EE Terminology

A5: See encryption.
AAC: Advanced Audio Coding. An ISO-endorsed audio data format with lossy compression, based on MPEG-2 and intended to supplant MP3. AC-2 is a sub-type used by Apple.
AACS: Advanced Access Content System. An aggressive AES-128 copy protection scheme replacing CSS for DVDs, also used for Blu-ray and HD DVD. Windows Vista complies with it.
AAFP: Analog Audio Front Panel. A 10-1 header (i.e., a 2x5 pin header with one pin missing) on PC motherboards for connecting audio. It supports either AC'97 or the newer HD Audio.
AAID: Android Advertising Identifier. A unique 32-character string assigned to each Android mobile-telephony device. Apps on the phone use it to track the activities and physical location of that device (and its owner), usually for advertising purposes. Starting with Android 12 in 2021, Google changed Android privacy settings to allow users to set the AAID to an all-zero string, which renders it largely ineffective if users are aware of the capability and choose to use it. The equivalent on Apple devices is the IDFA.
ABL: Airborne Laser.
A-bomb: Atomic bomb. All practical nuclear fission weapons are based on either uranium-235 or plutonium-239. U-235 constitutes about 0.7% of all naturally occurring uranium; almost all the rest is U-238. Pu-239 exists naturally only in trace quantities, and is manufactured by bombarding U-238 with neutrons. Handling sub-critical masses of these two isotopes is less dangerous than commonly believed. Gloves and normal clothing block alpha-particle radiation, which is almost all they emit. Inhalation or ingestion is still deadly.; Other materials can be used for weapons, but present greater practical difficulties – rarity, higher radiation and heat, harder to refine, larger quantities required for critical mass, lower explosive yields. Candidates include U-233, neptunium-237, different isotopes of uranium and plutonium, and many of the transuranic elements. Bombs can also be made from mixes of materials.; Highly enriched uranium (HEU) is at least 20% U-235. Weapons-grade uranium is 90% or more U-235, and weapons-grade plutonium should be 94% or more Pu-239. It takes at least 15-20 kg (33-44 pounds) of weapons-grade U-235 to make a bomb, depending on how rich the material is, and also on the mechanism. The very first A-bomb, the Trinity test, achieved a yield equivalent to 18.6 kilotons of TNT using the hollow-sphere implosion approach and a softball-sized, 8 kg (~18 lb.) mass of Pu-239. The Hiroshima device, “Little Boy”, using a simpler but less efficient gunlike mechanism and about 60 kg (132 pounds) of U-235, achieved a yield of 15 to 16 kT. “Fat Man”, the 21 kT bomb dropped on Nagasaki, was almost identical to the Trinity device. Compare H-bomb.; The Hiroshima bomb caused 140,000-150,000 immediate casualties, dead and injured; the Nagasaki bomb, 75,000-80,000; and the March 9-10, 1945 firebombing raid on Tokyo, with 334 planes dropping 1667 tons of conventional incendiary bombs, 185,000. In a six-week period during December 1937 and January 1938, using mostly bayonets and bullets, the Imperial Japanese Army massacred an estimated 260,000 to 350,000 people in the Chinese city of Nanjing.
ABS: (1); Acrylonitrile Butadiene Styrene. A light, tough, bad-smelling thermoplastic used, among other things, for Legos and 3D printing.; (2); Anti-lock Braking System.
abscissa: The horizontal (x) component of a point in the Cartesian plane. Contrast ordinate.
ABT: Advanced BiCMOS Technology. A logic family with TTL circuitry at the inputs and outputs, and CMOS in between.
A/B testing: A dot-com practice in which a new version of a Web page is served at random to some fraction of users, so that their experience and behavior can be compared to those of the standard page’s users. If the new version fares better, it becomes the standard.
AC: Alternating Current. Specialized applications use AC running at the hundreds or thousands of hertz, but basic line power in North and Central America is a 60 Hz, single-phase sine wave with a nominal peak value of ±177 volts, which converts to a RMS level of 125 V. It’s also called 120 V, 115 V, or 110 V power, because actual voltage peak varies depending on the power company, the neighborhood load at the time, and how close to the transformer it’s being measured. For North American electrical plug & outlet standards, see NEMA.; Most of the rest of the world uses 50 Hz, single-phase, 250-volt AC power, referred to as mains in the UK. As with the North American standard, its actual voltage varies, so it’s also called 220 V, 230 V, or 240 V. Only a few nations use 125V/50Hz or 250V/60Hz power.; Almost all power distribution grids carry AC rather than DC, because AC is what electrical generators produce, and also because the AC versions of essential hardware – step-up and step-down transformers, large motors, rectifiers – are inherently more efficient than their DC equivalents. (However, see HVDC.); A simplified explanation of US power distribution: A power plant generates three-phase AC power, with 120° difference between phases A, B, and C (each comes from a different generator coil). The plant’s step-up transformers raise the voltages to, typically, between 155 and 765kV, since power loss in transmission is proportional to the square of current, and higher voltages permit lower current. Long-distance power lines carry the power on four cables: one for each phase, and a common return. Distribution substations use step-down transformers to lower the voltage to, typically, 7200V. Local lines carry power plus return to neighborhoods.; In a typical installation (the image above), a building transformer steps down just one of the three hot lines to produce two 120V lines, 180° out of phase. These allow the building to be wired for single-phase AC at 120V (line 1 or 2 to neutral) or 240V (line 1 to line 2). A single neutral (return) line from the building returns to the center tap of the transformer, and is also bonded to earth ground at the building’s circuit panel. (Some older installations use metal water pipes as earth ground. If anything interrupts the neutral return, this can be a hazard.) See NEC, phasor.; Current and voltage throughout the grid must be in phase, but cycling equipment (motors, generators) drawing on the grid cause current to lag voltage. This can cause voltage levels to drop, crashing the grid. The grid needs the capability to quickly supply reactive power to correct the problem.
AC‘97: Audio Codec 1997. Intel’s 1997 sound coder/decoder specification. It can support up to six 48-kHz, 20-bit sampled audio channels. A PC motherboard implementing it has a 10-pin header (with pin #8 missing) for the cable to the PC’s stereo ports, which are sometimes, incorrectly, called AC‘97 ports. HD Audio is replacing it in the market. The pinout (functions marked with an asterisk * are optional, and might be disconnected):
ACAP: Application Configuration Access Protocol. A member of the TCP/IP protocol suite at the Application layer. This little-used protocol provides a way for IMAP to retain customized settings regardless of user location.
ACH: Automated Clearing House. A US network for electronic transfer of money.
ACI: Adjacent Channel Interference.
ACL: Access Control List. A set of permissions for users of a network, file system, computer, etc.
ACM: (1); Association for Computing Machinery. A non-profit organization founded in 1947 to support educational and scientific computing.; (2); Air Cycle Machine. A system on modern aircraft that generates cool, pressurized air for the cabin. For this reason, it’s sometimes called just an A/C pack. It uses engine bleed air, which is already pressurized. This saves a step and also provides a source of power – air pressure – for the ACM. It works in stages, which are typically: cooling via heat exchange with outside air, compression (which causes heating), more cooling via heat-exchange, expansion (which causes further cooling), and then achieving desired temperature and pressure by mixing with engine bleed air that has not been conditioned. The ACM also removes water to maintain low humidity.
ACPI: Advanced Configuration and Power Interface. Starting 1996 and gradually replacing APM, this Microsoft-Intel-Toshiba standard defines how a computer’s OS should manage power consumption. It defines four system-wide power states: G0 (working), G1 (sleeping), G2 (soft shutdown), and G3 (mechanical off). Under G0, there are sub-states for individual system components.
ActiveX: In 1996, reacting to Java, Microsoft renamed 32-bit COM and OLE controls ActiveX controls and presented them as a new Internet technology. ActiveX provides an API for sharing executable code objects over a network regardless of programming language – but only between Windows machines. Unlike the similar Java applets, ActiveX controls are Win32 components, with full access to the Windows OS on the machine that downloads them. This simplifies interoperability for legitimate applications, and also for malware.; In 1997, ActiveX was renamed COM. See COM, DCOM.
A/D: Analog to Digital. See ADC.
Ada: See programming language.
adapter: A device for connecting two physically incompatible connectors that carry the same signal. Not the same thing as a converter.
ADC: Analog-to-Digital Converter. A device that samples and digitizes an analog signal, representing an infinite number of possible input levels with a restricted set of digital values.
ad hoc network: A dynamically self-organizing network with no fixed structure or membership. The assumption is that such a network must be wireless. Bluetooth and types of IEEE 802.11 were the best-known ad hoc network standards as of 2002. See also ZigBee.
ADLC: Asynchronous Data Link Control. See data protocol.
ADM: Add-Drop Multiplexing. A multiplexing function used in SONET to add and drop signals without affecting ongoing transmissions. See OADM.
admittance: The reciprocal (1/Z) of impedance, represented as Y. It’s a complex value equal to G + jB, where G is conductance and B is susceptance. The SI standard unit of admittance is the siemens (s).
ADP: Automatic Data Processing.
ADPCM: Adaptive Differential Pulse Code Modulation. An encoding technique using differential encoding and variable-sized quantization steps. The variance in step size is based on estimates from past signal samples.
ADSL: Asymmetric Digital Subscriber Line. See DSL.
adware: See malware.
AED: Automated External Defibrillator. A portable electronic device with diagnostic capabilities that untrained operators can use to shock a heart out of cardiac arrest.
AES: Advanced Encryption Standard. See encryption.
AF: Antenna Feed. The input from an antenna.
AFC: (1); Automatic Frequency Control.; (2); Active Flow Control. A technology for enabling long-range lasers to overcome atmospheric disturbances from turbulence, dust, smoke, etc. in their immediate surroundings that interfere with targeting. It uses precisely timed pulses of air or plasma to shape these disturbances so that the targeting can anticipate their effects.
AFCI: Arc Fault Circuit Interrupter. Alternating current (AC) normally cycles at 50 or 60 Hz. Electrical arcing between the line and neutral wires in a circuit causes a huge jump to around 100 kHz. An AFCI is a device installed in a building’s electrical supply panel that will break a circuit if AC on that circuit surges to such high frequencies for more than a few milliseconds. Whereas GFIs are meant to protect people from electric shock, AFCIs prevent fires.
AFDX: Avionics Full-Duplex Switched Ethernet. Patented and trademarked by European aircraft manufacturer Airbus, this is an implementation of the ARINC 664 data communications standard, which uses an IEEE-802.3 (Ethernet) network as an aircraft data bus. AFDX supports up to 24 nodes exchanging data packets through two redundant switches referred to as Network A and Network B. To create a network with more than 24 nodes, switches can be connected to other switches instead of to end nodes.; The Ethernet-compliant AFDX data frames are formatted as shown below, and sent with a 12-byte interval between frames.; The padding is treated as part of the payload, and is there to ensure that total payload length is at least 17 bytes. The payload can carry frames for other data protocols, including ARINC 429. Frame sequence number is just a one-up byte. Although it starts at 0, it overflows from 255 back to 1.; Rather than use addresses for routing as Ethernet does, AFDX uses virtual links (VLs), each of which identifies one source node and one or more destinations. The network user, by allocation of these VLs, determines in advance which nodes can send to and receive from which other nodes. That’s why AFDX is referred to as a deterministic network: the allowed paths are a fixed subset of all possible paths, and, because the system allocates dedicated bandwidth to each VL, maximum network usage is known.; The 6-byte (48-bit) source address field of a frame contains a 24-bit constant, a 16-bit ID for the transmitting node’s Ethernet controller, a 3-bit interface ID specifying either Network A switch (001) or Network B switch (010), and a 5-bit field that’s always 00000. The 48-bit destination address field contains a 32-bit constant that identifies the network, and a 16-bit VL identifying the transmitting node and receiving node(s).
AFG: Arbitrary Function Generator. A laboratory instrument that generates a periodic waveform with characteristics set by the operator. Unlike a signal generator, it’s not limited to sinusoidal waves. Compare also AWG (2).
AFM: Atomic Force Microscope (or Microscopy). An instrument (or technique) invented in 1986 for measuring and manipulating materials at the atomic level using a surface-mapping laser. It can work with materials that the slightly older STM can’t. Used for atomic-scale MEMS.
AGC: Automatic Gain Control.
AGI: Artificial General Intelligence. Another term for broad AI, sometimes used to mean still-hypothetical AI that has self-awareness.
agile: For agile programming, see process management.
AGP: Accelerated Graphics Port. A private channel for the graphics controller in PCs to access main memory directly, as opposed to going through the local bus. It’s based on 32-bit, 66 MHz PCI, and uses a PCI form-factor slot, but isn’t compatible. It increases graphics throughput 2, 4, or even 8 times over a PCI card by using multiple data fetches per clock cycle, and so is called AGP 2x, 4x, or 8x. It’s being pushed out by PCIe graphics as of 2009.
AI: Artificial Intelligence. In engineering and computer science, this broadly means a computer system that performs tasks dealing with unpredictable input. Machine learning is a subset of AI that has the capability to adjust its own processes in an attempt to improve. Deep learning is, in turn, a subset of machine learning based on artificial neural networks.; AI is sometimes classified by what it’s for. Weak (narrow) AI is designed to perform just one task; strong (broad) AI, also called artificial general intelligence (AGI), is meant to perform multiple types of tasks. Machine-learning AI is also classified by how it learns and improves. In order of increasing complexity, these learning approaches are:; supervised learning – The example data sets fed to the AI model to train it include the known results, so that the model can learn to analyze data of the same sort for which the results aren’t known. A result is either a classification (e.g., does this set of data represent an apple, an orange, or a papaya?) or a prediction (e.g., given a set of data points about a house, what will it sell for?).; unsupervised learning – The AI model looks for patterns in unlabeled data.; reinforcement learning – The AI model uses trial and error to interact with the data it receives and come to decisions or actions, with desired outcomes reinforcing its approach and undesired outcomes driving major changes to or rejection of its current approach.; Even allowing for the difficulty of defining “consciousness” and “self-awareness” to everyone’s satisfaction, machines with those qualities remain science fiction.
AIDC: Automatic Identification and Data Collection. See RFID.
AIMM: AGP slot In-line Memory Module. Also called a Graphics Performance Accelerator (GPA) card. A specialized and little-used type of video RAM on a 3.75" × 1.5" circuit card. It plugs into the AGP slot on a motherboard that’s using integrated AGP graphics rather than a discrete graphics card.
AIN: Advanced Intelligent Network. A Bellcore telephone network architecture that separates service logic from switching equipment, allowing new services to be added without changing the switches.
AIO: All In One. Usually means a desktop PC with the user-interface hardware (monitor, speakers, keypad, mouse/trackball/touchpad) built into the same case as the CPU, motherboard, drives, & PSU. The term can be applied to other components or systems that eliminate the need for assembly and external connections.
air card: A wireless modem for establishing an Internet connection via 3G or later mobile phone networks. It plugs into a USB, micro-USB, or PC Card slot on the user’s computer.; Because it links to cell towers using the same signal as a mobile phone, it has much greater range than wireless networks such as the 802.11 standards, but smaller bandwidth and higher latency. Users are typically either residents of rural areas that lack wired broadband service (an air card is no match for DSL or a cable modem, but it beats dial-up), or people who want mobile Internet connectivity on a general-purpose computing platform rather than on a smartphone or tablet PC.; AirCard^® is a registered trademark of Sierra Wireless, the biggest vendor as of 2013.
AIS: (1); Application Image Script. Proprietary Texas Instruments 32-bit, little-endian format for processor bootup scripts. It holds binary images of runtime programs for storage in non-volatile memory. The TI boot process unpacks, verifies, loads, and starts these programs. TI provides a tool, genAIS, for creating them.; (2); Automatic Identification System. A tracking system for seafaring vessels that reports location, speed, vessel identification, and other information. It was originally for collision avoidance, but the data serves many other purposes. The signals are at 161.975 and 162.025 MHz, using NRZI signaling, 9600 b/s data rate, and GMSK modulation. They’re received by satellites as well as shore stations.
AIX: Advanced Interactive Executive. IBM’s open version of Unix, introduced in 1990.
ALAC: Apple Lossless Audio Codec. Apple’s lossless-compression codec (coder/decoder) for digital sound. It went open-source in 2011, but employs an MP4 wrapper, which includes support for DRM. For this reason, the files have the .m4a filename extension. Compare FLAC.
A-law: See companding.
aliasing: An effect of digitally sampling a RF signal at a rate less than half its bandwidth, a practice called under-sampling. Components of the original signal wrap around the Nyquist rate, appearing in the sampled signal at a much lower frequency. This is generally undesirable, but it can be exploited to make the ADC also serve as a downconverter.
Alpha: A 64-bit superscalar, superpipelined RISC microprocessor series, part of the microVAX family from DEC. DEC renamed this series Alpha AXP so they could copyright the name. Officially, AXP doesn’t stand for anything. After buying DEC, HP discontinued the series. Family members:; 21064 – (1992) EV4 core architecture (0.75 µm process) and 150-166 MHz clock, later EV4S and 200 MHz clock.; 21064A – EV45 core (meaning EV4 on 0.5 µm process), 233-300 MHz clock.; 21066(A) – Unsuccessful x86-compatible versions of the 21064 and 21064A.; 21164 – Workstation CPU with EV5 core, 0.5 µm process, 333, 466, & 600 MHz speeds, x86-emulation, 2 MB L3 cache SRAM, 128-bit data path, 96 kB L2 on-chip cache, and 16 kB L1 cache. A cheaper 1997 version meant for PCs, the 21164PC, runs at 400-533 MHz.; 21164A – A 21164 on the 0.35 µm process, 417 MHz clock.; 21264 – (1998) 500+ MHz, EV6 core, 0.35 µm. The 21264PC ran at 731 MHz as of May 2000.; 21364 – Released by Compaq in 2000, EV7 core, speeds of 1 to 1.1 GHz and up. Has 1.5 MB L2 cache, and Direct Rambus interfaces from chip to main memory. Also, 21364PC for PCs.; 21464 – EV8 core, 130 nm process, never released.
alpha particle: A helium (He) nucleus, α, consisting of two protons and two neutrons, traveling at sufficient speed to be classed as ionizing radiation (IR). Typically, that’s at least 4% of the speed of light.
ALS: Advanced Low-power Schottky. A variant of TTL.
Alt: Alternate. A standard key on English-language keyboards. Holding down the Alt key while pressing some other key or series of keys can perform a special action in the current application. For example, in MS Windows, the combination Alt-F4 closes the application. Compare Ctrl.; On MS Windows systems, Alt plus a four-digit code entered on the keyboard’s number pad (not on the row of number keys just above the letter keys) types the ASCII, extended 8-bit ASCII, or Unicode character corresponding to the code. Depending on the application, it might be necessary to first toggle the Num Lock key on for this to work.; The table below shows some of these special characters and the codes that produce them. Not all Windows applications accept codes past 0127 (the end of 7-bit ASCII) or past 0255 (the end of whatever 8-bit extension of ASCII the application is using), even if capable of displaying the character.
alternator: From “alternating current generator”. Any generator that produces AC, although in modern usage the term refers only to small, dedicated types, especially in vehicles; large or general-purpose types are just called generators. The typical alternator rotates a magnet (the rotor) within a stationary iron ring wound with conducting coils (the stator). The magnet is positioned so that its magnetic field cuts across the wire coils, inducing current in them as it moves. Where DC is required, modern alternators feed into rectifier circuits. See dynamo.
ALU: Arithmetic Logic Unit. A microprocessor section that carries out integer instructions – not only integer math but also logical operations such as AND, OR, ASL, etc. Contrast with FPU.
AM: (1); Amplitude Modulation. The oldest and simplest of the three types of RF analog modulation (FM and PM are the others): carrier wave amplitude varies with the amplitude of the modulating signal. In North America, the commercial AM radio broadcast band is 535-1605 kHz. Carrier frequencies are assigned every 10 kHz, starting at 540 kHz.; (2); Additive Manufacturing. Business-speak for 3D printing.
AMA: Ask Me Anything. A public forum, typically online, where an individual or group responds to real-time questions.
AMD: Advanced Micro Devices. A big chip company and sole surviving rival to Intel in the PC microprocessor market as of 2008. Some of their microprocessors are listed below. AMD doesn’t use the designations “Sempron 64” and “Athlon 64”, but they’re useful to distinguish 64-bit from earlier 32-bit processors with the same name.; 5x86 – Really a high-speed 80486 equivalent, with 133 MHz clock and 33 MHz system bus.; K5 – A late-to-market 5th generation x86 chip, running at 75-133 MHz with system bus at 50, 60, or 66 MHz (2/3 or 4/7 core speed).; K6 – (1997) 6th generation x86 chip, based on NexGen’s Nx686, with MMX support. Pin-compatible with P54C. Has a total of 64 kB L1 caches, external L2, 0.35µm process, 166/200/233 MHz. The K6-2 runs at 350 to 533 MHz, has a 100 MHz external bus, and was the first chip to incorporate a 3D graphics instruction set. The K6 III runs at 450MHz plus. The K6-2 and III interface with Super Socket 7 on the board.; Duron – A low-cost 32-bit processor with a 180 or 130 nm K7 core process, 1.5 V, Socket A. It has clock speeds of 1.0 to 1.8 GHz, a 200 to 266 MHz FSB, 128 kB L1 cache (64 kB each instructions and data), and 64 kB L2 cache. Phased out in 2004.; Athlon – 7th generation, 32-bit K7 core x86 chip on a cartridge with a Slot A or 462-pin Socket A interface. Debuted on 180 nm process at 700-1200 MHz, with 200 MHz FSB, 128 kB L1 cache (64 kB instruction/64 kB data), 256 kB on-chip L2 cache, and 11-stage pipelining. Later versions are the Athlon MP (for servers), Athlon XP (for desktops), and Athlon XP-M (for notebooks), all on 130 nm process. MP has clock speeds to 2225 MHz. XP comes in 1.47/1.53/1.67/1.8/2.0/2.13 GHz with 266 MHz FSB, 1833-2167 MHz with 333 MHz FSB and 512 kB L2, and to 2.2 GHz with 400 MHz FSB and 512 kB L2.; Sempron – Budget version of whatever the flagship 32-bit processor is, replacing Duron.; Opteron – Launched (4/2003) as a competitor to Intel’s Xeon and Itanium 2, based, like the Athlon 64, on X86-64 architecture and 130 or 90 nm process, with clock speeds 1.4 to 2.0 GHz. The later 45 nm quad-core version has shared 6 MB L3 cache, L1 and 512 kB L2 cache for each core, 2.3-2.9 GHz clock, uses Socket F and DDR2 memory, and has TDP of 75W. The 2009 six-core 45 nm version has 2.2-2.6 GHz clock, 2.4 GHz HyperTransport clock, and a decoupled 2.2 GHz north bridge/L3 cache clock. It’s still 75W, and still uses Socket F and DDR2.; Sempron 64 – 64-bit on 130, 90, or 65 nm process. 1.6-2.3 GHz clock, 1.2-1.4 V, 128-512 kB L2 cache, 128 kB L1, 1600-2000 MHz FSB, Socket 754 or AM2 or S1.; Athlon 64 – Debut 9/2003. A 64-bit, 8th-generation processor based originally on the X86-64 architecture and 130 nm process. Later versions went to 90 or 65 nm process, and featured 1.8-2.8 GHz clock, 1600-2000 MHz FSB, 128 kB L1 cache (64 kB instructions/64 kB data), 512-1024 kB L2, 1.25-1.5 V, on-chip memory controller, and Socket 754 or 939 or AM2.; Turion 64 – Notebook processor, 64-bit, 90 nm, 1.6-2.4 GHz, 128 kB L1 and 512-1024 kB L2 cache, Socket 754. The dual-core version is the Turion 64 X2, with each processor getting half the L2 cache rather than sharing the whole; it uses 638-pin Socket S1.; Athlon 64 FX – 64-bit processor with 2.2-3.0 GHz clock, 1600-2000 MHz FSB, 1.3-1.5 V, 128 kB L1, 1024 kB L2, 90 or 130 nm process, Socket 939 or 940, or Socket F for dual-core versions.; Athlon 64 X2 – Dual-core on 90 nm or 65 nm process, 1.9-3.2 GHz clock, 2000-2400 MHz FSB, 2× 128 kB L1, 2× 512 or 1024 kB L2, 1.025-1.4 V, Socket 939 or AM2 or AM2+. TDP is 65-95W, 45W for low-power types. Some models have 2MB shared L3 cache.; Phenom – 65 nm process and 1.0-1.3 V power, uses Socket AM2+. The X3 Triple-Core has 2.1-2.4 GHz clock, and 3.6 GHz FSB. The X4 Quad-Core has 1.8-2.6 GHz clock and 3.2-4.0 GHz FSB. Both have one 128 kB L1 and 512 kB L2 cache per core, and a shared 2 MB L3 cache.; Phenom II – (2009) 64-bit, 45 nm process using SOI and immersion lithography. The X2, X3, and X4 types have 2, 3, and 4 cores respectively, and TDP from 80W to 125W (65W for low-power models). 2.4-3.1 GHz clock, L1 cache 128 kB per core, L2 cache 512 kB per core, shared L3 cache 4-6 MB. Debuted for Socket AM2+ and DDR2 RAM; later versions are for Socket AM3 and DDR3 RAM, but retain backward compatibility.; Athlon II – (2009) 64-bit, 45 nm process, 938-pin, Socket AM3 (compatible with AM2+), uses DDR2 or DDR3 RAM. Debuted as dual-core X2 chip with 3.0 GHz clock, 65W TDP, L1 cache 128 kB per core, L2 cache 1 MB per core, no L3 cache.; Llano – (2011) 64-bit, 32 nm process, Socket FM1, 2 to 4 cores, 2.1-2.9 GHz clock, 1-4 MB L2 cache. One of the Fusion family of APUs, meaning the chip has integrated Radeon 63xx to 66xx series graphics. The graphics can cooperate with a discrete video card of the same series.; Bulldozer – (October 2011) Code name Zambezi. 64-bit, 32 nm SOI process, 940 pins fitting Socket AM3+, 4 to 8 cores, 2.8-4.2 GHz clock, L2 cache 2 MB per core, usually 4 or 8MB L3 cache. Works with up to 1866 MHz DDR3 SDRAM. TDP is 95 or 125 watts.; Piledriver – (2012) A minor improvement on Bulldozer with the same stats, code name Vishera.; Trinity – (2012) Second-generation APU, with on-chip Radeon HD 7000 series graphics. 64-bit, 32nm process like the 1st-gen Llano, but with improved performance and power consumption. It uses either one or two Piledriver modules, each of those being dual-core, and fits the FM2 socket.; Richland – (2013) An upgrade of the Trinity APU family with on-chip Radeon HD 8000 series graphics. 64-bit, 32nm process, 2 or 4 cores, 3.2-4.1 GHz base clock (with boost speeds as high as 4.4 GHz), 1-4 MB L2 cache, 65W or 100W TDP, and socket FM2.; Steamroller – (2014) 3rd generation of the Bulldozer architecture on 28nm process. Its Kaveri APUs have four processor cores, 3.5-3.7 GHz clock, 2 × 2MB L2 cache (each L2 shared by two cores), and 95 W TDP. They use DDR3 memory and the FM2+ Socket.; Carrizo – (2015) APU architecture on 28 nm process, using Socket FM2 or FM2+. Primarily aimed at laptop PCs.; Ryzen – A family of 64-bit multi-core CPUs using iterations of the Zen microarchitecture. Few members of this family have integrated graphics. Those that do are marketed as APUs.; Ryzen 1000 – (2017) Uses first-generation Zen architecture on 14-nm process, with support for DDR4 SDRAM and PCIe 3.0. Each core has 68 kB L1 cache and 512 kB L2 cache. Models made for the 1331-pin Socket AM4 are the Ryzen 3 (4 cores, 8MB L3 cache, base clock 3.1 to 3.5 GHz, TDP 65 W), Ryzen 5 (4 or 6 cores, 8 or 16MB L3 cache, 3.2 to 3.6 GHz, TDP 65 or 95 W), and Ryzen 7 (8 cores, 16MB L3 cache, 3.0 to 3.6 GHz, TDP 65 or 95 W). The more powerful Ryzen Threadripper processors (8 to 16 cores, 16 or 32 MB L3 cache, 3.2-3.8 GHz, TDP 140 or 180 W) use Socket sTR4.; Ryzen 2000 – (2018) Zen+ architecture on 12-nm process, with each core having 96 kB L1 cache and 512 kB L2 cache. Still using Socket AM4, or sTR4 for the Threadripper processors.; Ryzen 3000 – (2019) Zen 2 architecture, still on 12-nm process and using AM4 socket, but adding support for PCIe 4.0.; Ryzen 4000 – (2020) Zen 2 architecture again, but supporting only PCIe 3.0.; Ryzen 5000 – (2020) Zen 3 architecture on 7-nm process. PCIe 4.0, except for the APU versions. 512 kB L2 cache per core.; Ryzen 6000 – (2022) Zen 3+ architecture on 6-nm process, intended for the notebook market. Supports PCIe 4.0 and DDR5 SDRAM.; Ryzen 7000 – (2022) Zen 4 architecture on 5-nm process, with support for PCIe 5.0 and DDR5 SDRAM (no DDR4). 1 MB L2 cache per core.
AMD-V: AMD Virtualization. Also called SVM. AMD’s technology for creating a virtual machine to act as a security sandbox. Compare VT-x.
AMI: Alternate Mark Inversion. See NRZI, RZ-AMI.
Amiga: See Commodore.
AMLCD: Active-Matrix Liquid Crystal Display. See LCD.
AMOLED: Active Matrix Organic Light-Emitting Diode. See OLED.
AMP: (1); Asymmetric Multi-Processing. A shared-memory multiprocessing technology.; (2); Accelerated Mobile Pages. A 2015 Google technology to create Web sites specifically for the small screens of mobile phones. The content is presented, and therefore controlled, by Google, not by the source site, a detail that has drawn widespread criticism – especially paired with the fact that Google puts AMP-compliant sites in a special display above other search results.
ampere: Also amp, or A. Named for French scientist André-Marie Ampère (1775-1836). The standard unit of electrical current, and one of seven base units in SI.; Ampère’s law establishes that, given two parallel conductors of infinite length 1 m apart in a vacuum, 1 ampere is the amount of current in both that would produce a force between them equal to 2 × 10^-7 newtons/meter. If both currents are flowing in the same direction, the force is positive, hence attractive; if directions are opposite, the force is negative, hence repulsive. The amount of charge this current moves past a given point in a second is defined as 1 coulomb.; Electrons in a wire travel surprisingly slowly, taking more than a minute to cover an inch of distance.
AMPS: Advanced Mobile Phone System. From 1983 to the early 2000s, a US analog mobile phone technology using FM and FDMA signaling. It has 832 channels with 30 kHz channel spacing. Receiving channels occupy the 869-894 MHz spectrum, and transmitting channels are at 824-849 MHz.; Digital 2G and 3G systems, with their cost, security, and service advantages, put AMPS out of business by 2008.
AMR: Automatic Meter Reading.
AMS: Analog/Mixed Signal. A design method for circuits that contain analog elements. Software tools for AMS design can divide a circuit into blocks, and simulate each block at a different level of abstraction – from individual transistors, to cells, to functions or behaviors that encompass many transistors or cells. This makes them much faster and more flexible than traditional analog simulators, such as SPICE, which simulate only at the transistor level. See FPAA, Verilog HDL, VHDL.
AMT: Active Management Technology. A feature of Intel’s Management Engine that supports remote (i.e., over a network) management of PC security.
Android: Google’s Linux-derived, open-source operating system for mobile wireless platforms. Compare iOS.
angstrom: A unit of length equal to 10^-10 m, or 1/10^th of a nanometer, usually abbreviated Å.
ANN: Artificial Neural Network. Also called a deep neural network (DNN). A computer simulation of a collection of neurons. Each simulated neuron is a separate program that exchanges data and decisions with others. The state of a neuron depends on the sum of its inputs, with each input weighted in proportion to its significance as determined by the model-training process, and an activation function that it applies to this weighted sum to make a decision. Its output state then becomes input for other neurons. The neurons are grouped into layers, with each neuron connected only to neurons in the layers before and after it.; Advanced ANNs simulate learning with a combination of non-deterministic behavior – that is, using probabilities rather than absolute rules to make decisions – and adjusting decision probabilities based on past results. This ability of a program to make iterative changes to itself to better analyze its environment is called deep learning, and is currently (2023) the most promising way forward for artificial intelligence. Popular approaches to neural network design include the multilayer perceptron (MLP), the long short-term memory (LSTM) architecture, and the transformer model. All of them are computationally demanding and require enormous quantities of data to train them, meaning AI is still too expensive for many of the tasks that it could be applied to.; A large language model (LLM) is an ANN that’s been trained to understand natural-language queries and to respond in kind. The LLM powering ChatGPT was among the first.; Researchers have demonstrated neural networks with many of the simulated neurons altered to embed malicious behavior. Anti-malware programs don’t detect this.
anode: The terminal or electrode at which current enters a system, hence where electrons leave a system. If the system is a battery, fuel cell, or other DC source, the negative terminal is the anode. In a circuit, it’s the positive terminal, i.e. the current source. Compare cathode.
ANSI: American National Standards Institute. A very big standards-setting body. Sometimes used to mean an 8-bit expansion of ASCII used on English-language PCs; see data code for details.
antenna: The key parameters for an antenna are gain, beam pattern, polarization, and impedance, all at specified frequencies. Antenna gain is the peak value in the optimum direction, normally in dBi or dBd, and is proportional to antenna aperture size. If gain is given in plain dB, this normally means dBd.; Most antennas use electromagnetic (EM) waves. However, they can instead be designed for magnetic fields or even acoustic waves. These have advantages in certain applications. For example, acoustic waves have much shorter wavelengths than EM waves at the same frequency, because sound travels so much slower than light, so acoustic antennas can be much smaller. However, they require piezoelectric transducers to convert sound into electrical signals.; Non-directional (omnidirectional) antennas are the simplest. Directional antennas use arrays of passive elements, or incorporate active elements. Passive antennas are reciprocal, meaning the same gain applies to transmission and reception. Active antennas have electronic components that interact with the signal – chiefly amplifiers or (for transmission) oscillators, although as of 2022 researchers are considering the reconfigurable intelligent surface (RIS) as a way to boost gain of directional signals. See dB, isotropic, VSWR.; quarter-wave dipole – Also called a whip, monopole, or Marconi antenna. A straight, grounded antenna with length equal to one-fourth the wavelength of the signals it optimally receives or transmits.; A half-wave antenna is the shortest practical length for radiating into free space, but the quarter-wave dipole uses the phenomenon of ground reflection to effectively double its length. With a good earth ground (the classic Marconi design), it has an impedance of about 36 Ω. An artificial ground provided by drooping quarter-wave radials (a ground plane antenna) yields about 50 Ω. Artificial grounding on a large, reflective, metallic surface (e.g., a vehicle) is also common. The surface should extend at least one wavelength from the antenna. Incorporating a loading coil (inductor) into the antenna increases its electrical length, allowing it to be physically shorter but reducing gain slightly. If this allows it to be placed in a better ground, the tradeoff is worthwhile.; Maximum radiation and reception is in the plane normal to the antenna’s axis. This means that, when vertically mounted, which it usually is, it has an omnidirectional pattern with about 2.15 dBi gain for vertically polarized signals, and 3 dB beamwidth of about 55° (i.e., a toroid extending about 27.5° above and below the horizontal).; half-wave dipole – A straight, ungrounded antenna with length equal to half the wavelength of the signals it optimally receives or transmits. Independence from a ground plane makes it useful for portable radios, but it should be at least a quarter-wavelength above ground. It’s separated in the center by an insulator, and center-fed from a transmission line, commonly through a balun if the feed line is unbalanced. Impedance is about 75 Ω. Gain and pattern are typically the same as for a solidly grounded quarter-wave dipole, although some types achieve higher bi-directional gain (up to about 7 dBd). Often placed horizontally.; folded dipole – Two parallel half-wave dipoles connected at the ends, with one of them fused (no insulator in the middle) and the other fed at the center. Mainly used for TV signals, for its large bandwidth. Impedance is about 300 Ω.; co-linear – A cluster of parallel dipoles, typically with 3 to 4.5 dBd gain. Sometimes coaxial as well, i.e. arranged in a line.; loop – Inefficient, but useful for confined spaces and direction finding.; log-periodic – A directional, ungrounded, broadband antenna consisting of an array of active half-wave dipoles, usually with the shortest in front and the longest in back. Expect 7 to 10 dBi gain. Versions designed for the VHF/low UHF range are popular for TV reception.; Gray-Hoverman (GH) – Original patent 1959 by American engineer Doyt Hoverman (1913-1989). A UHF/high-VHF antenna optimized for receiving ATSC digital TV signals.; Yagi – Properly the Yagi-Uda, named for its Japanese inventor (Professor Uda) and the physicist who translated the design into English (Yagi). A high-gain directional antenna with three or more half-wave dipole elements – one driven, one reflector, & one or more directors – parallel to one another along a boom. The driven element is connected to the transmission or reception line, and can be a folded or a standard dipole. The other elements are isolated, and spaced at about 1/10th the target wavelength. The reflector is behind and slightly longer than the driven element, while the directors are in front and slightly shorter. Increasing the number of director elements increases antenna gain and narrows beamwidth (from 90° down to 20°).; aperture – Includes slot, parabolic reflector, and waveguide horn.; patch – A flat, directional type with optimum gain normal to its surface. Different designs for a given frequency can have different surface area, but need roughly the same volume to have the same gain.; array – Contains multiple elements. It can be linear, circular, rectangular, etc. See phased array.
antipodal: Refers to signal, codeword, or vector pairs that are opposites, such as sin(ω₀t) and –sin(ω₀t), so that their cross-correlation product (the integral of their product over time) is –1. An antipodal signal set is therefore restricted to the binary case, e.g. BPSK. Compare orthogonal.
AO: Adaptive Optics. Technology for correcting optical imagery to reduce the distortions caused by atmospheric turbulence and other interference, e.g., taking the twinkle out of starlight.
AOI: Automated Optical Inspection. Having a machine inspect a finished product.
AP: Access Point. An area with accessible wireless networking – pretty much the same as a hotspot.
APD: Avalanche Photo-Diode. See diode.
API: Application Programming Interface. The software (i.e., programming, not user) interface to a computer resource or capability. The resource or capability accessed by an API is called an endpoint.; Modern operating system kernels provide the APIs for essential hardware and software. Other software and (especially external) hardware require their own separately installed APIs. The abstraction provided by an API means that programmers don’t have to know the particulars of a resource to develop programs that use it.; Central to the design of an API is minimizing the number of function calls to the host device/system to perform any given task. Some examples of APIs:; IAPI – Internet API.; IDAPI – Integrated Database API. Promoted by Borland, IBM, Novell, and WordPerfect in 1992 to counter Microsoft’s ODBC.; JTAPI – Java Telephony API. Precisely what it sounds like.; MAPI – Messaging API. A Microsoft interface that provides a common set of function calls between client app or API and network messaging system.; SAPI – Speech API. Microsoft. Allows any speech engine to run on Windows.; TAPI – Telephony API. Microsoft. Allows Windows apps to access telephony functions.; TSAPI – Telephony Services API. Novell/AT&T rival to TAPI.; Win## – See Windows API, below.; Windows API (or WinAPI) – For creating software to run under the Windows OS. The original implemented 16-bit instructions on 16-bit processors. It was retroactively named Win16 due to the arrival of 32-bit Win32. The 64-bit implementation (with 32-bit compatibility) is sometimes called Win64, especially when being compared with earlier versions. It uses 64-bit memory addresses, and abstract data types that can be compiled into either 32-bit or 64-bit types depending on the system.; WinRT – A 32-bit Microsoft API introduced in 2012 as the sole API for Windows RT, and also available in Windows 8.; OpenAPI Specification and REST are not themselves APIs, but standards for creating and describing Web APIs. See also DirectX and OpenGL.
APM: Advanced Power Management. 1990s Intel/Microsoft technology for OS control of a computer’s power level. It recognizes five states, in descending order of power consumption: full on, APM enabled, APM standby, APM suspend, and off. ACPI replaced it.
APON: ATM-based Passive Optical Network. See PON.
APOP: Authenticated Post Office Protocol. A command feature added to POP to encrypt user name & password.
APPN: Advanced Peer-to-Peer Networking. A predictive network technology.
APT: Advanced Persistent Threat. Buzzword for criminal, foreign-government, or hacktivist organizations that employ sophisticated malware and a patient, long-term approach to infiltrate high-value online targets.
APU: Accelerated Processing Unit. AMD’s term for its Fusion multi-core processors with on-die integrated graphics – that is, a union of MPU with GPU – beginning with the Llano series (2011).
AR: Augmented Reality. Sometimes called extended reality (XR) or mixed reality (MR). In contrast to virtual reality, which presents the user with an artificial environment, AR adds data and data-gathering capabilities to the user’s view of the real world through a HUD, audio, overlays, etc.
Archie: From “archive”. An obsolete client/server application used to index FTP documents by title and keyword, and make them available via anonymous FTP hosts. Archie servers develop very large databases, which can be accessed via Telnet, Archie client applications, or even E-mail. See Gopher.
ARCnet: Attached Resource Computer network. Datapoint Corporation’s token-bus LAN for up to 255 nodes. The original runs at 2.5 Mb/s on RG-62 coaxial cable. The newer ARCnet Plus goes to 20 Mb/s on fiber.
Arduino: Designed in Ivrea, Italy, and named for Arduin of Ivrea, briefly king of Italy (1002-1004 AD). An open-source microcontroller module on a single circuit board for student and hobbyist firmware projects. It has an 8-bit Atmel microcontroller and multiple I/O connections, and comes with its own IDE, software library, and C-like programming language. A number of manufacturers produce Arduino kits. Compare Raspberry Pi.
ARIN: American Registry for Internet Numbers. See RIR.
ARINC: Aeronautical Radio, Inc. The original name of a company that developed a number of widely used data communications standards, including:; ARINC 429 – A simplex, asynchronous data bus with speeds from 12.5 to 100 kb/s using bipolar RZ format and 10V differential signaling over a twisted-pair line. Each bus has just one transmitter and up to 20 receivers. Very common in commercial aircraft, it’s cheaper to implement though less capable than MIL-STD-1553. ARINC 429 uses 32-bit data words, with a minimum of four null bits (zero voltage) between words for multi-word messages. A word consists of five fields: octal data-type label (bits 1-8), Source/Destination Identifier (bits 9-10), data (bits 11-29), Sign/Status Matrix (bits 30-31), and odd parity (bit 32). The standard use of the SSM field is to indicate one of four cases for the message data: normal operation (NO or NOP), functional test (FT), failure warning (FW), or no computed data (NCD). Some implementations don’t use SDI and SSM, increasing data field size to 23 bits. Unused bits are set to 0. The bus sends the label bits first and in reverse order to the rest of the word, as shown in the figure.; ARINC 664 – Specifies the use of an IEEE-802.3 (Ethernet) network as an avionics data bus. AFDX is an implementation of ARINC 664.; ARINC 717 – A more secure data bus used for flight data recorders.
ARM: Advanced (originally Acorn) RISC Machine. A family of (relatively) simple, low-power RISC microprocessor architectures. Also means the UK-based ARM Holdings, which designs but doesn’t manufacture the chips. The original 16-bit ARM1 (1985) had about 25,000 transistors, and used less than 0.1 W. The 32-bit StrongARM (1990s) developed in partnership with DEC ran at 200 MHz, and used less than 1 W.; ARM creates extensions of its newer 32-bit architectures, such as ARMv7’s Cortex family, and manufacturers implement these designs with additional changes. Some have multiple cores, and clock speeds topping 1 GHz. Their continuing focus on low power makes them popular for netbooks, smartphones, and embedded devices. Microsoft used a quad-core implementation of the 32-bit ARMv7 for its Surface RT tablet (2012), which runs Windows RT exclusively. AMD is developing server processors based on 2011’s 64-bit ARMv8 architecture.
ARP: Address Resolution Protocol. Part of the TCP/IP protocol suite at the Link or Internet layer (layers Data link or Network of the OSI model), ARP queries hardware (MAC) addresses of network nodes, and maintains a database mapping these MAC addresses to node IP addresses.
ARPA: Advanced Research Projects Agency. Government agency dedicated to identifying and promoting strategically transformational technologies. They launched ARPANet, the forerunner of the Internet, in 1969. They changed their name to DARPA (Defense ARPA) in 1972, back to ARPA in 1993, and back again to DARPA in 1996. They’re at https://www.darpa.mil.
ARQ: Automatic Repeat Request. A signal to a transmitter to retransmit a block of data.
ASCII: American Standard Code for Information Interchange. See data code for details. ASCII art refers to images made from ASCII characters, e.g.:
ASF: Advanced Systems (originally Streaming) Format. Microsoft proprietary audio-video container format for streaming media, using the .asf extension. It specifies the structure of the A-V data stream, not that of the source coding, but the latter is usually Microsoft proprietary WMA or WMV.
ASIC: Application-Specific Integrated Circuit. An IC chip designed for a particular system rather than a general range of functions. Typically, a system on an ASIC is faster, less power-hungry, and (in large volumes) cheaper to manufacture than the same system on a PLD. It’s not unusual, therefore, for a design to be developed and proved using a FPGA and then manufactured as an ASIC. ASICs are, however, more expensive than low-end microcontrollers, so they are not often used for embedded devices despite the potential power savings.
ASK: Amplitude Shift Keying. Modulating the amplitude of an RF carrier to convey analog fax or digital data. Obsolete. Non-discrete ASK is the same thing as AM (amplitude modulation).
ASL: (1); Above Sea Level.; (2); Arithmetic Shift Left. See binary logic.
ASLR: Address-Space Layout Randomization. An OS security feature that assigns processes to random memory locations, rather than letting them use the same addresses every time they run. This makes them harder targets for malware trying to inject malicious code. Microsoft started using it with Windows 7. See also DEP.
ASM: Abstract State Machine. See state machine.
asm.js: See JavaScript.
ASP: (1); Active Server Pages, originally. A Microsoft standard for using scripts to generate HTML dynamically. Microsoft’s VBScript is the default scripting language for ASP, but VBScript output is not always compatible with non-Microsoft browsers, so ASP developers can specify another language. When serving a Web page containing ASP, a server first passes the embedded scripts to its ASP engine, which runs them and generates the HTML content. Since the result is plain HTML, the requesting browser doesn’t need an ASP engine.; ASP began as an add-on to Microsoft’s IIS around 1998. The more advanced ASP .NET brings Microsoft’s .NET tool set and Common Language Runtime (CLR) environment to Web page development. A page using ASP can have the .asp or (for .NET) .aspx filename extension instead of .html, but this isn’t required.; (2); Application Software (or Service) Provider. A company that maintains and manages software applications for corporate customers. See MSP.
ASR: Arithmetic Shift Right. See binary logic.
assembler: A program that converts human-readable assembly language source code into machine language. An assembly language is a low-level programming language, i.e. specific to a particular type of microprocessor. Programs written in assembly have traditionally yielded faster, more efficient machine code than programs compiled from high-level languages, but smarter compilers are cutting into this advantage. See also linker.; A disassembler works the other way, converting machine language from a running program into assembly.
asynchronous: When referring to data protocols, this means one that allows bits to be transmitted at arbitrary times using an agreed-upon clock rate. Data must be framed by start and stop bits. Compare synchronous.
AT: (1); Advanced Technology. IBM’s 1984 PC-AT, with the Intel 80286 processor, IDE hard disk drive interface, 16-bit ISA system bus, 16 MB RAM, 5¼" 1.2 MB floppy disk drive, and MS DOS 3.0. It was the first machine to run MS Windows 1.0. The AT motherboard could be up to 12 × 12 inches. See PC.; (2); Attention. See AT command set.
ATA: Advanced Technology Attachment. Commonly but incorrectly called IDE, this family of data bus specifications based on Western Digital’s IDE technology was standard for PC hard disk drives from the 1980s into the late 2000s before giving way to SATA. It’s now called parallel ATA (PATA). It uses a 2-inch wide, 40- or 80-wire IDE ribbon cable with either two or three 40-pin connectors. The AT in the name refers to the IBM 80286 PC-AT, the machine on which it was first used.; The original ATA, finally standardized in 1994 and retroactively named ATA-1, supports one or two hard drives, up to 528 MB each. Subsequent versions of this standard are listed below.; ATA-2 – This improvement on ATA-1 supports CD-ROM drives, hard drives up to 8.4GB, and data rates of 4-16 MB/s. Unofficial terms for this standard (and some of its proprietary kin) include EIDE, Fast ATA, Fast ATA-2, and Fast IDE.; ATA-3 – (1997) A minor revision of ATA-2.; ATA/ATAPI-4 – (1998) The first standard to incorporate ATAPI. It specifies Ultra DMA modes 0, 1, and 2 (16.7, 25, and 33 MB/s), and introduces an optional 40-pin, 80-wire IDE cable (the 40 additional wires are for grounding to eliminate crosstalk). If the old 40-wire IDE cable is used for mode 2 (33 MB/s), it can be no longer than 18 inches. The terms UDMA, UDMA/33, DMA mode 33, ATA/33, Ultra ATA, and Ultra DMA are all used (incorrectly) to refer to this standard and to compliant devices. See DMA.; ATA/ATAPI-5 – (1999) This standard introduces Ultra DMA modes 3 and 4 (44.4 and 66.7 MB/s). In mode 4, the previously optional 80-wire IDE cable is mandatory; using the older 40-wire cable causes the device to default to mode 2. Again, ATA/66, Ultra ATA/66, and Ultra DMA/66 are common marketing terms for compliant devices.; ATA/ATAPI-6 – Similar to ATA/ATAPI-66, with slightly higher speed (Ultra DMA mode 5 at 100 MB/s) and enhanced error checking. Naturally, these devices are being called Ultra ATA/100.
ATAPI: Advanced Technology Attachment Packet Interface. A protocol to support devices other than HD drives – especially CD-ROM drives – on the ATA bus.
AT bus: Advanced Technology bus. IBM’s trademark name for the old ISA expansion bus.
AT command set: A set of ASCII text commands used to control most dial-up modems, based loosely on the command set for the 1977 Hayes Smartmodem 300. The eponymous AT (short for “attention”) precedes most commands.
ATE: Automated Test Equipment.
Athlon: See AMD.
ATL: Active Template Library. A Microsoft library (e.g. atl.dll, atl70.dll) of C++ classes for writing COM and ActiveX objects rather than full applications. See MFC, .NET, STL, WTL.
ATM: Asynchronous Transfer Mode. An obsolete network technology that transmits digital signals in 53-byte packets (cells), each consisting of a 5-byte address and 48 bytes of information. “Asynchronous” refers to the fact that the signals are multiplexed by the address labels rather than by time. Being time-critical, voice and video signal packets have priority. In 1993, LAN bandwidth varieties included 45, 100, and 155 Mb/s full duplex.; ATM is intended for running a wide range of services and protocols over one network. It creates virtual circuits for routing, as opposed to the connectionless store-and-forward of the standard packet-switched (e.g. IP) network. However, it lost out to IP and other networking and broadband technologies.
atmosphere: The gaseous shell around Earth comprises five layers defined by temperature, hence not including the ionosphere. Weather happens only in the lowest two layers: the troposphere (7-17 km thick, thinnest over the poles), and the stratosphere (to 50-55 km). Above that are the mesosphere, the thermosphere, and finally the exosphere.; In the 1970s, perfectly dry, perfectly clean air at sea level consisted by volume of about 78.03% nitrogen (N₂), 20.99% oxygen (O₂), 0.94% argon, and 0.033% CO₂, with traces of methane (CH₄), ozone (O₃), and the noble gases Ne, He, Kr, and Xe. As of 2021, CO2’s share has risen to about 0.041%, and is still rising. Realistically, there’s always some amount of pollutants present, along with water vapor. The latter varies from 0.3% to about 4% of total volume, averaging 0.8%.
ATSC: Advanced Television Systems Committee. The US digital television (DTV) standard, as well as the name of the working group that established it. Compare DVB, ISDB, NTSC.; ATSC uses the same 6 MHz wide channels defined for NTSC, but in place of the analog carriers, it has a broadband digital signal with a 310 kHz pilot tone, carrying 19.2895 Mb/s of data after overhead. It also uses the old NTSC spectrum, with channel numbers 2-13 (VHF) and 14-83 (UHF), although channels 37 and 52-69 are unavailable. Each channel can have sub-channels, denoted as #-1, #-2, #-3, etc., where # is the channel number.; The changeover from analog to digital led to the phenomenon of the virtual channel: A station would broadcast ATSC in a new physical channel, but with a Program and Station Information Protocol (PSIP) identifier giving some other channel number – normally the station’s original NTSC channel. This other channel is the virtual channel. The digital receiver scans its environment and creates a virtual channel map of all such signals, so the end user doesn’t need to see or know about the physical channels used.; Many stations did this rather than changing their original channels to digital. It allowed them to A) begin digital broadcasting while continuing analog service, and B) use higher-frequency channels (which require more power but use smaller antennas and are less susceptible to interference) without giving up the channel numbers that viewers know them by.; The digital carrier uses 8-level, trellis-coded, vestigial sideband modulation (8VSB, or 8T-VSB), a type of VSB in which the upper sideband is switched rapidly through eight amplitude levels. The data is video coded with MPEG, and audio with Dolby AC-3. 8VSB is more vulnerable than DVB’s COFDM to EMI and rapidly changing multipath, creating problems with reception in moving vehicles. On the other hand, it achieves a higher data rate with lower power.; ATSC’s MPEG-2 video signals use 18 different formats, defined by just three properties:; The shorthand for signal format is pixel height, scan type, and fps. For example, Blu-ray uses 1080p24.; The four pixel resolutions for signals are 640 × 480, 704 × 480, 1280 × 720, and 1920 × 1280. The first two are used for standard definition and enhanced definition TV (SDTV and EDTV). SDTV is equivalent to analog NTSC, with the old 4:3 aspect ratio. EDTV (typical for DVD players) has slightly higher quality because it doesn’t use interlaced scan; some EDTV signals achieve the high-definition TV (HDTV) 16:9 ratio using 704 × 480 resolution and non-square pixels. The 1280 × 720 and 1920 × 1280 (2K) signals are HDTV, exclusively in 16:9.; Manufacturers and vendors describe TV display equipment by just its native pixel height and scan type: 480p, 480i, 720p, 1080i, 1080p. (There is no 720i, since no ATSC 1280 × 720 signal uses interlaced scan. Interlacing in any case is falling out of use for TVs, just as it did for PC monitors.) If a system has non-standard physical resolution – e.g. 1366 × 768, 1024 × 768, or 852 × 480 – they’ll advertise it as one of the standards. This sounds like a cheat, but it’s not. DTV sets are expected to display most or all ATSC formats, so they’re usually scaling the signal anyway to make it fit the physical screen without distortion. PCs that display DTV have to do the same thing for their monitors (see graphics).; TVs don’t have a specified frame rate. That’s a trait of the signal, not of the display. Instead, they have a refresh rate in hertz (Hz), indicating the number of times per second they re-write the image to the screen, whether it has changed or not. If the hardware’s refresh rate isn’t an integer multiple of the signal’s frame rate, it introduces irregularities into the timing of the frame changes, degrading display quality. That’s why manufacturers are moving beyond the old 60 Hz refresh rate: it’s fine for 30 fps and 60 fps HDTV signals, but not for, say, 24 fps Blu-ray. Higher rates such as 120 Hz smoothly accommodate both.; Display quality, in general, will be the least common denominator of signal quality and TV capability.
ATX: Advanced Technology Extended. Started as a 1995 PC motherboard standard from Intel, based on IBM’s 8.5 × 12 inch Baby AT, which in turn derived from the original AT. In 1999-2000 Intel released it as ATX12V version 1.0, adding specifications for a mechanically & electrically compatible PC chassis and power supply unit (PSU). It’s now an industry standard. Intel released ATX12V version 2.3 in March 2007.; The ATX12V motherboard, still usually called just “ATX”, can’t exceed 305 × 244 mm (12 × 9.6 inches), has a 20-pin power header (24-pin since v2.0), and can host five or more expansion slots. Along with its micro-ATX or µ-ATX variant, which is limited to 244 × 244 mm and four expansion slots, it’s the most common motherboard type as of 2009. The mini-ATX and Extended ATX are uncommon.; The ATX12V chassis is also normally called just “ATX”, and is also the most common type as of 2009. There are tower models of varying size – mini-tower, mid-tower, and full tower – in addition to the desktop types, which lay flat. Most models can accommodate both ATX and micro-ATX motherboards. The smaller micro-ATX cases, as the name implies, won’t take a full ATX board.; The ATX12V PSU provides +3.3, +5, and ±12 VDC power to the motherboard via a 20-pin (original) or 24-pin (since v2.0) connector. This includes 5VSB (5-volt standby) power to support quick wakeup. Many PSUs have a two-part 20+4 connector to accommodate either 20- or 24-pin motherboard headers. The PSU also has Molex connectors for disk drives and some peripherals. The +3.3 and +5 V levels support motherboard electronics, while +12 V runs fans and disk-drive motors. The -12 V level is little used. Standards prior to ATX12V v1.3 included a -5 V level for the now-defunct ISA bus. An ATX12V PSU measures 86 mm high × 150 mm wide × 140 mm deep, and has either an 80 mm exhaust fan on the rear or a 120 mm intake fan on the top or bottom. (Being bigger, an intake fan rotates more slowly and hence runs quieter, but an exhaust fan provides slightly better cooling.)
ATX12VO: Same as 12VO.
AU: Astronomical Unit. The distance from the Earth to the Sun, standardized to serve as a unit of astronomical measurement. The original figure, dating to 1672, was 140,000,000 km. In the 20th century, it became a calculation based on the mass of the Sun and other measurements, and could vary slightly. In 2012, the International Astronomical Union redefined it to be 149,597,870,700 m.
AUI: Attachment Unit Interface. Defined by the IEEE 802.3 Ethernet standard as a DA-15 (see D-sub) connection between a switch, router, or other controller and a transceiver (medium attachment unit, or MAU). The transceiver, which can be external or internal, allows the controller to transmit to and receive from the network medium. An Ethernet transceiver also provides electrical isolation between stations, and detects collisions. See XAUI.
autoclave: A vessel that generates high internal temperature and pressure. Used to make some kinds of crystals.
autocorrelation: The complex product of a given binary sequence and a time-shifted version of itself.
AUTOVON: Automatic Voice Network. A US military system.
AUX: Auxiliary. On audio equipment, a 3.5mm TRS jack for a stereo (two-channel) sound signal.
AV: Anti-Virus. Software to detect, block, or remove malware.
AVI: Audio-Video Interleave. A Microsoft standard for multimedia files, using a .avi file extension.
Avogadro’s number: Named for Italian physicist Amedeo Avogadro (1776-1856). See mole.
AVX: Advanced Vector Extensions. (2008) A 256-bit extension to the x86 instruction set.
AWG: (1); American Wire Gauge, formerly called the Brown and Sharpe (B+S) gauge. A standard for wire sizes dating to 1857. The smallest is AWG 50, which is 1 mil (.001 inch) in diameter. For every doubling of wire diameter, the AWG number goes down by six: .002" = AWG 44, .004" = AWG 38, and so on, down to .256" = AWG 2. Put another way, each increase in diameter results from a 26% increase in the wire cross-section. Solid copper AWG 14 can safely conduct up to 20 amps.; (2); Arbitrary Waveform Generator. A laboratory instrument that generates electrical signals with characteristics chosen by the operator: frequency, amplitude, duration, pulse shape, DC bias. Unlike other signal generators including the AFB, it can add modulation and other time-dependent changes, but might not achieve the same frequency precision.
AWGN: Additive White Gaussian Noise. Noise that’s completely uniform over a given frequency band. Thermal (Johnson) noise is AWGN up to about 1 THz (1,000,000,000,000 Hz), after which it falls off. See PN.
AWS: Amazon Web Services. Amazon.com’s cloud computing service, older and more popular than its rival Azure.
axion: See dark matter.
Azure: Microsoft’s cloud computing service and interface. Compare AWS.

Pin	Function	Pin	Function
1	microphone input	2	GND (analog)
3	microphone bias power	4	GND*
5	headphone output, right channel	6	headphone jack detect, right channel*
7	+5V power*	8	(no pin)
9	headphone output, left channel	10	headphone jack detect, left channel*

Alt- Code	Character
0009	tab
0150	–
0151	—
0153	™
0160	non-breaking space
0169	©
0174	®
0176	°
0177	±
0181	µ
0188	¼
0189	½
0190	¾
0215	×
0223	ß
0233	é
0241	ñ
0247	÷
0252	ü
8208	non-breaking hyphen
8486	Ω
8531	⅓
8532	⅔
8800	≠
8804	≤
8805	≥