Named for Swiss mathematician Leonhard Euler (1707-1783), this transcendental number is the base of the natural logarithm loge[x], which is written as ln[x]. The natural logarithm is so called because, at any value x1, a line tangent to the curve of ln[x] has a slope of 1/x1. The value of e can be calculated as the limit of (1 + 1/n)n or as the sum of the infinite series 1/n!, with n beginning at 0 and increasing without bound. (See factorial.) These terms both converge to the irrational number 2.718281828459… The expression “e to the power of n” is often written as exp(n) rather than en.
A set of European and South American digital carrier standards based on the AT&T DS-#. An E-1 circuit carries 2.048 Mb/s, enough for 30 DS-0 PCM 64 kb/s voice grade channels (1-15 and 17-31) and 2 64 kb/s control channels (0 and 16), and uses lines with 120Ω or 75Ω characteristic impedance. An E-2 is 8.448 Mb/s (four E-1 carriers, plus overhead), an E-3 is 34.368 Mb/s (four E-2 carriers, plus overhead), an E-4 is 139.264 Mb/s (four E-3 carriers, plus overhead), and an E-5 is 565.148 Mb/s (four E-4 carriers, plus overhead). Compare DS-#, J-#.
End-To-End Encryption. Encryption that’s applied at the source of a message and decrypted at the destination, so that the message is secure at all points in transit and even the ISPs can’t read it.
A mother. The father of geography was the North African scientist Eratosthenes, who, around 240 BCE, calculated Earth’s polar circumference as 252,000 stadia. Depending on the conversion, this is either 46,620 km (too large but not a bad try) or 39,690 km (amazingly close). Earth has a mean radius of about 6371 km (equatorial radius 6378.137 km, polar radius 6356.8 km), and a mass of about 5.976 × 1024 kg. By definition, one minute of latitude is equal to 1 international nautical mile (1.852 km), which implies a polar circumference of 1.852 × 360 × 60 = 40,003.2 km. The measured equatorial circumference is 40,075 km, so one minute of longitude is slightly more than 1.852 km at the equator. See g.
Discounting its waters and atmosphere, Earth consists of a molten core, a solid mantle some 3000 km thick, and a crust varying in thickness from 7 km under the ocean floor to 35 km beneath the continents. The crust is primarily igneous rock, which is formed by cooling magma, with the type determined by pressure and rate of cooling as well as by mineral content: rapid cooling at Earth’s surface yields extrusive igneous rock (e.g. basalt, rhyolite, obsidian, pumice), while slow cooling under pressure creates intrusive igneous or plutonic rock (e.g. granite, diorite, gabbro), with its characteristic large mineral crystals and coarse texture. For this reason, the continents are mostly granite, while the ocean floor, which is much younger than the continents, is primarily the denser basalt.
There are two other classes of rock besides igneous. Sedimentary rock is compressed deposits of eroded minerals (sand, silt, clay) and biological remnants (peat from plant matter, calcium carbonate from shells and skeletons of tiny marine animals). Metamorphic rock arises from geological pressure and heat acting on existing rock. With time and compression, sand, clay/silt, peat, and calcium carbonate become, respectively, the sedimentary rocks sandstone, shale, lignite, and limestone. More pressure and heat turn them into the metamorphic rocks quartzite, slate/mica/schist/gneiss, coal/anthracite/jet, and marble. Increasing heat and/or decreasing pressure reduces them to magma, which can then form new igneous rock.
Geological time uses, from largest to smallest, five time scales: eon, era, period, epoch, and age/stage. Just three eras span the last 542 million years: the Paleozoic (consisting of, in order, the Cambrian, Ordovician, Silurian, Devonian, Mississippian Carboniferous, Pennsylvanian Carboniferous, and Permian periods), the Mesozoic (Triassic, Jurassic, and Cretaceous periods), and the Cenozoic (Paleogene and Neogene periods). The Paleogene period consists of the Paleocene, Eocene, and Oligocene epochs, while the Neogene period is the Miocene, Pliocene, Pleistocene, and Holocene epochs. The Holocene epoch began with the end of the last ice age, about 12,000 years ago. Many geologists believe a new epoch, the Anthropocene, has recently begun, characterized by rising erosion, temperatures, and marine acidity, and lower soil fertility.
An archaeological rather than geological or zoological view divides the time since the emergence of hominids into the Stone Age, Bronze Age, and Iron Age, which are necessarily inexact terms, since different cultures entered them at different times, or not at all. The Stone Age further divides into three or four periods. The Paleolithic Period began about 2.6 million years ago with the appearance in Africa of what were once considered the earliest stone tools, and continued through the end of the last geological Ice Age to 12,000-10,000 BCE. (Much older stone tools have since been found, but archaeologists are in no hurry to redefine the Paleolithic.) The Mesolithic and Neolithic Periods, identified with agriculture, villages, trade, and increasingly sophisticated stone tools, carried through about 4500 BCE in the Fertile Crescent, followed by a Chalcolithic Period (to about 3300 BCE) marked by the use of copper, the rise of kingdoms and city-states, and the earliest writing. The Bronze Age ran from the end of the late Stone Age to about 1200 BCE in Europe and the Middle East. The Iron Age in Europe spanned approximately 1200 BCE to 400 CE.
After that, we’re down to calendars. Currently, a single revolution of the Earth around the Sun (one sidereal year) takes 365 days, 5 hours, 49 minutes, 9.55 seconds, not 365.25 days as assumed by the Julian calendar. That’s why the Gregorian calendar was created; it compensates for this slight difference by making years that are evenly divisible by 100 but NOT by 400 (e.g. 1400, 1500, 1700, 1800, 1900, 2100) non-leap years. For the four states that adopted the new calendar after 4 October 1582 as decreed by Pope Gregory XIII, the date rolled over to 15 October, leaving 10 calendar days that never existed. Other nations adopted the calendar later, requiring a larger adjustment. In the United Kingdom and its North American colonies, for example, it was the 11 days of 3 September to 13 September 1752 that never happened. Russia hung on until 31 January 1918, and skipped the first 13 days of February 1918.
A benign feature hidden in software, firmware, or even a CD or DVD, and activated by some unlikely trick. Two famous examples: the flight simulator in MS Excel 97 (open a new worksheet, press F5, type X97:L97 in the dialog box’s Reference field, press Tab to highlight cell M97, and hold down the Ctrl and Shift keys while clicking the menu bar’s chart wizard icon), and the arcade games (Centipede, Tetris) on HP oscilloscopes.
THE figure of merit for a digital signal: the ratio of energy per transmitted bit Eb to noise power spectral density N0 (joules to watts/Hz), measured at receiver input. It’s equivalent to per-bit normalized SNR; Eb/N0 = [C/N0] [1/R], where C (or Pr) is average signal power at the receiver and R is the information bit rate.
Eb/N0 is the best predictor of link performance – that is, BER – for a given modulation type. The receiver detector actually uses Es/N0 (symbol energy to noise), but that changes with code and modulation. For FEC-coded data, there’s also Ec/N0 (energy per coded bit to noise PSD), which is equal to (k/n) [Ec/N0] for a (k/n)-rate code. This means that each coded bit actually has a higher probability Pc of detection error than the uncoded bit-error probability Pb, but the coding gain should more than compensate.
Embedded Board Expandable. A form-factor standard for single-board computers. It defines size (5.75" × 8", or 146mm × 203mm), mounting hole pattern, power connector, and the use and placement of a PC/104-Plus bus connector.
Echo Cancellation. Essential for full duplex systems where both ends share the same bandwidth. It prevents each end from receiving its own reflected transmission. Widely used in VGC modems; see V.##.
Error Checking and Correction. A term for computer main memory (DRAM) that employs FEC techniques to reduce the risk of errors. This makes it more expensive and slightly slower, so it’s common only on systems with especially low tolerance for error – especially supercomputers, with their huge banks of memory that are susceptible to bit flips triggered by cosmic rays.
External Cavity Diode Laser. A semiconductor laser that uses one or two external mirrors to reflect laser light back into the cavity. A wavelength-selective component, such as a prism or diffraction grating, emits the desired wavelength. It can produce high output power, and a very tight spectrum over a broad range.
Earth-Centered Earth-Fixed. A 3D Cartesian coordinate system with its origin at the center of the Earth. The X-axis passes through the Prime Meridian at the Equator (a spot in the Gulf of Guinea), the Y-axis passes through the Equator 90° east of the X-axis point (in the Indian Ocean west of Indonesia), and the Z-axis is aligned with Earth’s axis of rotation. This is the coordinate system used by GPS.
Electro-Cardiograph (or -Cardiogram). A device that measures the electrical activity of the heart muscle using sensors affixed to the chest. Sometimes “EKG” to agree with the traditional Greek-derived spelling.
The path across the sky followed by an orbiting body. Earth is in the same orbital plane as all of the other planets except Pluto, so they (and the Sun and Moon) are all on roughly the same ecliptic.
Electronic (originally Engine) Control Unit. Also called an ECM (M for module). Any of the scores of microprocessors and microcontrollers that monitor and control the engine and other systems of every modern automotive vehicle, and display problem alerts on the dashboard. They’re accessible by way of an OBD-II output connector, which they share with the EDR. They might also be accessible via the wireless connection capabilities that almost all new vehicles have.
While the fine-tuned control they provide makes vehicles safer, more efficient, and less polluting, the many ECUs in a modern vehicle also make it expensive to design and repair, and vastly more difficult to fully test. Hacks of vehicle ECUs via wireless have been demonstrated.
Electronic Design Automation. Using software to design circuits, including virtual circuits for a PLD, FPGA, or simulation. The EDA sequence can be divided into design entry (schematic capture, HDL & synthesis, functional simulation), implementation (map, place, & route), and verification (timing analysis & in-circuit testing). Most EDA tools are made for digital design; analog is harder. See RTL, GDS.
Erbium-Doped Fiber Amplifier. A very important late 1980s development: a linear optical amplifier consisting of several meters of optical fiber with erbium doping. (There are other types of doped fiber amplifiers (DFAs), but EDFAs are the most common.) A diode IR laser spliced into the fiber injects light at 980 nm or 1480 nm wavelength, pumping up the erbium ions, which then lase when stimulated by signals in a roughly 3 THz bandwidth around 1550 nm, providing 20-40 dB or more of amplification. EDFA devices with wider amplification are being developed in an attempt to push more data through fiber. The EDFA tends to exhibit gain tilt, amplifying longer wavelengths more than shorter ones. Gain-flattening methods are applied to compensate. See WDM.
Enhanced Data rates for GSM (or Global) Evolution. Also called UWC-136. A modulation technology that enables existing GSM to operate at higher speeds. It was introduced as an intermediate stage between 2G and 3G mobile networks, but some implementations are called 3G.
EDGE employs 3π/8-8PSK modulation, tripling the GSM data rate but also increasing vulnerability to noise, and retains GSM’s GMSK modulation as a fallback. It provides a theoretical maximum rate of 474 kb/s, with average rates around 100 kb/s uplink and downlink. A version called EDGE Compact was proposed for the US to enhance IS-136, but flopped. A later upgrade, Evolved EDGE (EDGE II), uses 16-QAM and 32-QAM with slightly faster signaling to increase theoretical speeds to around 1 Mb/s.
In contrast to cloud computing, this refers to the use of computing resources that are located elsewhere but still physically near to users, minimizing latency.
Electronic Design Interchange Format. An ASCII text file standard for describing circuits, used in PLD design. It usually uses the .edf file extension. See netlist.
Event Data Recorder. The much simpler automotive equivalent of the aircraft FDR, the so-called black box. It constantly monitors vehicle speed, braking, steering, seat-belt use, airbag status, and other information. In the event of airbag deployment, it records at least 15 seconds’ worth of this data to non-volatile memory for later retrieval.
The EDR is readable via an OBD-II connector, which also connects to the vehicle ECU. EDR capability is sometimes part of the ECU module, sometimes physically separate.
Although EDRs are not yet (2011) mandatory, most cars in the US have one. Models that actually transmit the information they gather, including vehicle location, are common in rental cars and encouraged by some insurance companies. Some newer retail cars have models with this capability, which might or might not be active.
Electrically Erasable Programmable Read-Only Memory. Largely obsolete. It uses the same non-volatile memory (NVM) technology as the EPROM, but each bit of storage has an extra transistor that allows it to be individually and electronically reset. This dispenses with the need for UV light erasure, and the costly ceramic case with quartz window.
EEPROM is bulkier and slower than EPROM, but can be programmed without removing it from the system. Unlike its NAND-gate-based descendant, flash memory, it retains the ability to write individual bits. Flash is much more common due to its compactness and lower cost.
The really obsolete serial EEPROM was a much smaller variant (typically an 8-pin package), with data I/O one bit at a time, mandating access via a communications protocol. It was used when cost mattered more than speed.
European Geostationary Navigation Overlay Service. A European Space Agency program similar to the American WAAS. Its ranging and integrity monitoring stations (RIMS) on the ground generate regional error correction for satellite navigation signals including the USA’s GPS, Russia’s GLONASS, and Europe’s Galileo, and send this data to satellite uplinks. Three dedicated GEO satellites transmit it for EGNOS-capable receivers. It improves GPS accuracy to 1-2 meters.
Effective Isotropic Radiated Power. The power of a transmitter signal in a specific direction, assuming an isotropic radiator. This assumption is needed because a stationary receiver has no way to know what the transmitter’s actual propagation model is. EIRP is calculated as transmitter power Pt plus actual antenna gain Gt. Example: In a particular direction, an antenna being fed a 2W signal achieves 24 dB gain, so EIRP = [10 log(2W)] + 24 dB = 3.01 dBW + 24 dB = 27.01 dBW. A receiver in the chosen direction will see a signal strength equal to [EIRP – free space loss] dB.
Extended Industry Standard Architecture. A backward-compatible 16 MHz, 32-bit version of ISA, marginalized by PCI and now functionally dead. It used slots with 97 connectors, one less than ISA. See bus.
Electroluminescent. Having a phosphorescent coating that glows in response to electric current or electric fields. EL materials use less power than neon or fluorescent lights, and are becoming common in signs.
Executable and Linkable Format. (Used to be Extensible Linking Format.) Since 1999, the Unix and Linux standard format for binary code files, i.e., program executables, object files, and libraries. Flexible and extensible, it has largely replaced the more complicated COFF in non-Unix operating systems as well, especially embedded systems. It includes source code information to support debugging. If this is not needed, it can be stripped out during the build process to produce a smaller binary.
An ELF file begins with a header, which is either 52 or 64 bytes long depending on whether the file is 32- or 64-bit. These two options, selected by the value of the fifth header byte, are commonly denoted ELF-32 and ELF-64. The files can have the .elf filename extension, but often have a different extension to declare their content, such as .bin for binary, .o for object, or .so for shared object (i.e., library).
Electromagnetic. An electric charge differential, or voltage, creates an electric field E, which acts on other electric charges. A moving electric charge, or current, creates a magnetic field H, which acts on magnetic poles. (See Maxwell’s equations.) These effects radiate into space and certain other media as EM waves, which are categorized by frequency range as RF waves, microwaves, infra-red, visible light, ultra-violet, X-rays, gamma rays, cosmic rays, etc.
A transverse electromagnetic (TEM) wave, the case that occurs in free space and is informally called a plane wave, has its electric and magnetic fields oscillating in phase, and entirely perpendicular to one another and to the direction of propagation. It travels at speed c with direction of propagation given by the cross product [E × H]. (At the level of individual photons, quantum uncertainty leaves field phase undefined.)
TEM wave
In higher-order modes – optical fibers, hollow single-conductor waveguides, dielectric rods – E and H are not necessarily in phase, and there are components of one or both fields in the direction of propagation. These cases are classed as TE, TM, and hybrid modes.
The above descriptions of EM wave behavior are for the wave’s far field, which is the standard case. Within the near field, which extends one wavelength distance from the source, things get more complicated: E and H exhibit varying relative phase shift, and detection or reception of the signal feeds back into the transmitter. The near field is further divided into the reactive and radiative (Fresnel region) near field.
Electro-Mechanical Actuator. An electrically powered device for creating straight-line mechanical force. Most types use an electric motor to turn a threaded shaft, which applies a pushing or pulling force as it moves in its housing.
Electronic Mail. The TCP/IP protocols for handling Internet e-mail are SMTP for sending, and POP or the newer IMAP for receiving. Web-based e-mail, aka Webmail, communicates with mail servers through HTTP, which eliminates the need for a user e-mail client app but leaves the user without local storage.
For security and privacy, it’s a good idea to set the e-mail client or Webmail interface to default to plaintext-only display. Users can change the display for individual messages that they deem trustworthy. An e-mail with formatting, besides potentially containing embedded malware or links to dangerous sites, can contain tiny, effectively invisible images that the client tries to retrieve when the user opens the message, which notifies the originator(s) that the message has been opened. This trick, called pixel tracking, is used by spammers and also by commercial services that compile and sell information about users.
A device controlled by a microcontroller or microprocessor, or sometimes just the controller/processor itself. Coffee-makers, dishwashers, digital clocks, vending machines, gas pumps, cars, and many other things now depend on built-in microcontrollers to make their operating decisions. An Internet appliance means an embedded device of this sort with the capability to connect to the Internet. The Internet of Things (IoT) is a trendy term for all these online embedded devices, usually communicating with other machines rather than with people. (Routers, modems, hubs, and switches are run by microcontrollers too, but aren’t called Internet appliances because networking is their purpose, not just a feature.)
An embedded device is typically programmed via a JTAG bus or some other interface from a standard computer. The embedded device’s microcontroller loads its program at power-up (from ROM, NVRAM, flash, whatever). Microcontrollers aren’t built for multi-tasking, so if there’s an operating system (OS), it’s compiled with the program into a single executable. Because the typical embedded device has to respond immediately to events or user input, almost every embedded OS is a real-time OS (RTOS) that guarantees responses within some very short interval. There are literally hundreds of RTOSes in use, most of them subsets of GNU/Linux.
An embedded system is simply a collection of embedded devices working together. Such a description covers a wide range of systems, however, from the simple to the very complex.
Equilibrium Mode (Modal) Distribution. A condition of multi-mode fiber in which the relative power distribution among the modes is constant. This occurs after some propagation distance called the equilibrium length, which can be more than a kilometer.
Encrypted Media Extensions. An addition to HTML 5 to support closed, proprietary content decryption modules (CDMs) that control decryption of DRM-protected Web video. It was approved by the W3C in 2017 amid widespread criticism that it tramples on the doctrine of fair use.
Electromagnetic Interference. EM waves from an external source affecting a system’s operation. Also called EMH (electromagnetic hazard) or RFI. Intentional EMI (IEMI) refers to RF transmitters intended to degrade the performance of electronics, especially communications; see HPM.
Embedded Multi-die Interconnect Bridge. Intel memory technology to improve memory density and speed, using silicon bridges instead of the problematic TSVs used in similar technologies such as HBM.
Also called common collector. A logic circuit output from the emitter terminal of a transistor, with an internal pull-down resistor. It provides unity gain at best, but has very high input impedance and so doesn’t draw much power.
Electromagnetic Pulse. A powerful surge of EM waves that can fry unshielded electronics. Doing this at a distance takes a nuclear detonation or a high-power, narrow-beam transmitter. See HPM.
Extraordinary Magneto-Resistance. A huge change in resistance exhibited by certain non-magnetic hybrid materials in response to a magnetic field. This effect, discovered in the late 1990s, replaced GMR-based technology in hard disk drive read heads. Its non-magnetic read heads are not susceptible to magnetic noise, and so can be much smaller and more sensitive.
Europay, MasterCard, Visa. A worldwide standard created in 1995 for so-called chip-and-PIN cards: more secure credit or debit cards with embedded IC chips that support encryption, in addition to or in place of the strip of magnetic tape on the back. Some but not all versions require use of a 4- to 6-digit PIN. So far, EMV is an obstacle to point-of-sale abuses, but not to online credit-card fraud.
Encryption systems divide broadly into two types: symmetric (a single password to encrypt and decrypt) and asymmetric (different passwords for encryption and decryption). Symmetric encryption further divides into stream ciphers, which apply a pseudo-random encrypted sequence that operates on a single bit at a time, and block ciphers, which apply encryption to bits in blocks of some fixed size.
Asymmetric encryption requires longer keys and much more processing power, making it slower. It’s also called public key encryption, since it enables users to make one key public while keeping the other private. Security rests upon the extreme computational difficulty of performing the reverse calculations to find the secret key even when the algorithm and public key(s) are known. For example, it’s much easier to calculate the product of two factors than to calculate the two factors of a product, a difficulty gap that increases with the size of the numbers.
The user of a public-key system encrypts messages with the private key. Anyone can decrypt them with the public key, verifying that the messages are from the user. Conversely, if someone encrypts a message with the public key, only the private key can decrypt it, ensuring that only the user will receive it. This assumes that the public key is verified as genuine – a problem for which there is still no definitive solution.
Hybrid encryption refers to approaches that use both symmetric and asymmetric encryption. The objective is to provide the speed and robustness of the former with the convenience of the latter.
Some common commercial encryption standards:
A5 – A stream cipher used in GSM. A5/0 is unencrypted, A5/1 is the standard, A5/2 is a weakened version of A5/1 for export, and A5/3 is a newer algorithm beginning to replace A5/1, which is known to be vulnerable to sophisticated attack. Three Fibonacci LFSRs – 19, 22, and 23 bits long – are loaded with a 64-bit key. The middle bits of the LFSRs are examined at each clock cycle to determine which registers shift. The output bits are XORed to produce the code, which is XORed with the data stream to encrypt or decrypt.
AES – Advanced Encryption Standard. A NIST-selected block cipher algorithm approved in 2001 to replace the old DES. It permits 128-bit, 192-bit, and 256-bit encryption keys, and its Rijndael (pronounced “Rhine-doll”) encryption algorithm uses the key to shift rows and columns in blocks of numbers.
CDMF – Commercial Data Masking Facility. The same as DES, except that it uses a 40-bit key and expands it to 56 bits.
DES – Data Encryption Standard. A 1977 US government algorithm that uses a 56-bit private key in a series of steps to convert 64-bit input blocks into 64-bit output blocks. It uses the same steps and key to reverse the process. Triple DES is a 168-bit descendant.
D-H – Diffie-Hellman. Also Diffie-Hellman-Merkle. Named for inventors Whitfield Diffie and Martin Hellman, based on a concept by Ralph Merkle. A hybrid encryption algorithm from 1976, still used in 2013 for Web encryption (HTTPS) among other things. Each end uses two publicly known prime numbers plus its own secret, randomly selected value to generate its own public key. They then exchange public keys, and from them calculate the same unique private key, which they use to encrypt messages between them. The advance of computing power has made the use of D-H with 512-bit prime numbers insecure, and even 1024-bit D-H might be vulnerable to organizations with NSA-level resources.
ECC – Elliptic Curve Cryptography. A highly scalable, public-key approach based on the algebraic structure of elliptical curves over finite fields.
MD# – Message Digest. A family of cryptographic hash algorithms originated by Ronald Rivest. Members are the 128-bit MD2 (1989, for 8-bit processors), MD4 (1990), and MD5 (1991), all of which are now considered insecure, and variable-length MD6 (2008).
PGP – Pretty Good Privacy. A popular type of hybrid encryption originally created by Phil Zimmerman in 1991. It uses a randomly generated, one-time symmetric key (the session key) to encrypt a message, and then uses the intended recipient’s public key to encrypt the session key. The recipient uses their private key to decrypt the session key, and finally uses the session key to decrypt the message.
In contrast to the hierarchical public key infrastructure required for SSL and TLS, PGP relies on a web of trust to distribute public keys in peer-to-peer fashion: Each user of the system accumulates the public keys of other, trusted individuals, and can act as certifier for those who don’t know them.
RC4 – Ron’s Code #4, pronounced “arc four”. A stream cipher developed by Ronald Rivest of RSA fame, and used by the TLS and older SSL browser encryption protocols. It uses key sizes from 40 to 2,048 bits. It’s now considered vulnerable to sophisticated attack.
RSA – Rivest-Shamir-Adleman. A public-key encryption and authentication system developed for the Internet in 1977 and named for its creators. It’s still in widespread use. The private key is a pair of very large prime numbers. The public key is the product of these numbers. Its security is based on the extreme computational difficulty of factoring a public key to reveal the prime numbers that produced it. Public key size is variable, up to 4096 bits so far (2015). Current best practice is to have a key at least 2048 bits long, and to add randomized padding to the message as protection against hostile decryption analysis.
SHA-# – Secure Hash Algorithm. A series of NIST-approved cryptographic hash functions. SHA-0 (1993) is no longer used. The 160-bit SHA-1 (1995) is based on MD5, and, like it, is no longer secure. SHA-2 (2002) uses either of two functions, called SHA-256 and SHA-512 for the bit lengths of their data blocks. Bitcoin uses SHA-256. SHA-3 (2012) XORs each block of message data with a 5×5 array of 64-bit words called the system state, which changes after each XOR; it uses hash lengths of 224 to 512 bits.
From Gulliver’s Travels. In digital devices that store data as 8-bit bytes, which is almost all of them, data types with more than 256 possible values must be broken into multiple bytes. Big-endian (a.k.a. Motorola order) systems, including Motorola-based Apple computers and most mainframes, store the most significant byte first, giving it the lowest address in the sequence. Little-endian (a.k.a. Intel order) systems, such as PCs, store the least significant byte first. Bit order within a byte is a separate issue, and (WARNING) has nothing to do with the way a byte is read or written by people. Endianness is sometimes called byte sex.
A programmer can demonstrate endianness by reading just the first byte of a multi-byte integer. For example, in C, assuming 16-bit integers (int), one could create an integer, get the address of the integer and recast it as a pointer to character, and read the value pointed to, which is the first byte stored:
int num1 = 0x1234; char *ptr1 = (char *)&num1; if (*ptr1 == 0x12) // BIG-ENDIAN else if (*ptr1 == 0x34) // LITTLE-ENDIAN
A device that expends fuel to create mechanical motion. Compare generator, motor.
Under ideal conditions, each hydrocarbon fuel molecule combusts with one oxygen molecule, producing only CO2 and H2O. However, even at a 1:1 balance of fuel to air, a condition called lambda 1, combustion is not perfectly efficient. A lean mixture (lambda > 1), with more oxygen than fuel, burns with higher efficiency and slightly cooler, but produces less power. A rich mixture burns more reliably but less efficiently.
Common types of internal combustion engine (ICE):
Reciprocating internal-combustion engines are the most common. They include the four-stroke engines of most cars, and the two-stroke engines used not only for small machines such as lawnmowers and scooters but also for very large, slow-revving machines. Such an engine has one or more cylinders, each with its own reciprocating piston, in which combustion takes place. A two-stroke engine performs a full power cycle with just two strokes of the piston – one for compression, one in the opposite direction for combustion. It’s a relatively light, simple design, but emits a lot of exhaust gas. The power cycle of a four-stroke engine is two full revolutions of the crankshaft, hence four piston strokes for its four stages: intake (of fuel & air), compression, power (ignition), and exhaust, vulgarly but colorfully summarized as “suck, squeeze, bang, blow”. Both subtypes employ spark generators, commonly spark plugs, to provide ignition.
Diesel engines, named for German inventor Rudolph Diesel (1858-1913), ignite their fuel by heat of compression rather than a spark generator. They have better fuel efficiency than reciprocating four-stroke designs, but also emit more pollutants, because the much higher pressure required for ignition means a less homogenous fuel-air mix that burns less completely.
Rotary engines separate the four stages of intake, compression, ignition, and exhaust into four different physical spaces. The Wankel rotary engine, named for German engineer Felix Wankel (1902-1988), is the only commercially successful version. A rounded-triangle rotor turning in a roughly oval housing creates three volumes of gas that are constantly expanding and then contracting, alternately drawing in the fuel-air mix and expelling the exhaust through dedicated ports in the housing. It still uses spark plugs for ignition. It has a higher power-to-weight ratio than reciprocating-piston engines, but is less fuel-efficient and more polluting. (The rotating-cylinder-block engines of some very early aircraft were also called rotary engines, but were very different beasts.)
Radial engines are a variant reciprocating-piston design used for decades in propeller-driven aircraft. The cylinders are arranged in a ring converging on the propeller shaft, which they drive via a single rotating cam connected to their pistons. The engine is air-cooled, and so requires a distinctive open-front cowling. Some designs use multiple rings of cylinders to generate greater power, which makes cooling more difficult.
Gas turbine engines use a rotating compressor to take in and compress air, then inject fuel and burn the fuel-air mix. The high pressure from this combustion drives a turbine that boosts the rotation of the compressor. These engines produce more power relative to their weight than reciprocating-piston designs. Because they must operate at very high turbine speeds, with high internal temperatures and pressures, they’re expensive and need frequent maintenance. Subtypes include the turbojet, turboprop, and turbofan engines of modern fixed-wing aircraft, and the turboshaft engines of helicopters.
A set of reusable development tools for handling the computation, interface, and display tasks common to most computer games. Just as you can build different cars around the same mechanical engine, so with games and game engines. A 3D video rendering engine, for example, is normally built on a standard graphics API (Direct3D, OpenGL) to provide an abstracted hook for GPUs. Interface with other hardware is usually abstracted through libraries such as DirectX, SDL, and OpenAL.
Electronic Numerical Integrater And Computer. The first fully electronic general-purpose computer, built during WWII and publicly unveiled in 1946. It contained 17,468 vacuum tubes, weighed more than 27 metric tons, occupied about 200 square meters, and used 150 kW of power. It malfunctioned every day or two because of vacuum tube burnouts.
End Of Line. For indicating the end of a line in text files, Windows uses carriage return followed by line feed, which is byte values 0x0D 0x0A. Unix and the current macOS use just line feed, and the old Mac OS (versions through 1999) uses just carriage return. For fax signals, the end-of-line sequence is typically eleven 0s and a 1 (000000000001). See CR/LF.
Explicitly Parallel Instruction Computing. The approach Intel & HP used for the IA-64 instruction set & architecture, the basis for the unsuccessful Itanium processor series intended to replace the x86 family. It combines long or very long instruction words (VLIW), instruction-level parallelism, and branch predication.
Instruction-level parallelism (ILP) means that a compiler arranges the instructions at compile time so that they are pre-scheduled for parallel execution, and exposes this parallelism to the CPU in a template field of the code. This eliminates the need for the dynamic scheduling that is used in out-of-order execution. It also means that this type of CPU is heavily dependent on the compilers that produce its code.
Branch predication is similar to branch prediction, but the CPU processes alternative paths, and chooses between the results once it knows the correct path. This means wasting some cycles executing instructions that are never used, but it’s cheaper than the penalty associated with having to start over after a wrong prediction.
Ethernet Passive Optical Network. A service model for subscriber access to the Internet via Ethernet protocols. It uses a single fiber as the trunk line from the central office, and a passive (no power required) optical splitter to connect it with a number of fiber drops to individual subscribers. It’s now competing with DSL and cable modems. In 2009, 10 Gb/s EPON became IEEE standard 802.3av. See PON.
Erasable Programmable Read-Only Memory. An obsolete type of non-volatile memory that uses charge-storing transistors instead of the fuse-blowing connections of a PROM, allowing it to be programmed more than once. An EPROM chip has a ceramic case with a quartz window. Strong UV light beamed through the window dissipates the stored charges on the transistors, erasing the memory so that it can be reprogrammed in a device programmer. See also EEPROM.
EPROM
The EPROM can hold data for 10-20 years. It was used to store code and constant data to be accessed at runtime. Since the ceramic packaging was expensive, the same chip design was usually available with a plastic case and no window, making it functionally a PROM.
Entry-level Power Supply. A 1998 PSU standard from an industry working group, creating an extension of the ATX PSU to support servers with high power needs. Now elaborated as EPS12V, it defines three physical sizes for power supplies: 86 × 150 × 140 mm (the same as ATX), 86 × 150 × 180 mm, and 86 × 150 × 230 mm. Most full ATX cases fit all three, but it’s best to measure first for the two longer types. “EPS” is also used to mean the 24-pin ATX power headers and 8-pin CPU power headers on PC motherboards.
A system of coordinates for astronomy, using the convenient fiction of a celestial sphere borrowed from the long-defunct Greek model of a geocentric universe. Its equator is defined by Earth’s rotation, and its meridian is directly overhead at noon on the first day of spring – the vernal equinox. Locations on the celestial sphere are specified by declination (δ), which is degrees north or south of the celestial equator, and right ascension (α), which is the hour angle (HA, in hours, minutes, & seconds) east from the celestial meridian. The right ascension and declination of a celestial object therefore remain almost constant (the vernal equinox shifts about 50" west each year) as the Earth orbits the sun.
There are heliocentric, geocentric, and topocentric (observer’s actual location) systems of celestial coordinates, depending on the degree of error that can be tolerated. For example, planets cannot be accurately tracked using heliocentric coordinates.
A unit of network traffic density. One erlang is the density at which one traffic path would be continuously busy, e.g. one call-hour/hour or one call-minute/minute.
Enterprise Resource Planning. Also called Enterprise Application Suite (EAS). A business term for database software that organizes how an organization obtains and manages the resources – facilities, employees, equipment, supplies, etc. – it needs to function. The most widely used is SAP.
Electro-Static Discharge. A formal term for static electricity, which can accidentally influence, damage, or destroy electronics. The damage is often undetectable to the eye. Raising humidity reduces ESD, although if it gets too high it creates condensation, which has its own problems.
In the US, both the ESD Association and the EIA set industry standards for ESD testing and safety. JEDEC and the IEC set tolerance standards for integrated circuits.
The JEDEC human body model (HBM) for ESD allows for human contact to create voltages as high as 30 kV. JEDEC also has a machine model (MM) and charged device model (CDM) for ESD, although the latter is considered obsolete. As electronics feature size has decreased over time, circuits have become more sensitive to ESD, so the tolerances assumed by the models have had to be lowered.
The anti-static mats used during electronics assembly have a non-conductive upper surface and a conductive underside, albeit less conductive than, say, a wire. The operator wears a wrist strap connected to the mat’s underside, which is in turn connected to earth ground, commonly the grounding wire of an AC socket. This prevents both static buildup on the operator and voltage differential between operator and work surface. A powered ESD monitor is sometimes connected to the mat underside as well, to provide an audible alarm if the mat isn’t grounded.
Enhanced Small Device Interface. Maxtor’s obsolete 1980s disk drive interface. The ESDI controller connects to one or two drives using a single 34-pin control cable, and a separate 20-pin data cable for each drive.
Equivalent Series Inductance (L is for inductance in honor of Russian physicist Heinrich Lenz). The same idea as ESR, except, of course, the property being modeled as separate from the ideal component is inductance.
Electronic System Level. The conceptual or architectural level of IC design, one step more abstract than RTL. Description languages for EDA at this level exist, but RTL approaches are more popular (2004).
Electronic Serial Number. A 32-bit ID number (8-bit manufacturer code + 24 bits for the individual device) programmed into cell phones at manufacture starting in the early 1980s. A cellular base station verifies the ESN before connecting a device. There weren’t enough unique ESNs for the explosive growth of the mobile wireless market, hence the later IMEI and MEID. Some wireless systems still use pESN (pseudo-ESN), which sets manufacturer code to hexadecimal 0x80, and generates the other 24 bits from a hash of the device’s MEID.
Equivalent Series Resistance. No real-world electronic component is perfect. A capacitor, resonant crystal, or other device can be modeled, at some benchmark frequency, as an ideal component of its type in series with a fixed resistance. The value of this hypothetical ESR accounts for all of the losses generated within the component. Its effects increase with frequency.
High-quality capacitors have ESR << 1 Ω. This generally desirable property means circuit designs must take higher inrush currents into account.
From French étalon, a standard of measurement, deriving in turn from the Greek eidolon (ειδωλον). Also called a Fabry-Pérot interferometer (FPI). An optical cavity consisting of two parallel, partially reflective mirrors. Light waves must be in resonance with the cavity to overcome the self-interference it causes and pass through, so, as well as being an instrument for analyzing the spectrum of incident light, it can act as an optical band-pass filter. Strictly speaking, it’s an etalon if the length and therefore the resonant frequency is fixed, and an interferometer if it’s tunable, but not everyone observes this distinction.
CH3CH2OH, also called ethyl alcohol or grain alcohol. It’s produced from fermentation of sugars by yeast or, less often, bacteria, and purified by distillation. In the US, the proof of an alcoholic liquid is 2 times its percentage of ethanol by volume, so 200 proof means 100% pure ethanol. The EU uses Gay-Lussac proof, which is equal to the ethanol percentage, while the old British proof was 1.75 times the percentage. Industrial or denatured ethanol contains small amounts of toxic or vile-tasting additives to make it unsuited for drinking. Compare methanol.
Originally a 1978 bus LAN standard, with data frames containing a 22-byte header, 46-1500 bytes of data, and 4-byte Frame Check Sequence (FCS). Now refers to the IEEE 802.3 family of standards starting with 10BaseT and using Cat3 or better cable.
The Alexandrian mathematician Euclid (~300 BCE) proposed five fundamental axioms of geometry, stated in two dimensions (a plane) and extrapolated to 3D. The first four axioms make up what’s called basic or absolute geometry, stated as:
1) For any two points, there is just one line segment that connects them.
2) Any line segment can be extended indefinitely and remain a straight line.
3) A circle is uniquely described by its center and radius.
4) All right angles are equal to one another.
The fifth axiom states that, for any given line and a point not on the line, there exists just one other line that passes through the point and never intersects the first line. This is also called the parallel postulate, and is equivalent to the Pythagorean theorem. It’s the defining characteristic of Euclidean (or parabolic) geometry.
Any definition of lines and points that obeys the first four axioms but not the fifth is a non-Euclidean geometry. For example, marine navigation employs elliptical geometry, operating on the surface of a sphere in 2D. The line segment between two points is the shortest distance between them, and must fall along a great circle. However, in elliptical geometry, any two lines must intersect. Hyperbolic geometry (on the surface of a pseudosphere) is essential to General Relativity. The orbit of Mercury around the Sun can be predicted with slightly more accuracy using hyperbolic as opposed to Euclidean geometry.
Having the lowest melting point that can be achieved in a mixture or alloy of a given set of ingredients. Solder is a eutectic alloy. Lead-free solders have higher eutectic temperature, and a troublesome tendency to sprout fine metal whiskers.
Extreme Ultra-Violet. The long-awaited future of IC lithography, for creating extremely small circuit features using high-energy UV radiation. It beat rival technologies such as ion-projection, direct-write e-beam, and electron-projection lithography to finally reach production in 2018. It came too late for 10 nm and 8 nm process, deploying initially at 7 nm.
One difficulty is that EUV wavelengths tend to be absorbed by most materials, including lenses and even air. Another is creating a strong EUV source, and the photoresist material to go with it. The leading EUV source technologies are called laser-produced plasma (LPP) and laser-assisted discharge plasma (LDP); the latter is also called discharge-produced plasma (DPP).
Electric Vehicle. Unlike the more common hybrid electric, which powers its drivetrain with an electric motor but backs it up with a fuel-burning internal combustion engine, an EV has only an electric motor. The most common type is the battery EV (BEV), running entirely on stored charge from its internal battery, with a standard SAE J1772 receptacle for plugging it in to recharge. The 5-contact SAE J1772 is also called the IEC Type 2 connector. The Combined Charging System (CCS) connector is a variant with two additional contacts for fast DC charging. The US and European versions of CCS have somehow managed to be incompatible. In 2023, the SAE accepted Tesla’s North American Charging Standard (NACS) connector as a standard.
J1772 receptacleCCS receptacleNACS receptacle
The US has two recharging standards for EVs. Level 1, which uses an adapter cable to connect a residential 120 VAC outlet directly to the vehicle’s SAE J1772, provides enough charge in one hour to drive just 3 to 8 kilometers. Level 2 also draws on 120 VAC power, but uses an EVSE (electric vehicle service equipment), an external charging system with SAE J1772 plug. Level 2 converts residential power to 240 VAC and up to 30 amps, providing 16 to 40 km of driving range per hour of charging. EV owners for whom Level 1 isn’t acceptable can buy EVSEs to install at their homes.
There is as yet no Level 3 charging standard, but for EVs with a CCS or NACS receptacle, there are proprietary DC fast-charge systems that can charge batteries to about 80% in 30 minutes. These are found at commercial charging stations, as the equipment is too expensive for most private owners.
As of 2017, most EVs get less than 150 km of driving out of a fully charged battery. Swapping out the vehicle battery for a fully charged one is potentially much faster than any charging cycle, but manufacturers have not embraced the standardization that this would require.
Error Vector Magnitude. In digital communications, the error vector (EV) is the differential between the ideal and the received modulation symbol, measured for one symbol or averaged over many. EVM, the absolute value of the EV, is a common figure of merit for modulation quality.
Evaluation Module. A circuit board that hosts a modular component, for use by engineers and programmers developing software that runs on the component or other systems that interact with it.
Extended File Allocation Table. A 2006 Microsoft file system used for thumb drives and SD cards. Windows and later versions of Mac OS X support it. It permits vastly larger files and storage capacities, longer file names, and greater numbers of files than the FAT32 system on which it’s based.
In MS-DOS, PC memory up to 32 MB beyond the basic 640 kB RAM address space, assigned according to the Expanded Memory Specification (EMS) standard. It was application-specific, hence rarely used even before Windows 95, which killed it.
PC main memory beyond the first 1024 kB (1 MB), for MS-DOS on 80286 or later CPUs. (8086/8088 systems can address only 1 MB of RAM, hence can’t use it.) It’s accessible only when the processor is in protected mode. It became irrelevant for later versions of Windows that don’t run on top of DOS. See XMS.
A common way of viewing digital signal quality in real time. An oscilloscope synchronized to the symbol rate displays the overlapping symbol traces. The eyes are spaces between the amplitude levels of the signal. Pure FSK and PSK have just two levels (see image) because they don’t modulate the amplitude of the sinusoidal carrier wave, but 16-QAM has four amplitude levels, hence three eyes. If signal quality declines, the eyes narrow or close. See constellation.
