmable read-only memory) is a type of ROM that can be written to once (only) after the circuit has been manufactured. It is often used when a device is made in small numbers, so it would not be cost-effective to write the contents during manufacture. The PROM is supplied empty, and the supplier of the device writes the data to it. This also lets the supplier write the latest version of the software without an expensive redesign of the manufacturing process. PROM is used in microcontrollers, games consoles, mobile phones and RFID tags.

EPROM (erasable programmable read-only memory) can be changed. The memory chip is removed from the system and ultraviolet (UV) light is used to erase the data; all the contents are erased in one operation. This is a specialist job that needs special equipment and an expert operator.

EPROM has been replaced by EEPROM (electrically erasable programmable read-only memory). Individual bytes can be erased electrically, using a voltage, and the chip can then be reprogrammed. This can be done while the chip is still inside the device, so the device’s software (its firmware) can be updated without taking it apart. EEPROM is used in smart cards and remote keyless systems. Some EEPROM chips have extra security features to stop hackers reprogramming them.

Whenever data is transferred from one part of a computer system to another, there is a problem if the data can be sent more quickly than it can be received. The solution is a buffer: an area of memory, usually in RAM, where data is held on its way to its destination. A buffer works as a queue, so data comes out of it in the same order that it went in.

A printer is a good example, because a printer is much slower than the processor:

1. The processor sends the data to be printed to the printer buffer, as fast as it can.
2. The printer takes data from the buffer and prints it at its own, slower speed.
3. Meanwhile the processor is free to carry on with other tasks instead of waiting for the printer.
4. When the buffer is empty, the printer sends an interrupt to the processor to ask for more data (interrupts are explained in 4.1).

Buffers are used in the same way with other devices:

• Hard disk or optical disc writer: data waits in a buffer until the disk is ready to write it, so the processor does not have to wait for the slower device.
• Speakers, streamed video and VR headsets: incoming data is stored in a buffer before it is played. The device reads from the buffer at a steady rate, so the sound or video plays smoothly without pauses, even if the data arrives unevenly.
• A video doorbell or a 3D printer: the data (the video being recorded, or the print data) is held in a buffer until it can be transmitted, stored or printed.

An HDD is a hard disk drive. It is a mechanical device with many moving parts, so in theory it is more likely to fail. However, HDDs are sold as sealed units, so the disk and the mechanism are well protected from dirt and from wear and tear. Data is stored on the disk surface (a metal platter) as magnetised dots: binary data is represented by the two states "magnetised" and "not magnetised".

To read or write, the disk must spin at high speed. The speed of rotation is a crucial factor in how fast data can be accessed: the faster the disk spins, the faster the data can be read. Standard drives spin at 5,400 or 7,200 rpm (revolutions per minute). Higher-specification drives, such as those used in servers, can spin at 15,000 rpm.

Standard HDDs read and write at between 80 and 200 MB/s (megabytes per second). The speed depends on how the disk has been set up, the size of the file, and how fragmented the disk or file is. Fragmentation happens when there is not enough contiguous (continuous) space to store a file. A fragmented file is stored in separate places, each with a link to the next, and it is read by following the links until an end-of-file marker is reached. When a disk is nearly full, it takes longer to find space to write a file, and longer to read a file spread across several places.

The main advantages of a magnetic hard disk are its capacity and cost: it still gives much more storage for a much lower cost per gigabyte than an SSD, although the gap is narrowing fast. Magnetic hard disks are best for desktop and laptop computers where a low purchase price matters most.

The disk unit has one or more metal platters with a magnetic coating. Each platter is double-sided, and the number of platters decides the total storage capacity. Binary data is stored on the platters as a series of magnetised (polarised) states, each representing 1 or 0.

Each platter is divided into a set of concentric rings called tracks, and the tracks are divided into sectors. A contiguous group of sectors is called a cluster (also called a block or an allocation unit). This is the smallest unit of storage the file system can address, and its size depends on the size of the disk and how it was formatted.

The location of every file is recorded in a file allocation record. This lets the file be found again, and lets free space be found for new files. When the disk is fairly full, files are often split across many blocks that are not next to each other, so the record must keep track of where the parts of the file are and the order they go in. If this record is lost or corrupted, the data is still on the disk, but it cannot be accessed in a meaningful way unless the record is rebuilt.

To read and write data, there is an arm with a read-write head for each side of each platter. The head is a tiny magnet. To reduce wear from friction, the head does not touch the platter: there is a very thin gap between the head and the disk surface. The platters spin, typically at 7,200 rpm. The head sweeps across the tracks, while the sectors pass underneath it as the disk turns. This means any block can be accessed directly, without reading through the whole disk. Magnetic tape, which disks replaced, only allows serial access.

SSD stands for solid-state disk, though it is also called a solid-state drive or solid-state device. An SSD is an electronic device that uses flash memory, and it has no moving parts.

SSDs have many advantages over magnetic hard disks. The main one is speed of access: a standard SSD reads and writes at about 200 to 500 MB/s, and SSDs always outperform magnetic or optical drives.

With no moving parts, mechanical failure is far less likely, and an SSD is less likely to be damaged by everyday bumps and knocks, which makes it a better choice for portable devices. Having no moving parts also means it uses less energy, so less heat is produced.

However, an SSD has a limited lifespan: it can only be written to a limited number of times, and memory cells can leak their charge if the device is not powered up from time to time. So SSDs are not suitable for long-term archiving of data. SSD storage is also still more expensive per gigabyte than magnetic storage.

SSDs are usually built into a case the same shape as a magnetic disk drive, so a computer can be upgraded by swapping the drive or adding a second one. SSDs are best for files that are read or changed often: putting the operating system on an SSD makes the computer start up much faster, and frequently used programs and data can go on it too. Many computers now have both: a small SSD for the operating system and frequently used files, and a larger magnetic disk for files used less often.

A solid-state disk is made of a controller and a bank of millions of NAND flash memory cells. Flash memory can be built using NAND or NOR technology (named after the NAND and NOR logic gates); SSDs use NAND flash.

Each cell has a floating gate transistor that lets an electrical charge be trapped. A cell with trapped charge represents 0, and a cell with no trapped charge represents 1. Each cell has two gates, often described in exam questions as two transistors:

• The control gate controls whether current can pass through the cell, from the source to the drain.
• The floating gate sits between two insulating layers, so it keeps its electrons (its charge) even when there is no power. This is why flash memory is non-volatile.

NAND flash cells are organised into pages and blocks. A page is a group of cells (4 KiB in the diagram), and a block is a group of pages (256 KiB in the diagram).

Data is read from an SSD a page at a time. It can also be written a page at a time, as long as the cells around it are empty. A file cannot be updated where it is: the block containing its pages must be copied to main memory and updated, then written to a new area of the disk, and the original block is erased.

SSDs become slower as they fill up. A new, empty drive is full of pages that can be written to straight away. When the disk is nearly full, writing and updating become slow, because existing pages in a block have to be copied out and rewritten.

Optical discs are a large family of disc types. The most common are the CD (compact disc), the DVD (digital versatile disc) and the Blu-ray Disc. For optical storage the word is usually spelt "disc", unlike other types of disk storage.

• CD-ROM and DVD-ROM: ROM means read-only memory. These discs are usually mass-produced to distribute music and films. Each one is pressed from a mould (mastered) in one operation, so tens of thousands of copies can be made in a day. They were used to deliver application software, because they are light, easy to carry and take up little space, but now software is usually distributed over the internet.
• CD-R and DVD-R: R means recordable. These discs are also described as WORM (write once, read many). They were popular for archiving data, such as photos. Archiving means making a permanent copy, so it needs a storage medium that cannot be overwritten.
• CD-RW and DVD-RW: RW means rewritable. These discs were popular for backing up files. Backing up means making a temporary copy, so it needs a storage medium that can be overwritten later. Today, most systems are backed up to high-speed digital tape or to cloud storage.

The speed rating of an optical disc is given as a multiple of a base speed, written as 1×. For CDs, the base is the speed needed to play a music (audio) CD. For DVDs, it is the speed needed to play a film.

The actual speed of reading or writing depends on two things: the speed rating of the disc and the speed rating of the drive. If a disc is rated 16× but the drive is rated 8×, the lower of the two is used. The same applies to writing.

CD-ROMs and DVD-ROMs are fairly robust, but they can be scratched if the surface is not protected. Heat can make the polycarbonate or metal layers of any disc warp. Discs with a layer of chemical dye can be damaged by light: long exposure to sunlight damages the dye, making it more likely that data is misread, and any chemical layer breaks down over time.

The biggest problem with optical discs is finding a drive to read them. Most computers no longer come with an optical drive, and they are disappearing from music players and cars. Valuable data on a CD or DVD should be copied to cloud storage.

An optical disc is made of several layers, depending on the type of disc. The base is usually polycarbonate, which gives the disc its structure, and there is always a thin reflective layer of aluminium. Recordable (R) and rewritable (RW) discs also have a layer of chemical dye on top of the reflective layer.

Each disc has one continuous track of data arranged in a spiral (not a set of concentric tracks like a hard disk). Data is stored as pits (troughs) and lands (peaks) in the surface of the disc.

An embedded system still needs a processor, memory and input/output. If these are all built on one chip, it is called a microcontroller. In some systems the input and output are only connected to the internal workings of the host device. In others, for example a monitoring or control system, the input comes from within the system but some output is given to the user. An embedded system can also have a full user interface, as in a mobile phone.

The major advantage of embedded systems is that they are special-purpose, perhaps doing only a single task. That task is likely to be needed in many different products, so embedded systems are mass-produced, which brings economies of scale: the more are made, the cheaper each one becomes.

In the early years, embedded systems were difficult to program because there was little memory to store the program, so programs had to be short. Also, if errors were found after the system was installed, new chips had to be made to replace the faulty ones.

These problems are less likely in modern systems, but a new problem has appeared. Embedded systems are now part of the Internet of Things: more and more of them have a network connection. This can make a product much more useful, for example by sending information and updates to the owner. But being reachable over a network is a security concern, because embedded systems are less likely to be protected against attack than general-purpose computers.

Many embedded systems use sensors to measure the physical world. They are either monitoring systems or control systems.

A monitoring system uses sensors to measure physical quantities, such as temperature or movement. The processor records and analyses the readings, and may display them or give a warning (for example sound an alarm or show a sign) when a reading is outside its set limits. A monitoring system does not change the conditions it is measuring. Examples: a burglar alarm, a health monitor that reads a person’s heart rate, a video doorbell that records visitors, a warning sign that tells an overweight or over-height vehicle to stop before a bridge.

A control system also uses sensors, but the processor then uses actuators to change the physical conditions, to bring them back to (or keep them at) the required values. The output of the system affects its next input, which is called feedback. Examples: a refrigerator that switches the cooling on and off to keep the temperature between 4 °C and 6 °C, automatic doors that open when a person is near, a car that brakes automatically when an object in front is too close.

1. A sensor measures a physical quantity and sends an analogue signal.
2. An analogue-to-digital converter (ADC) turns the reading into digital data the processor can use.
3. The processor compares the reading with the stored set value or range.
4. Monitoring system: the processor records or displays the data, and gives a warning if the reading is outside the range. Control system: if the reading is outside the range, the processor sends a signal to an actuator to change the conditions (for example switch a heater, a motor or a valve on or off).
5. The cycle repeats continuously.

A sensor is an input device that measures a physical quantity and sends the reading to the processor. Most sensors produce an analogue signal, which is converted to digital before the processor uses it. Common sensors include:

An actuator is an output device that turns a signal from the processor into physical movement or action, such as a motor, a valve, a pump, a heater or a switch. For example, in a factory, when a sensor shows that a chocolate bar has the wrong weight, the processor sends a signal to an actuator, which moves an arm to push the bar off the conveyor belt.

Feedback is when the output of a control system affects its next input. After the processor tells an actuator to act, the sensors keep measuring, so the processor can see the effect of what it did and adjust its next output.

For example, a refrigerator’s temperature sensor reads 7 °C, which is too high, so the processor switches the cooling on. The sensor keeps reading the temperature as it falls. When it reaches 4 °C, the processor switches the cooling off. Without feedback, the processor would not know when to stop.