EE Terminology

S2S: Server To Server.
SAE: Society of Automotive (originally Automobile) Engineers. A standards-setting and professional body for automotive and aerospace engineering worldwide.; Their most widely recognized standard is for motor oil viscosity. Motor oil has a SAE base rating from 20 to 60, proportional to its viscosity at 100° C. All-season oil, for engines that operate in temperatures below freezing, has lower viscosity at colder temperatures. The SAE grade of all-season oil therefore has a ##W- prefix (W = winter), with the low-temperature viscosity number ranging from 0 to 25. This is because high viscosity better protects an engine from wear, but in cold weather a less viscous oil reaches all parts of the engine faster, shortening the period of harmful dry running that occurs at startup.; The SAE also created the SAE J1772 connector and associated recharging standards for electric vehicles.
SAGE: Semi Automatic Ground Environment. An innovative 1950s military project: a computerized electronic defense network supporting the DEW Line. Not a network in the modern sense, since the computers didn’t communicate directly.
Salutation: A late 1990s standard for network service discovery protocols (SDP), to simplify network assembly.
SAM: Spin Angular Momentum. See OAM.
SAN: (1); Storage Area Network. A LAN optimized for RAID storage and retrieval. It reduces overhead to facilitate the transfer of large blocks of data rather than many shorter message packets. Compare NAS.; (2); Satellite Access Node. What some folks call a ground station or earth station.
sandbox: Security quarantine for a program, isolating it from the rest of the system that runs it and tightly controlling what it can do, often by hosting it in a virtual machine.
SAP: Systems, Applications, and Products in Data Processing. A very widely used ERP software package sold by the German company SAP AG.
SAR: (1); Synthetic Aperture Radar. A moving radar platform can take a series of range and azimuth (cross-track) readings as it moves. DSP techniques can process the azimuth data as if it came from a single narrow-beam antenna with a physical length equal to the distance of travel. This distance is the synthetic aperture. The resulting images have much better azimuth resolution than those of a conventional radar antenna.; (2); Successive Approximation Register. A popular structure for analog-to-digital converters (ADCs). It uses a divide-and-search approach to convert the analog value to digital form. This means that the ADC sample rate is usually some fraction of the system clock rate.; (3); Specific Absorption Rate. A FCC safety guideline for exposure to RF radiation.
SAS: Serial Attached SCSI. A newer, serial version of the old parallel SCSI, with speeds of 3, 6, or 12 Gb/s, and cable lengths up to 10 m. It uses several different physical connectors for ports & cables, including one compatible with SATA, because SAS controllers can accept SATA drives.
SATA: Serial Advanced Technology Attachment. A high-speed serial bus standard supplanting the old parallel ATA (PATA) used for PC disk drives, with which it’s software-compatible. It uses a smaller cable with a 7-pin connector to support two balanced-line pairs, one (A) to transmit and one (B) to receive. SATA drives use a 15-pin power input, but some also have the Molex 4-pin power port long used for ATA drives in case the supporting system doesn’t have SATA power connections.; The original 2003 SATA runs at a maximum of 1.5 Gb/s, although its use of [10, 8] FEC coding limits throughput to 1.2 Gb/s (150 MB/s). The 2005 SATA 3Gb/s and 2009 SATA 6Gb/s are commonly called SATA II or 2.0 and SATA 3.0, respectively, over the objections of SATA-IO, the standards organization. Their maximum throughputs are 2.4 Gb/s (300 MB/s) and 4.8 Gb/s (600 MB/s). Devices that support these standards often settle on speeds below the limit.; Rather than trying to double SATA speed yet again in a new specification, the SATA-IO turned to SATA Express (part of the SATA 3.2 specification), which provides both SATA and PCIe connections. This led to the M.2 bus, intended to replace both.; External SATA (eSATA) just means a connection to an external device.; Mini-SATA (mSATA) is a 2009 variation that implements 1.5 or 3.0 Gb/s SATA on a circuit card with a 52-contact tab identical to that of a mini-PCIe board. It’s intended as a data storage platform in systems where space is tight. It mates with a mini-PCIe slot that supports SATA.
satellite band: See microwave band.
satnav: Satellite Navigation. A system that uses signals from artificial satellites to determine the receiver’s position on the globe, or a portion of the globe. GPS was the first. Satnav systems in operation as of 2022:; BeiDou – The Chinese name for the Big Dipper. The BeiDou Navigation Satellite System (BDS), sometimes called COMPASS, is a Chinese-built geopositioning system first launched in 2000. The original four satellites are now called BeiDou-1. They cover China and neighboring regions. BeiDou-2 is intended to provide global coverage by 2020. Unlike other systems, BeiDou uses geostationary as well as MEO satellites.; Galileo – Named for Italian scientist Galileo Galilei (1564-1642). The European Union’s neutral, civilian alternative to the other satnav systems, which are each controlled by the military of a single nation and could be made unavailable in a crisis. Operational as of 2016, it has two European control centers, plus 10 uplink stations around the world. The plan is for a constellation of 30 satellites (24 operational and 6 spare) in three circular, 23,616 km MEO planes inclined 56° relative to the equator. As of 2019, it has 22 in operation. There are three different signals to provide different categories of service: E1 (1575.42 MHz), E5 (1191.795 MHz), and E6 (1278.75 MHz).; GLONASS – Russian ГЛОНАСС, from Глобальная навигационная спутниковая система (GLObal’naya Navigatsionnay Sputnikovaya Sistema), or Global Navigation Satellite System in English. A Russian (formerly Soviet) system. In theory, it has 21 MEO satellites (plus three spares) in three 19,100 km circular orbits with a period of 11 hours 15 minutes 44 seconds, inclined 64.8° to the equator, giving it good coverage at higher latitudes. One ground station is in Brazil, the rest in Russia.; Like GPS, GLONASS has unencrypted L1 signals and more precise, encrypted L2 signals. Both use BPSK modulation with DSSS. The L1 signal has 15 562.5 kHz FDMA bands ranged around a 1602 MHz RHCP carrier, and achieves resolution of roughly 5-10 meters. The L2 signal has 15 437.5 kHz bands around a 1246 MHz carrier.; GPS – Global Positioning System. The US military NAVSTAR (Navigation Signal Timing and Ranging) system of 24 MEO satellites (actually 29 as of 2007) controlled from Colorado Springs that continuously transmit low-power time-and-position signals. The first NAVSTAR satellites were just for testing, and were dubbed Block I. The newer ones are Blocks II, IIa, and IIR. They’re in six circular 12-hour orbits, with an altitude of 20,200 km and a 55° inclination to the equator, and use 10.23 MHz atomic clocks. The orbits are meant to ensure that at least six satellites are in view at any given time and location.; When GPS time began at 00:00:00 UTC on 6 January 1980, it agreed precisely with UTC, and GPS compensates for the relativistic effects that cause a clock in orbit to run slightly faster than one on Earth. However, the very gradual slowing of Earth’s rotation requires that UTC be delayed by a second every year or three to keep it in sync. Because so many users are depending on it for precise navigation at any given moment, GPS time can’t be safely changed in this way. Therefore, as of 1 January 2017, GPS time is 18 seconds ahead of UTC. There are proposals to sever the link between UTC and georotation, but that hasn’t happened.; In a separate time-related issue, the GPS week number is just 10 bits, so it rolls over from 1023 to 0 approximately every 19 years 7½ months. It did this at 23:59:47 UTC on 21 August 1999 and at 23:59:42 UTC on 6 April 2019, and will do it again shortly before midnight UTC on 20 Nov 2038.; The L1 Standard Positioning Service (SPS) signal, at 1575.42 MHz (154 × 10.23 MHz), RHCP, is available to all users, although parts of the data frame are encrypted. Each satellite’s L1 uses its own Coarse Acquisition (C/A) Code, a 1023-bit, pseudo-random spreading code running at 1.023 Mb/s (hence frame period 1 ms) that uniquely identifies the originating satellite. The navigation message frame, containing ephemeris and almanac data, is modulated onto this coded stream using 50 b/s BPSK, so each data bit is spread over 20 C/A frames. The almanac is information about the satellite’s status and time. Ephemeris specifies orbit. A GPS receiver needs ephemeris from four satellites to accurately determine its own latitude, longitude, and altitude (a 3D fix). It takes 12.5 minutes (750 seconds) for a satellite to transmit its entire 37,500-bit navigation message frame at 50 b/s, so L1 ephemeris data repeats every 30 seconds. Originally, a feature called Selective Availability (SA) was applied to degrade the accuracy of L1 ephemeris. The military turned it off in 2000, so unimproved L1 position fixes are now accurate to about 15m; however, see EGNOS, WAAS.; The L2 1227.6 MHz (120 × 10.23 MHz) Precise Positioning Service (PPS) signal is encrypted. It uses a different spreading code, the Precise (P) Code, at a 10.23 Mb/s chip rate. This allows it to be more accurate than L1.; Later generations of GPS satellites added new broadcast signals called L3 (1381.05 MHz), L4 (1379.913 MHz), and L5 (1176.45). Satnav systems using one or more of these in addition to L1 can achieve accuracies of less than a meter.
SAW: Surface Acoustic Wave. A metal film inter-digital transducer (IDT) on an elastic, piezoelectric solid can convert signal voltage variations into acoustic waves at the free surface of the solid. Another IDT can do the opposite, converting surface acoustic waves into voltage variations. This SAW device is used for very compact RF filters and RF oscillators that work from the interaction of acoustic rather than EM waves.
SBC: Single Board Computer. A removable circuit board containing the processor, chip set, RAM, video, and everything needed to run a computer. Used in backplane systems. The blade server is a common type of SBC.
SBL: Space-Based Laser.
SBOM: Software Bill of Materials. A list of all the distinct components that make up a software package. It’s become standard practice to include a SBOM with for-purchase software, and it’s mandatory when dealing with the US government.
SBSRAM: Synchronous Burst Static Random Access Memory. See RAM.
SC: (1); Square Connector. See fiber.; (2); Switched Capacitor. Just what it sounds like, a capacitor with a connection to the rest of the circuit that’s closed and opened by a switch. Common in S/H and T/H circuitry.
SC#: Single Chip. See JEDEC.
SCA: Software Communications Architecture. A standard for implementing software-defined radio, developed for JTRS. It uses CORBA and driver software to eliminate hardware dependencies from software design.
SCADA: Supervisory Control And Data Acquisition. A.k.a. distributed control system (DCS) or industrial control system (ICS). Any automated system that monitors and controls remote sites, particularly utilities (e.g., power, water, or gas plants), industrial systems (e.g., factories, refineries, pipelines), and facility support nodes (e.g., climate-control and power systems throughout a building or building complex). A SCADA command-and-control node is called a remote terminal unit (RTU) or programmable logic controller (PLC).; The evolution of the SCADA from a proprietary stand-alone system into an Internet-connected WAN has made it a security issue. Most SCADA networks are part of the Dark Internet, hence don’t have a Web presence, but there are specialized search engines for finding them. A very sophisticated worm called W32.Stuxnet, released in 2009-2010, infected many SCADA systems. Others have followed. Older industry security standards, such as the Critical Infrastructure Protection (CIP) standard released in 2006 by the North American Electric Reliability Corporation (NERC), were not equal to the threat.
scalar processor: A processor with a single pipeline.
schema: A description of an XML document type, laying down rules for the structure, content, etc. that XML documents of the defined type must have. There are schema languages for writing schemas, such as DTD and the W3C’s XML Schema. A software tool called a schema processor validates instances of a document type by testing them against the schema.; HTML Web pages written to one of the DTD schemas (Strict, Transitional, or Frameset) are the best-known XML document type, but certainly not the only one.
Schmidt-Cassegrain: See telescope.
Schottky diode: See diode.
SCIP: Secure Communications Interoperability Protocol. US government standard for encrypted data and voice comms, adopted in 2004. It can work with a wide range of systems, including wireless and STE but not the older STU. Like a fax protocol, it allows the called and calling devices to negotiate a mutually acceptable data rate and other parameters.
SCM: (1); Single-Chip Module. A circuit board hosting a single IC chip.; (2); Source Code Management. Also called a version control system. Refers to software applications, such as CVS, Git, and Subversion, that retain multiple versions of protected files and track what changes have been made to them, by whom, and when.; (3); Software Configuration Management. An activity that almost certainly uses (2) but is not itself software.
SCOP: Seasonal Coefficient Of Performance. See COP.
SCORM: Shareable Content Object Reference Model. A standard for online education software, originated by the US Department of Defense in 2000 but since overtaken by other, more capable systems. The original government site for it, http://www.adlnet.org/scorm, has moved to https://adlnet.gov/past-projects/scorm.
SCP: Secure Copy Protocol. Both a network protocol and any of several command-line utilities to implement the protocol, SCP is based on SSH. It’s used to securely transfer files between hosts through TCP port 22, but is considered obsolete.
SCPI: Standard Commands for Programmable Instrumentation. A standard set of ASCII commands to control programmable test and measurement devices. It’s based on IEEE-488.2 (GPIB), but works over a variety of connections.
.scr: Under filename extension, see .scr.
SCR: Silicon-Controlled Rectifier. See thyristor.
scrambler: A pseudo-randomizer used by a RF transmitter to spread signal energy evenly across the operating bandwidth. This renders the signal less susceptible to noise and reduces unwanted spectral components outside the band.
scramjet: Supersonic Combustion Ramjet. A ramjet without an inlet manifold to slow down incoming air, so that combustion takes place at supersonic speeds.
scripting language: Any programming language that is A) relying on an interpreter as opposed to a compiler, and B) generating real-time commands or managing connections between other programs and objects. It might lack the Turing-complete features of a full programming language, but that’s not a rule. Scripts – programs written in one of these languages – can be stand-alone text files, or embedded in text files of other types. They can also exist purely in memory.; On the World-Wide Web, server side scripts (a.k.a. back-end scripting) are those that the Web server is responsible for running, so the client’s browser receives just the HTML-format results, not the actual code. Client side scripts (front-end scripting) are passed to the requesting browser to be run. Most scripting languages for the Web are designed for one end or the other. Popular server-side languages include PHP, Python, Ruby, and ASP.NET. JavaScript is standard for the client side.; Like all interpreted code, scripts can be created and modified more quickly than compiled software, but aren’t as efficient at runtime. They save human time at the expense of computer time. They’re most frequently encountered nowadays in the form of Web scripts for dynamically generating HTML. However, scripts have a long history, and many other uses, including malicious ones (see XSS).
Scrum: See process management.
SCSI: Small Computer System Interface. A set of obsolete parallel data bus standards originating from the 1981 Shugart Associates Systems Interface (SASI). There are three iterations of the SCSI standard (SCSI-1, SCSI-2, and SCSI-3), each of which supports a number of different features. Terms such as Fast, Wide, and Ultra have specific meanings when referring to a SCSI implementation:; Narrow – A retroactively applied term for the original 50-pin, 8-bit SCSI cable.; Wide – A 68-pin cable carrying 16 bits of parallel data.; Fast – 10 MHz signaling speed, as opposed to the original SCSI-1 5 MHz.; Ultra – 20 MHz signaling speed. Also called SPI, or Fast20.; Ultra2 – 40 MHz signaling speed. Also called SPI-2, or Fast40.; Ultra3 – 80 MHz signaling speed. Requires LVD. Also called SPI-3, or Fast80.; Ultra160 – A standardized subset of Ultra3 SCSI features established by Adaptec, HP, and IBM. It’s guaranteed to run at 160 MB/s.; Ultra320 – Still in development (2000). Also called SPI-4.; Differential – SCSI-2 introduced High Voltage Differential (HVD) signaling. SCSI-3 added Low Voltage Differential (LVD) signaling. If an implementation uses HVD, it’s called just “Differential”. If it uses LVD, it should specifically say so.
SCT: Single-Chip Transceiver.
SCXI: Signal Conditioning Extensions for Instrumentation. A line of National Instruments signal conditioning and data acquisition products for PC-based instrumentation systems.
SCXML: State Chart Extensible Markup Language. XML offspring for describing state machines.
SD: (1); Secure Digital (card). The SD Card Association’s standards for removable flash memory cards, derived from the MMC. Despite similarities, these are not the same as the SIM cards used for mobile wireless devices.; The original SD Card measures 24 × 32 × 2.1 mm, with 9 contacts on one edge. The 11-contact miniSD is 21.5 × 20 × 1.4 mm, and the 8-contact microSD 15 × 11 × 1.0 mm. All of these use 3.3 VDC power and logic levels, and support both the SPI bus and the standard’s own SD bus. There are adapters to physically connect the cards to different slot sizes.; These form factors and their slots are used for four different storage standards:; SDSC (SD Standard Capacity) – The original. The card can hold from 1 MB to 2 GB. 4 GB versions exist, but aren’t the same as the 4 GB SDHC models.; SDHC (SD High Capacity) – From 4 to 32 GB. The slots are backward-compatible with SDSC.; SDXC (SD Extended Capacity) – (2009) From 32 GB to 2 TB, although no one has actually made cards for the higher capacities. Not available in miniSD.; SDIO (SD Input/Output) – Provides other capabilities to the host device in addition to storage. The cards fit SD or miniSD (not microSD) slots, but the card sizes can vary depending on the additional hardware they host, e.g., antennas and wireless adapters.; In addition to several proprietary variations of the SD card standards, there’s the Embedded SD standard for non-removable cards, to provide low-power, high-capacity storage for embedded devices. Though electrically compatible with SDHC, it uses its own set of physical sizes. SD Express (2018) and microSD Express (2019) are bus standards that use a NVMe or PCIe interface to transfer data to/from a SD card at up to 985 MB/s full duplex.; (2); See standard deviation.
SDD: Secure Data Device. A STU without voice capability.
SDH: (1); Synchronous Digital Hierarchy. A European standard for optical telecommunications networks, created around 1986 by ETSI for Europe, and adopted by ITU for everywhere but North America (where a variant called SONET rules) and Japan (where another variant of SDH is used). It permits different digital carrier standards (DS-#, E-#, J-#) to interoperate.; The fundamental SDH carrier is the 155.52 Mb/s STM-1. What makes SDH synchronous is the use of a single, highly accurate master clock (the PRC). Any other clocks in the system must be slaved to the master. See SEC.; (2); Secure Digital Host. A device that hosts a SD card.
SDHC: Secure Digital High-Capacity. See SD.
SDI: (1); Serial Digital Interface. A digital video signal format that carries up to 2.97 Gb/s on 75 Ω BNC coaxial cable. It’s standard for professional video systems such as TV stations and movie studios. Because it lacks DRM, licensing restrictions have been used to keep it out of consumer products.; (2); Single Document Interface. The simplest sort of GUI for Windows apps. It consists of a text output panel and a basic file access/print/edit menu & toolbar at the top.; (3); Source/Destination Identifier. See ARINC-429.
SDIO: Secure Digital Input/Output. See SD.
SDIP: Shrink Dual In-line Package. See JEDEC.
SDK: Software Development Kit. A set of utilities for writing software applications.
SDLC: (1); Synchronous Data Link Control. See data protocol.; (2); Software Development Life Cycle.
SDMA: Space Division Multiple Access. A simple approach to frequency re-use: use directional antennas and aim them in different directions.
SDN: Software-Defined Networking. An alternative approach to computer networks that defines layers of functionality, separating data from routing.
SDP: (1); Service Discovery Protocol. A protocol that allows devices on a network to find out what other devices are on the network, and what their capabilities are. Bluetooth and Jini use SDPs.; (2); Scenario Design Power. See TDP.
SDP4: Simplified Deep-space Perturbations 4. See SGP.
SDR: Software Defined Radio. Radio that processes RF signals mostly or entirely in software rather than dedicated hardware. The objective is a smaller radio that handles multiple signal formats and is easy to modify. High-speed ADC and DAC are software tasks anyway. It still needs a hardware antenna, and usually but not always a front-end LNA and hardware filter.; Advanced SDR that analyzes and adapts to its RF environment using an artificial neural network is dubbed cognitive radio.
SDRAM: Synchronous DRAM. Refers to both the original SDRAM memory technology and its descendants.
SD-ROM: Super Density Read-Only Memory. A SanDisk Corporation implementation of the SD standard, with very long life (100+ years) but lower capacity (128 MB). It supports the SPI and SD buses.
SDSC: Secure Digital Standard Capacity. See SD.
SDSL: Symmetric Digital Subscriber Line. See DSL.
SDU: (1); Spectrum Display Unit. A device that provides a frequency-domain display of a portion of the radio spectrum. Useful for tuning.; (2); Service Data Unit.
SDXC: Secure Digital Extended Capacity. See SD.
SEC: (1); Single Edge Contact. See Slot #.; (2); Synchronous Equipment Clock. A slave clock in a SDH network element, driven by a master clock. See PRC.
SECAM: Séquentiel Couleur Avec Mémoire. An obsolete analog color TV broadcasting standard in France, Eastern Europe, and Russia. It employs 625 lines refreshed at 50 Hz, with 220VAC power, and is quite similar to PAL. See NTSC.
SecOps: Security Operations. See process management.
Secure Boot: See Trusted Computing.
SED: (1); Surface-conduction Electron-emitter Display. A mid-2000s flat panel display (FPD) technology, similar to the FED, that never made it to the consumer market. It uses a single nanometer-scale electron emitter for each pixel. It uses less power than a LCD, with color superior to a PDP. See FPD.; (2); Self-Encrypting Drive. A hard drive with built-in support for encryption to be managed by system firmware (see TPM), as a defense against malware. This encryption is not normally apparent to users.
SEM: Scanning Electron Microscope. It uses a focused electron beam for microscopic imaging at resolutions to less than 1 nm, much finer than an optical microscope. For even finer resolution, there’s the STM.
semaphore: A program-control bit flag or status variable, especially for a RTOS.
semiconductor: A material that can act as either an electrical conductor or an electrical insulator. Semiconductors, as the raw material of diodes and transistors, are the basis for virtually all modern electronics.; Electrons in free space occupy a continuum of energy states. Electrons in a solid, exhibiting their wave nature, have energy states that they occupy (bands) separated by energy states that they can’t occupy (band gaps). Each material has distinct band gaps set by its atomic lattice. A large band gap means more energy is needed to generate free electrons, hence conductivity.; The band gaps of semiconductors can be dramatically changed by doping: adding small, precise amounts of certain impurities to the semiconductor during manufacture. Negative doping means using a dopant that adds free electrons, hence an overall negative charge, to the crystal lattice. Positive doping uses a dopant that binds free electrons, creating regions of positive charge (holes).; Although an n-type semiconductor has more free electrons, and p-type has more holes, both kinds of charge carrier are present in both types of semiconductor. The charge carriers are free to flow as current in response to an applied electrical field. What’s physically flowing in both types is electrons, but the fiction that the positive charge carriers flow means that the same analysis and design approach works for either type.; Crystalline silicon is by far the most common semiconductor material. Its sub-types are monocrystalline silicon (mono-Si), with its uniform coloring, superior charge mobility, and higher cost; and polycrystalline silicon (poly-Si), recognizable by the irregular, flake-like grain boundaries running through it. Germanium is a distant second to silicon. It has higher charge mobility, hence can handle higher frequencies, but is more expensive and less heat-tolerant. Some other materials:; GaAs – Gallium Arsenide. Outperforms silicon at high frequencies and power levels due to higher charge mobility, but costs much more. Researchers have developed high-performance GaAs microcircuits on a substrate of epoxy-coated cellulose nanofibril, made from wood pulp. This technology has the potential to greatly decrease GaAs fabrication costs without loss of performance.; GaN – Gallium Nitride. Pronounced “gan”. The replacement for GaAs and silicon in high-end applications, notably RF amplifiers, that require higher power or frequency, greater efficiency, or greater heat and EM tolerance. GaN-based devices also replaced vacuum tubes in some of their remaining uses, such as the TWTAs common in radar systems and satellites. Because of its large 3.4-eV band gap energy, GaN is optically clear (photons at optical wavelengths aren’t energetic enough to be absorbed) and can emit light in the green to UV range. This makes it useful for LED lighting.; Researchers hope to replace GaN with semiconductors based on aluminum nitride (AlN) or diamond.; Ga₂O₃ – Gallium Oxide. Up-and-coming n-type-only semiconductor as of 2021, with a nearly 5-eV band gap and exceptionally high breakdown voltage, but poor thermal conductivity.; IGZO – Indium Gallium Zinc Oxide. A good n-type but not p-type semiconductor, used for very thin, flexible, high-resolution LCDs and OLED displays.; InGaAs – Indium Gallium Arsenide. For special applications, which sometimes also use InP (indium phosphide) and InAlGaAs (indium aluminum gallium arsenide).; InGaN – Indium Gallium Nitride. More efficient for LED lighting than GaN.; InSb – Indium Tin. Has the highest known electron mobility in a semiconductor.; ITO – Indium Tin Oxide. Being researched for use in high-resolution flat-panel displays.; SiC – Silicon Carbide. Also known as carborundum. A very hard material that tolerates extreme temperatures. It handles higher voltages than GaN, but can’t run at the high frequencies GaN can, and its electron mobility is actually lower than that of silicon.; SnO – Tin Oxide. As of 2016, the only stable, 2-dimensional (i.e., one atom thick) p-type semiconductor.; TiS₃ – Titanium Trisulfide. Experimental. In layers just a few molecules thick it has a bandgap of 1.02 eV, similar to that of silicon, but with much higher carrier mobility.
SEO: Search Engine Optimization. Designing a Web site to improve how search engines rank it. (See content farm.) This is known as SEO poisoning when malware sites do it.
SEP: Spherical Error Probable. The 3D version of CEP.
SEPIC: Single-Ended Primary Inductance Converter. A type of switching-mode, non-isolated, non-inverting, buck-boost DC power regulator.
sequential code: A type of trellis code that pre-dates the Viterbi algorithm. Its decoding complexity is a linear rather than an exponential function of constraint length, so it uses large constraint lengths. It’s vulnerable to buffer overflow when E_b/N₀ is poor.
SerDes: Serializer-Deserializer. A device that converts parallel data to a serial form for transmission, and converts incoming serialized data to parallel form for processing. As of 2020, there are no non-proprietary standards for doing this.
serialization: Conversion of a software data object, such as a database or a set of program data structures, into a format that can be handled by networks and file systems. Many programming languages have this feature. Various proprietary and standard formats have been created for the purpose. Some languages or applications instead use an established general-purpose format, such as XML, which has the advantage of being human-readable. The reverse process is called de-serialization.
serial port: A computer port for two-way devices, using asynchronous signaling. Serial data is necessary for transmission over distances. It almost always uses a DE-9P physical port and the RS-232 signaling protocol. The spread of USB and FireWire have made serial ports an endangered species.
server: A computer that provides some service to multiple clients over a network. For services with many users, it will be an array of blade servers working together to supply the very large bandwidth and processing power the job demands. The word “server” is also used to mean just the software program that the hardware runs to manage the service. Some types:; database server – Provides access to a database, sometimes including the capability for users to modify it.; file server – Maintains stored files to be shared by users. Making these files available over the Internet involves a TCP/IP protocol such as FTP or HTTP. The NAS is designed to be a file server, as opposed to a general-purpose server assigned to the job.; license server – Also called a key server. To ensure that purchasers of their software don’t run more copies than they pay for, many commercial software vendors use floating licenses. Starting the installed software requires the customer’s computer to contact a license server and claim one of the available licenses. The software releases the license to the server when it exits. The license server can be on the customer’s own network, or, less usually, it can be maintained by the vendor.; mail server – Receives, stores, and forwards e-mail, using IMAP, MIME, POP, and SMTP.; proxy server – An Internet server that acts as an intermediary, performing operations or accessing resources on behalf of remote users who can’t or don’t wish to perform those actions directly. A common use is to provide anonymity: it sends requests or data to the destination with the proxy server’s IP address rather than the user’s. Of course, this kind of anonymity assumes the proxy itself is trustworthy; there are many free proxy services on the Internet, but it’s best not to trust a free product or service if you don’t know how it’s funded. Users seeking stronger anonymity can establish accounts on multiple proxies, ideally located in different countries, and link the accounts in a chain, or use a darknet that does this. Each link of the chain increases the protection but further slows the user’s connection. This is a problem for, say, streaming video, much less so for e-mail or file-sharing.; Web server – Handles Web page requests. The old HTML-only Web pages could be supplied by a file server, but Web pages with scripts and other, increasingly complex interactive content are more efficiently and securely presented by a specialized Web server. Its job is becoming more demanding, and also risky (see CGI). The most popular Web server softwares as of 2013 are Apache, Nginx, GWS (Google Web Server), and Microsoft’s proprietary IIS.; See also virtual private server.
SET: Secure Electronic Transaction. An Internet protocol for payments.
SETI: Search for Extraterrestrial Intelligence. Nothing yet...
SFDR: Spurious-Free Dynamic Range. A figure of merit for DACs and ADCs: the ratio of carrier RMS amplitude to peak noise RMS amplitude in the bandwidth of interest, usually in dBc (dB relative to the carrier).
SFF: Small Form Factor. Originally, any of a set of narrow standards established by a MSA called the Small Form Factor Committee, which was founded in 1990 to develop mechanical standards for laptop hard drives but has done many other things. The term has since come to mean any device or system smaller than the standard for its field, but it has at least two specific meanings:; For computers, it means a miniaturized PC based on a motherboard smaller than the micro-ATX.; For optics, it means the SFF Committee’s fiber-optic transceiver module standards (SFF-8472, INF-8074i, etc.) or fiber connector standards, some of which are also SFP. Board footprints for pin-through-hole (PTH) SFF pin arrays include 1×9, 2×5, 2×9, and 2×10.
SFODB: Spaceborne Fiber-Optic Data Bus. Created by DoD and NASA and defined in IEEE 1393 as a 1 Gb/s fiber-optic network to support the real-time, on-board data handling requirements of remote sensing spacecraft. The network is fault-tolerant, highly reliable, and capable of withstanding launch stresses and the space environment. It emphasizes small size, light weight, and low power.
SFP: Small Form-Factor Pluggable. Standards established by the Small Form Factor Committee for fiber-optic transceiver modules that plug into a board without soldering. INF-8074i seems to be the principal one. INF-8077i is for XFP, a 10 Gb/s SFF pluggable module; “X” is the Roman 10, for 10 Gb/s. Similar to what happened with SFF, it has since come to mean any smaller-than-standard module that can be plugged into a circuit without soldering.
SFR: (1); Special Function Register. A microprocessor register reserved for pre-defined uses, as opposed to a general-purpose register (GPR) that the software can use any way it chooses.; (2); Sodium-cooled Fast Reactor. See reactor.
SFTP: (1); A secure version of FTP for sending files over untrusted networks. It’s now a standard part of SSH, so it’s also called SSH File Transfer Protocol.; (2); Simple File Transfer Protocol. Proposed Internet protocol that was never adopted.
SFX: Small Form-factor Extended (more or less). Intel’s 1997 standard for a smaller PSU, originally to support PCs built around micro-ATX motherboards. It specifies the same +3.3, +5, and ±12 VDC levels (no -5 V) and 20-pin motherboard connector as ATX. It dropped the -12 V level and added the 24-pin connector when it evolved into the 2003 SFX12V standard.; SFX12V v3.1 (March 2005) defines five different physical configurations of PSU for compact PCs. They include the 63.5 mm high x 100 mm wide x 125 mm deep standard, with a top-mounted 80 mm or rear-mounted 60 mm intake fan; the PS3, essentially a shortened ATX12V (86 × 101.4 × 150 mm) for micro-ATX systems with higher power needs; and the low-profile 50 × 100 × 125 mm with a rear-mounted 40 mm intake fan.
SGML: Standardized General Markup Language. The mother of all markup languages, standardized in 1986 by the ISO. It was too big and complex for the subsequent development of the WWW, so other markup languages (HTML, XML) in use on the Web are simplified applications of SGML.
SGP: Simplified General Perturbations. The earliest of five models for satellite orbits, developed in 1966 for near-Earth (orbital period less than 225 minutes) objects. The “perturbations” are the influences of the Sun and Moon, and complicated distortions of Earth’s gravity. Later versions of the model are SGP4 (1970) and SGP8 (1980). SDP4 and SDP8 were derived from SGP4 and SGP8, respectively, for modeling deep-space orbits (D stands for “Deep-space”). The NORAD standard in the 1990s was to use a combination of SGP4 and SDP4. See 2LMES.
SGRAM: Synchronous Graphics Random Access Memory. A type of SDRAM designed specifically for graphics.
SGX: Software Guard Extensions. A security feature of newer Intel processors that establishes execution areas in on-chip cache memory to which the computer’s operating system has no access. This protects the CPU in the event that the operating system is compromised. Hacks for SGX exist, but are difficult.
S/H: Sample and Hold. A step in A/D conversion that samples the amplitude of the input analog signal (typically with a switched capacitor), and then opens the switch and holds its output signal at the sampled level for some period of time before taking the next sample. Subsequent circuitry converts the stepped levels to digital values. Compare T/H.
SHA-#: Secure Hash Algorithm. See encryption.
shadow RAM: A write-protected area of RAM that loads a copy of the system BIOS at bootup. Default settings activate shadow RAM on a system with more than 2 MB of memory. It can load the video BIOS too if this function is enabled in the system configuration.
Shannon limit: In his seminal 1948 paper, American engineer & mathematician Claude E. Shannon (1916-2001) demonstrated that any digital communications channel has a maximum achievable error-free capacity for a given power level: C = W log₂ [1 + P/N], where C is the capacity (in bits/second), W is the bandwidth (in hertz), P is transmitter power (in watts), and N is noise power (in watts) – assuming only additive white Gaussian noise (AWGN). This, also called the Shannon-Hartley theorem because it built on the work of American engineer Ralph Hartley (1888-1970), began the science of information theory. When comparing a transmission to the Shannon limit, it’s standard to compare the power actually used to the theoretical minimum power requirement. See also Nyquist rate.
SHARC: Super Harvard Architecture. A 1990s family of 32-bit, floating-point, programmable DSPs from Analog Devices.
SHF: Super High Frequency. 3-30 GHz. See RF.
shot noise: The noise caused by random fluctuations in the motion of charge carriers in a conductor, or the emission of photons in a wireless transmission.
.shs: Shell Scrap. See filename extension.
SI: Système International d’Unités. Better known as the metric system. It has seven base units (meter, kilogram, second, ampere, kelvin, mole, candela) from which it derives all the others. Here’s a JavaScript-powered metric conversions page for converting between metric and English/Imperial units.; SI defines prefixes for all of the units to indicate powers of ten. As of 2022, these prefixes and their meanings are:; quecto (q) – 10^-30, or 0.000000000000000000000000000001; ronto (r) – 10^-27, or 0.000000000000000000000000001; yocto (y) – 10^-24, or 0.000000000000000000000001; zepto (z) – 10^-21, or 0.000000000000000000001; atto (a) – 10^-18, or 0.000000000000000001; femto (f) – 10^-15, or 0.000000000000001; pico (p) – 10^-12, or 0.000000000001; nano (n) – 10^-9, or 0.000000001; micro (µ) – 10^-6, or 0.000001; milli (m) – 10^-3, or 0.001; centi (c) – 10^-2, or 0.01; deci (d) – 10^-1, or 0.1; deca (da) – 10¹, or 10; hecto (h) – 10², or 100; kilo (k) – 10³, or 1,000; mega (M) – 10⁶, or 1,000,000; giga (G) – 10⁹, or 1,000,000,000; tera (T) – 10¹², or 1,000,000,000,000; peta (P) – 10¹⁵, or 1,000,000,000,000,000; exa (E) – 10¹⁸, or 1,000,000,000,000,000,000; zetta (Z) – 10²¹, or 1,000,000,000,000,000,000,000; yotta (Y) – 10²⁴, or 1,000,000,000,000,000,000,000,000; ronna (R) – 10²⁷, or 1,000,000,000,000,000,000,000,000,000; quetta (Q) – 10³⁰, or 1,000,000,000,000,000,000,000,000,000,000
sidereal: Having to do with stars or constellations. Earth’s rotation defines the north and south poles of the equatorial system, which describes star positions. For any location on Earth, the local meridian is an arc extending between these poles and passing through zenith (directly overhead), and sidereal time is the hour angle from this arc to the celestial meridian – the local meridian at noon on the vernal equinox. The sidereal day, reckoned by the time it takes for a star to cross the meridian twice, is four minutes shorter than the solar day. The sidereal year averaged 365 days 6 hours 9 minutes 9.55 seconds as of 1955, and is increasing at the rate of 0.000095 sec/year.
SIEM: Security Information and Event Management. A comprehensive approach to preventing or containing cyber-attacks, from conventional viruses and worms to phishing and other forms of social engineering.
signal generator: A laboratory instrument that generates electrical signals with sinusoidal waveform and operator-selected frequency and amplitude. It might have other configurable features, but is less capable than an AFG or AWG. Compare also pulse generator.
SIL: Software In the Loop. A type of system testing in which software simulates the actions of hardware system components. SIL is typically an initial stage of testing prior to the more time-consuming and potentially risky HIL.
SIM: Subscriber Identity Module. An ETSI standard integrated circuit that authenticates the user to mobile wireless networks with a stored IMSI. This circuit is usually on a removable SIM card with a UICC form factor – originally the 1FF (SIM), then the smaller 2FF (mini-SIM), 3FF (micro-SIM), and 4FF (nano-SIM). The SIM card also provides storage for applications and user-generated data.; The SIM swap is a popular type of cybercrime: trick a mobile-phone service provider into transferring someone’s number to a new SIM card held by the criminal, and then use the stolen phone account to gain access to the owner’s other accounts by, for example, requesting a password reset, or intercepting and validating authentication requests.; GSM was the first system to require SIMs. GSM phones are usually configured to accept SIMs for just one particular country or network provider (but see root).; The trend is toward embedded SIMs (eSIMs) or integrated SIMs (iSIMs), which subscribers can’t remove. An eSIM is a firmware version of a SIM, residing in a dedicated reprogrammable storage chip on the device motherboard. An iSIM is built into the system-on-a-chip (SoC) that runs the device. These types of SIM make it easier to switch service providers, but they take away the user’s ability to use the same SIM on multiple devices, and make it harder for users to prevent service providers from tracking them.
SIMD: Single Instruction Multiple Data. A method that allows a processor to perform the same operation, using a single instruction, on multiple independent sets of data operands. Helpful for the vector operations used in multimedia and signal-processing applications. Intel uses SIMD for multimedia instructions in the Pentium III. See SSE, and compare MIMD.
SIMM: Single In-line Memory Module. An obsolete type of DRAM printed circuit card with either 30 or 72 flat contacts along each side of the bottom edge. Each pair of opposing contacts is linked as a single functional contact.; The 30-contact SIMM has 256 kB to 16 MB of 8-bit RAM. Like the older SIPP it replaced, it measures about 3.5" × .75", and must be installed in pairs for 16-bit 80286 PCs and groups of four for 32-bit 80386 or 80486 PCs.; The 72-pin SIMM has 4 to 32 MB of 32-bit EDO DRAM or FPM DRAM, so it must be installed in pairs on 64-bit Pentium systems. It measures 4.25" × 1" (height can vary), with an off-center notch dividing the contacts into two groups of 36 and preventing incorrect installation. It snaps into place at an angle to the motherboard.
SIMO: (1); Single Input, Multiple Output. An RF link that uses multiple receiving antennas to maximize reception of a signal coming from one transmitting antenna. It can follow a phased array approach, or something simpler. Compare MIMO, MISO, SISO.; (2); Slave Input/Master Output. See SPI.
single mode: See SM.
single-phase: See three-phase.
SINR: Signal to Interference-plus-Noise Ratio. A more specific form of SNR.
SiP: (1); System in Package. Vertically stacking a set of IC chips to permit shorter interconnects and higher density than a MCM, more cheaply than a SoC. Compare also SoP.; (2); Silicon Photonics. Optical signaling in silicon-based circuits.
SIP: Session Initiation Protocol. Developed by the IETF as an Internet signaling protocol to establish point-to-point connections for conferencing, telephony, IM, and events notification. It can provide authentication and encryption, and runs over TCP or UDP. It’s seen as having applications for Internet appliances.
SIPP: Single In-line Pin Package. A one-side printed circuit board (PCB) with a single row of 30 pins projecting like a comb, making it fragile and tricky to insert. It measures about 3.5" × .75", minus the pins. Used for PC DRAM in the mid to late 1980s, it holds 256 to 1024 kB of 8-bit memory. That means 16-bit 80286 PCs require an even number of SIPP modules, and 32-bit 80386 PCs have to install them in sets of four. The later, more robust SIMM inherited these issues.
SISO: Single Input, Single Output. A conventional RF link with one antenna at each end. Compare MIMO, MISO, SIMO.
SKU: Stock Keeping Unit. A tracking number, typically used to identify a type of product or to track an individual unit. It’s most often found as a barcode, sometimes a QR code.
skywiring: Connecting points on a circuit board with wires, rather than through traces integral to the board. It’s the hardware equivalent of a kluge.
SL: Under Intel, see 80386, 80387, 80486, 80487.
SLA: Sealed Lead-Acid. See battery.
SLAC: Subscriber Line Audio-processing Circuit.
SLAR: Side-Looking Airborne Radar.
SLC: (1); Single-Level Cell. See flash memory.; (2); A designation introduced by IBM for its hybrid 16/32-bit 386 and 486 CPU chips (see bus). It was a power-saving design with an integrated cache. “SLC2” indicated a chip with its internal clock doubled. Some versions were made by Cyrix and Intel. Under Intel, see 80386, 80486.; (3); Single-Layer Ceramic. Refers to a type of ceramic capacitor.
SLF: Super Low Frequency. 30-300 Hz, usually. See RF.
SLI: (1); Scalable Link Interface. A 2004 PCIe-based technology from NVIDIA Corp. that allows graphics cards with identical processors to collaborate on a single video output.; (2); Scan-Line Interleave. A defunct video technology from a company bought by NVIDIA. No relation to (1).
SLIC: Subscriber Line Interface Circuit. A circuit that connects a phone or other device to a telephone line.
SLIP: Serial Line Internet Protocol. A member of the TCP/IP protocol suite at the Link layer (the Data link layer of the OSI model). It allows a computer to exchange IP data with a networked host over a serial cable or dial-up modem connection, thereby linking into the host’s network. It was supplanted by the more versatile and complex PPP.
Slot #: In 1997-1998, Intel came out with Slot 1, a 242-contact motherboard interface for Pentium II CPUs. It’s electrically identical to Socket 8, but with a different physical interface to accept a single-edge contact (SEC) cartridge. Intel claimed this was for technical reasons, but it was widely viewed as an attempt to squeeze out rival CPU manufacturers, who had been making their chips compatible with Intel chip sockets. The cartridge contains the CPU and a separate L2 cache with its own backside bus at half or full core speed.; Slot 1 debuted with a 66 MHz external bus to the motherboard. Later versions go to 100 MHz. Versions exist for Pentium III and Celeron SECs, and also for laptops. Slot 2 is for the Xeon II and III server chips. It has a 100 MHz external bus and can handle larger L2 caches. AMD responded to all this with Slot A for its Athlon (K7) chip. See Socket #.
SLR: Single-Lens Reflex. A camera technology that uses a mirror and prism to let the photographer view the actual lens image, rather than a potentially different image from a viewfinder. Modern digital cameras instead use LCD viewfinders; see DSLR.
SM: Single-Mode. A type of fiber-optic cable with a very small (9 µm) central core. This guarantees a straighter propagation path than MM, so delay variations are smaller, permitting higher bandwidth and transmission rate. On the other hand, the electronics at the end must be more precise, hence expensive, and a laser rather than a cheaper LED is required as a source. Maximum transmission distance varies, but is in excess of 2000 m.
SM#: See JEDEC.
SMA: (1); See cable connector.; (2); Shape Memory Alloy. A metal alloy that, when heated above some critical temperature, reverts to a shape to which it was previously set while heated, removing any deformation that has occurred. The shape-memory effect degrades with time and repeated deformation.
smart card: See ICC.
smartdrv.exe: If loaded as a device in the MS-DOS config.sys, smartdrv performs double buffering, providing compatibility with some hard drive controllers that can’t work with EMM386 or Windows memory. It uses 2 kB of conventional memory. If called by autoexec.bat, smartdrv also performs disk caching to speed the hard drive. The default setting will take up to 2 MB of extended memory.
SMB: See cable connector.
SMBus: System Management Bus. An application of the I2C bus. It’s commonly used on PC motherboards to control voltage regulators and clock frequency.
SMC: See cable connector.
SMD: Surface-Mount Device. See SMT.
SMDS: Switched Multimegabit Data Service. A WAN (public) packet-switching service intended for connecting LANs at T3 rates. It can carry non-LAN traffic, and rates are approximately 56 kb/s to 34 Mb/s.
SMF: Single-Mode Fiber. See SM.
SMIL: Synchronized Multimedia Integration Language. A language for creating Web pages, released in 1998.
Smith chart: A polar chart devised in 1939 by American engineer Phillip H. Smith (1905-1987) to represent graphically, rather than mathematically, the complex impedance anywhere along a transmission line. It has circles of constant resistance and arcs of constant reactance. It’s used for impedance matching, where the center of the chart is a normalized resistance of 1.0 – a match. See characteristic impedance, VSWR.
SMM: System Management Mode. A high-privilege state built into x86 CPU firmware since 1993 to simplify troubleshooting at a level above the OS and kernel. This is the mode the BIOS uses, and UEFI also uses it to provide support for legacy BIOS features. See also ASLR, DEP.
SMP: Symmetric Multi-Processing. Two or more similar processors controlled by a single OS and sharing memory and I/O resources – as opposed to asymmetric multi-processing, where the processors or cores have master/slave roles and perhaps different operating systems. Not to be confused with SMT.
SMPS: Switching Mode Power Supply. Any switching power regulator.
SMR: (1); Specialized Mobile Radio. A dispatch radio and interconnect service for businesses. Covers frequencies in the 220 MHz, 800 MHz and 900 MHz bands.; (2); Shingled Magnetic Recording. (2014) A Seagate HDD technology offering high storage density and reliability, at the cost of lower speed.; (3); Small Modular Reactor. See reactor.
SMS: Short Message Service. A near-universal protocol developed in the 1980s for sending short, text-only messages (up to 1120 bits, enough for 160 7-bit characters or 140 8-bit octets) on mobile wireless devices, with storage and forwarding handled by the wireless provider’s Short Message Service Center (SMSC). This protocol, first used with GSM phones, is the reason for the 140-character limit imposed by Twitter and some other social media services. Its data is carried over the same digital voice channels as mobile phone calls. In part because of this, message delivery isn’t guaranteed.; Wireless providers tried extensions to get around the limitations of basic SMS. Concatenated SMS could send longer messages broken into a series of 1120-bit packets. Enhanced Messaging Service (EMS) supported still longer messages with text formatting and multimedia content. However, these variations were rendered obsolete by the MMS and RCS industry standards, and proprietary services such as Apple’s iMessage and Facebook’s WhatsApp and Messenger, all of which are multimedia-capable and use IP rather than riding a voice channel.; Despite its limitations, and the complications of supporting it on 4G and later generations of mobile telephony that handle voice traffic in a different way, SMS remains in wide use in 2023. It’s a long-established standard, not proprietary, and it’s cheaper to use than the newer, more capable protocols.
SMT: (1); Surface-Mount Technology. Soldering circuit components to the surface of the board, without the holes and leads required by the old PTH approach. This enables higher circuit density. The SOJ, SOP, PLCC, and PQFP are examples of surface-mount devices (SMDs). See JEDEC for form factors.; (2); Simultaneous Multi-Threading. A software technique for pipelining CPUs that wrings more performance gain out of their multi-threading capability through simultaneous execution of multiple process threads.
SMTP: Simple Mail Transport Protocol. A member of the TCP/IP protocol suite at the Application layer, this is the original e-mail transmission and storage protocol for the Internet, still in use. Mail servers use it to store e-mail addressed to the local domain for later retrieval by a POP or IMAP mail client or a Webmail browser, and to forward non-local e-mail. SMTP uses TCP port 25 or 2525 to connect to the network, or port 465 for encrypted connections.; SMTP is designed to send just 7-bit ASCII text, so the IETF later approved MIME to supplement it.
SNA: Systems Network Architecture. An IBM network protocol.
SNAP: Sub-Network Access Protocol.
Snell’s Law: See TIR.
SNL: Sandia National Laboratories. US Department of Energy research facility in Albuquerque, NM.
SNMP: Simple Network Management Protocol. A member of the TCP/IP protocol suite at the Application layer, this protocol provides tools for managing a network, with four functions for network troubleshooting. The latest version is SNMPv3.
SNR: Signal to Noise Ratio. The standard measure of analog signal quality. For digital, see E_b/N₀.
SO#: See JEDEC.
SOA: Service-Oriented Architecture. A set of design principles for producing distributed (network) applications as a set of standalone services that interact as needed. Compare REST.
SOAP: Simple Object Access Protocol. An XML-based protocol for sharing data between applications using HTTP. A SOAP message consists of a header, a body, and a fault element wrapped in XML tags.
SoC: System on a Chip. An IC chip that implements all the functions of a discrete electronic device. See embedded device. The term is increasingly used to mean single-chip computers, although these devices are necessarily less capable than the microprocessors that run full-featured PCs, and they still require external memory. Other, less dense design approaches include MCM, SiP, and SoP.
socket: An API (package of subroutines) in the OS kernel that allows programs to access TCP/IP network connections. A raw socket is one that allows an application to handle its own IP formatting and unwrapping rather than relying on the kernel. See Winsock.
Socket #: Any of numerous motherboard CPU chip sockets, starting with Intel’s Socket 1 in 1989. In the late 1990s, after Socket 7 and the proprietary Socket 8, Intel switched to proprietary motherboard slots that accommodate CPU chips in single-edge contact (SEC) cartridges – see Slot #. When they returned to the use of sockets with Socket 370, Intel no longer licensed them to rivals. Also, the names gave the number of contacts, which is a good idea.; Most CPU sockets have used PGA (pin grid array) technology, meaning the socket presents a grid of holes and the chip has a set of matching pins on its underside. Around 2006, Intel began favoring the LGA (land grid array) – the fragile pins are part of the socket, making them the problem of the motherboard manufacturer. About 10 years later, AMD followed suit. A few sockets are compatible with processors made for other, very similar sockets, but this is not common and not always seamless.; Intel-licensed:; Socket 1 – 17×17 rectangular 169-pin PGA, 5V, for 80486 processors.; Socket 2 – 19×19 rectangular 238-pin PGA, 5V, for 80486 and Pentium Overdrive processors.; Socket 3 – 19×19 rectangular 237-pin PGA, 5V/3.3V, for 80486, Overdrive, and 5x86 processors.; Socket 4 – (1993) 21×21 rectangular 273-pin PGA, 5V, for Intel’s 60 and 66 MHz Pentiums.; Socket 5 – 37×37 diagonal 320-pin PGA, 3.3V, for Intel’s 75-133 MHz Pentiums.; Socket 6 – For 80486 chips, never much used.; Socket 7 – 37×37 diagonal 321-pin diagonal PGA, 2.5V-3.3V, for Intel’s 75-300 MHz Pentiums, AMD’s K5 and K6, and Cyrix’s 6x86 and 6x86MX. Industry standard, zero insertion force (ZIF) socket. It supports a 64-bit, 66.6 or 100 MHz I/O bus.; Intel only:; Socket 8 – Proprietary 387-pin PGA, 3.1V/3.3V, for the Pentium Pro. It requires a six-layer motherboard.; Socket 370 – For Pentium III and Celeron, in two PGA variations.; Socket 423 – Used from November 2000 to August 2001 for the Pentium 4 Willamette.; Socket 478 – Rectangular PGA, for Intel Pentium 4 Northwood, Celeron, and Xeon (P4 Extreme Addition).; Socket 603 – For Intel Xeon.; Socket 604 – For Intel Xeon.; LGA 775 (Socket T, Socket 775) – (2004) For desktop Pentium, Celeron, and Core 2 processors. With an adapter, it supports some LGA 771 processors.; Socket M – (2006) 479 pins but incompatible with Socket 479. For Core Solo/Duo, Core 2 Duo.; LGA 771 – (2006) Primarily for Xeon workstation and server CPUs.; Socket P – (May 2007) 479 pins, replacing Socket 479 and Socket M for Core 2.; LGA 1366 (Socket B) – (November 2008) 1366 pins, for Core i7 and Core i9, requires DDR3 memory. The big increase in pin count supports the chip’s on-die memory controller.; LGA 1156 (Socket H) – (2009) For Core i7, Core i5, and Core i3. Supports the integrated graphics of i5 and i3.; µPGA-989 – (2009) For laptops.; LGA 1155 – (2011) For chips based on Sandy Bridge, a revision of the 32nm Core architecture with fully integrated graphics, and for later 22nm Ivy Bridge processors. Not compatible with LGA 1156.; LGA 2011 (Socket R) – (November 2011) 2011 pins, for high-end Sandy Bridge and Ivy Bridge processors. Meant to replace the LGA 1366 in the market.; LGA 1150 (Socket H3) – (2013) 1150 pins, for Haswell and planned Broadwell processors. It has five fewer pins than LGA 1155 because Intel moved the voltage regulator from the motherboard onto the processor die.; LGA 1151 – (2015/2016) For the Skylake processor family. Supports DDR4 SDRAM for main memory as well as DDR3.; LGA 2066 – (2017) For Skylake, Kaby Lake, and Cascade Lake processors (Xeon and high-end Core i5 through Core i9), replacing the LGA-2011 in the server/workstation market. It supports DDR4 DRAM.; LGA 1200 – (2020) For 10th- and 11th-generation (Comet Lake & Rocket Lake series) Pentium, Celeron, and Core i3 through Core i9 desktop CPUs, replacing the LGA 1151. It supports DDR4 DRAM.; LGA 1700 – (2021) For 12th- and 13th-generation (Alder Lake & Raptor Lake series) Pentium, Celeron, and Core i3 through Core i9 desktop CPUs, replacing the LGA 1200. It supports DDR4 and DDR5 DRAM.; LGA 1851 – (expected 2023) For the chiplet-based, 14th-generation Meteor Lake series, replacing the LGA 1700.; AMD only:; Super Socket 7 – For the K6 (2 and III).; Socket 462 (PGA-462, Socket A) – For Sempron, Athlon, Athlon XP, and Duron.; Socket 563 – For Athlon and Duron.; Socket 754 – For Sempron, Opteron, and Athlon 64.; Socket 939 – (2004) For Opteron, Athlon 64, Athlon 64 FX, and dual-core Athlon 64 X2.; Socket 940 – (2004) For Opteron and Athlon 64 FX. Supports DDR (DDR1) memory.; Socket AM2 – (2006) A 31 × 31 PGA with 21 holes absent, so it accepts a 940-pin CPU – but is incompatible with Socket 940 CPUs. It’s for the single-core Sempron, Opteron, Athlon 64, and Athlon 64 FX, and the dual-core Athlon 64 X2. It requires DDR2 memory. The physically identical Socket AM2+, for the multi-core Phenom and Phenom II series, can host an AM2 processor if the motherboard supports that. AM2+ and some AM2 sockets can also run AM3 processors.; Socket F – (August 2006) 1207-pin LGA, for some high-end versions of Athlon 64 FX and Opteron quad-core. Supports DDR2 memory.; Socket S1 – 638 pins, for notebooks, with Turion 64 and Athlon 64 Mobile.; Socket AM3 – (2009) A 31 × 31 PGA grid with 941 holes open, for an AM3 CPU with 938 pins. It requires DDR3 memory, and supports some multi-core Phenom II & Athlon II CPUs and single-core Semprons. It does not accept AM2 or AM2+ processors; motherboards billed as AM2/AM2+/AM3 are really AM2+.; Socket G34 – (2010) 1974-contact LGA, rectangular, supporting up to four channels of DDR3 memory for up to quad-core CPU with 1944 pins.; Socket FM1 – (2011) A 31 × 31 PGA grid with 905 holes open. The FM1 CPU has 905 pins. Used for budget 32 nm processors such as the Llano series.; Socket AM3+ – (2011-2012) A 31 × 31 PGA grid with 942 holes open, for Bulldozer core processors, which have 940 pins. The AM3+ socket still runs AM3 CPUs, but AM3+ CPUs aren’t likely to work in AM3 sockets.; FM2 – (2012) A 31 × 31 PGA grid with 904 holes open, matching 904 pins on the FM2 CPU. Not compatible with FM1 or FM2+ APUs. Used for budget APUs starting with the 32 nm Trinity Fusion series, and including the later Richland series.; FM2+ – (2013) A 31 × 31 PGA grid with 906 holes open, for AMD’s Kaveri series APUs with 28 nm Steamroller core. Accepts FM2 APUs.; AM4 – (2016) A 39 × 39 PGA grid with 1331 holes open, for AMD processors using the Zen, Excavator, and Ryzen architectures. Also called PGA 1331. It supports DDR4 SDRAM, PCIe 3.0 and later 4.0, and USB 3.1.; sTR4 – (2017) The first LGA socket from AMD for desktop CPUs, namely the heavy-duty Threadripper models of the Zen microarchitecture’s Ryzen family. It’s... big.; AM5 – (2021) Also called LGA-1718, a ZIF socket with 1718 contacts, continuing AMD’s embrace of the LGA approach that began with the sTR4. Released for the AMD Ryzen 7000 Series with its Zen 4 architecture, it supports DDR5 SDRAM and PCIe 5.0.
SO-DIMM: Small Outline Dual In-line Memory Module. A DRAM circuit card with 72, 144, 200, 204, 260, or 262 flat contacts in paired sets on both sides of the bottom edge, derived for notebook PCs from the larger DIMM. Like the DIMM, the SO-DIMM has unbuffered ECC (usually denoted SO-UDIMM or SO-CDIMM) and registered (SO-RDIMM) versions for special applications, and CSODIMM for improved stability at the high signaling frequencies used for DDR5.; 72-contact – Measures 2.375" × 1" (height varies), and has 36 contacts on each side, with no separating notch. Used for 32-bit FPM DRAM or EDO DRAM.; 100-contact – For printers and other devices rather than PCs. It measures 2.625" × 1", and has two notches dividing its contacts into groups of 12, 32, and 56.; 144-contact – 2.625" × 1" (height varies), with a notch separating its contacts into groups of 60 and 84. Used for 64-bit FPM DRAM, EDO DRAM, or SDRAM (PC66, PC100, PC133).; 200-contact – 2.625" × 1" (height varies). The notch separates the contacts into groups of 60 and 140 (DDR SDRAM) or 40 and 160 (DDR2 SDRAM).; 204-contact – 2.625" long, for DDR3 SDRAM. The notch separates the contacts into groups of 72 and 132.; 260-contact – (2015) For DDR4 SDRAM. The notch separates the contacts into groups of 144 and 116.; 262-contact – (2020) For DDR5 SDRAM. The notch separates the contacts into groups of 136 and 126.
SOE: Soap-Opera Effect. Also called motion smoothing or post-processed motion interpolation. A digital video technology that creates and inserts additional motion-capture frames to compensate for the display having a higher frame rate than the source. Although it’s meant to reduce blur and make moving images look smoother and more realistic, many viewers find the effect off-putting for movies and TV shows.
SOFC: Solid Oxide Fuel Cell. See fuel cell.
SOHO: Small Office/Home Office. Refers to a LAN supporting a small business.
SOI: Silicon-On-Insulator. A chip technology that offers higher speed and lower power than CMOS, but also higher cost and more defects. The wafers consist of thick layers of an insulating substrate such as SiO₂ (quartz) or Al₂O₃ (sapphire) alternating with thin layers of single-crystal silicon containing the circuitry.; In thick-film or partially depleted SOI (PD-SOI), the silicon layer is 1 µm or more thick, so that the depletion region around each gate doesn’t extend into the insulating substrate. Thin-layer or fully depleted SOI (FD-SOI), with lower voltages and greater speeds, uses silicon layers typically less than 0.25 µm, and oxide layers less than 0.5 µm. Strained SOI (sSOI) uses silicon germanium (SiGe) as the insulating substrate; its larger lattice strains the crystalline structure of the overlying silicon, improving charge mobility, hence performance and power consumption.
SOIC: Small Outline Integrated Circuit. See JEDEC.
SOJ: Small Outline J-lead. See JEDEC.
Solaris: The Sun workstation OS, a version of Unix. Called SunOS before 1992.
solar power: There are approaches to solar power that are not photovoltaic, but they fall outside the EE domain.
solenoid: An inductor coil of insulated or enameled wire around a ferrite core. The core is free to move in response to the magnetic field created by current through the coil, so it can trip a mechanical circuit breaker, valve, or other device.
soliton: A class of RZ (return-to-zero) optical pulses with a hyperbolic secant shape, for which dispersion of wavelengths longer than the zero-dispersion wavelength cancels and is canceled by self-phase modulation, so that the pulse will travel great distances in optical fiber without spreading.
SoM: System On a Module. A less extreme approach than the SoC, using a small circuit board rather than a single chip.
SOMI: Slave Output/Master Input. See SPI.
SONET: Synchronous Optical Network. A standard for optical telecommunications networks, created in 1986 by ECSA on behalf of ANSI for use in North America only. It’s a subset of SDH, which was under development at the same time. SONET has four optical interface layers: the path layer, the line layer, the section layer, and the photonic layer. Its fundamental carrier standards are the STS-1 (51.84 Mb/s) and OC-1 (the optical counterpart to STS-1). SDH has equivalent carrier standards, starting with STM-1, which is equal to an STS-3.
SoP: System-On-Package. A newer device design approach that embeds components and traces in parallel, interconnected layers of dielectric rather than placing them in individual packaging on a circuit board. This has the potential to drastically reduce device size relative to a MCM or SiP. See also SoC. Circuits with differing requirements (e.g., optical or RF rather than electronic) get their own layers, which can use differing dielectrics.
SOP: Small Outline Package. See JEDEC.
SOT-##: Small Outline Transistor. See JEDEC.
sound: The range of sound frequencies that most people can hear is roughly 30 to 20,000 Hz. Most mammals have a wider hearing range; dogs hear frequencies out to about 45,000 Hz. Ultrasonic devices can generate sound frequencies into the low GHz range.; Multi-channel (polyphonic) sound systems are denoted by two numbers separated by a period: 2.0, 3.0, 2.1, 3.1, 4.1, etc. This indicates the number of full channels and the presence (.1) or absence (.0) of a LFE (low frequency effects) channel for powerful low-frequency sound. The LFE drives the sub-woofer if one is present. However, it can go directly to speakers, and other channels can also feed to the sub-woofer, so it’s not accurate to call the LFE the sub-woofer channel.; Professional sound systems can go to 22.2 or beyond. Even modern PCs commonly have 6 or 8 channels (5.1 or 7.1) integrated into the motherboard. However, a motherboard sound port has 3 to 5 watts of power, which won’t sound like much on passive speakers. Taking full advantage of such a system requires independently powered speakers and sub-woofer. Some standard sound configurations:; monaural sound – Aka monophonic or high-fidelity. One channel. Eclipsed by stereo in the 1950s.; stereo (stereophonic) sound – Two full channels, hence, 2.0 or 2.1. See AUX.; surround sound – 5.1 or better, with two of the speakers behind the listener. Virtual surround sound (VSS) attempts to simulate this effect with a 2.1 configuration.; Under the widely used 1999 version of the Intel/Microsoft PC System Design Guide, the 3.5mm TRS analog sound ports on PCs are color-coded as shown below. Less powerful speaker/subwoofer sets use just the lime-green port, rather than connecting the subwoofer separately to the orange port.; Common digital audio formats: AAC, ALAC, FLAC, MIDI, MP3, WAV, WMA.
source coding: Converting analog source data (text, images, sound, etc.) to digital form. This assumes some efficiency, such as compression; otherwise, it’s just formatting. Not to be confused with source code, which refers to files written in a programming language and used to create a program. See data code.
southbridge: Named for being below (south of) the northbridge component in circuit diagrams. Introduced in 1991 with the Intel Hub Architecture, this part of the PC motherboard chipset controls I/O and system buses, for which reason it’s also called the I/O controller hub (ICH). It’s subordinate to the northbridge, to which it links via the PCI bus.; The Platform Controller Hub (PCH) has replaced the Hub Architecture.
spam: Named for a certain Monty Python’s Flying Circus sketch, in which the eponymous canned meat is way too much of something no one wants any of, spam is unsolicited, mass-distribution messaging. In the early years of the Internet, it was exclusively e-mail, and was usually just selling something or otherwise trying to get your money directly. Modern spam is almost all sent via botnets, and includes social media postings as well as e-mail. E-mail spam is still mostly after your money, but now it usually carries malware, or links to a phishing site or compromised site that pushes malware. Social-media spam is about trolling, including for political purposes.; More recently, spamming has become slang for indiscriminately sending or outputting almost anything.; SPF is a technical approach to blocking e-mail spam.
S-parameters: Scattering parameters. Given some two-port device, such as a filter, a signal input to one port results in some of the signal being reflected, and some being transmitted out the other port. In matrix algebra terms, two output signal vectors (one from each port) at a given frequency can be represented as two input signal vectors multiplied by a 2×2 array of four parameters. The members of this 2×2 array (the scattering matrix) are the S-parameters. Two are reflection coefficients, and two are transmission coefficients. They characterize the signal behavior of the device across the tested frequency range. A swept-frequency network analyzer is the usual measurement tool.
SPARC: Scalable Processor Architecture. A family of Sun Microsystems RISC microprocessors, the basis for a successful line of minicomputers in the late 1980s and 1990s that usually ran SunOS or Solaris OS. Includes MicroSPARC I, MicroSPARC II, SuperSPARC, and UltraSPARC, and SPARC64. Still around but losing ground as of 2009.
S/PDIF: Sony/Philips Digital Interconnect Format (or Digital Interface). A digital sound I/O technology, using either 75Ω coaxial cable (usually with BNC or RCA connectors) or fiber-optic cable (with TOSLINK connectors). It carries two channels of 20-bit PCM audio sampled at 48, 44.1, or 32 kHz, or multi-channel compressed audio using the same set of bitrates. It implements anti-copying features like its audio-video cousin HDMI. Some PC motherboards have a 3-pin header to provide an internal S/PDIF connection.
SPDT: Single Pole, Double Throw. See switch.
spectrum analyzer: Spec-an for short. An instrument that displays and measures input amplitude across a range of frequencies. Contrast the oscilloscope, which measures amplitude (voltage) as a function of time rather than of frequency.
SPF: Sender Policy Framework. A 2006 IETF standard intended to stop most E-mail spoofing. The host of a SPF-compliant Internet domain configures its DNS record to specify which servers are allowed to send E-mail from that domain. Receiving nodes compare the originating server given by a message’s IP packets to the list of permitted servers. See http://www.open-spf.org/ for more.
SPI: (1); Serial Peripheral Interface. A full-duplex, synchronous, serial, master-slave data bus protocol created by Motorola in the 1980s. It’s used to link the processor with other components in embedded systems, commonly including the BIOS. It’s faster than the similar I²C, up to several Mb/s. There’s a nearly identical subset of SPI called Microwire. Compare also CAN, LIN.; SPI has four signals: a serial clock (SCLK, from the master), a master output/slave input (MOSI or SIMO), a master input/slave output (MISO or SOMI), and a slave select (SS) or chip select (CS) that goes low to initiate a command exchange. Data is written to the output register in blocks of some preconfigured length, often but not always 8 bits depending on the hardware implementation of the standard, and the SPI logic clocks it out MSb-first on the MOSI or MISO line.; Some implementations call the data lines SDI and SDO (serial data in and serial data out) or SI and SO (serial in and serial out). If more than one slave exists, the master needs a separate control line for each SS.; (2); SCSI Parallel Interface. See SCSI.
SPICE: Simulation Program with Integrated Circuit Emphasis. The granddaddy of all analog circuit simulator programs, created at UCal/Berkeley in the early 1970s. A commercial version, PSpice, is distributed by the makers of OrCAD. SPICE, the newer SPICE2, and PSpice all simulate circuits at the transistor level, which makes them impractical for dealing with large designs.
spin: A fundamental but poorly understood property of subatomic particles. It has magnitude and direction, but is strictly a quantum-mechanical property, with no classical equivalent. An electron, for example, acts like a magnet, with north and south poles. If its spin property is up, it will orient itself with an external magnetic field; spin down means it will align to oppose the field.; It’s common to say that a subatomic particle can have superposed spin states, meaning up and down simultaneously. This is a confusing description of quantum-mechanical probability at work. In quantum physics, the spin of a particle does not exist until it’s measured. In fact, the particle is not even a particle until it’s measured. Spin state is a wave function that uses the complex-valued quantum amplitudes α and β. The squares of the quantum amplitudes are the probability of the electron being spin up or spin down when measured. Since it must be one or the other, α² + β² = 1.0.; The spin state of individual particles is unstable, lasting only nanoseconds even if not disturbed. Just as a magnet can align electrons, a spin-polarized current can change or flip the direction of a magnet it passes through.; Fermions (e.g. electrons) have spin values of ½, 3/2, 5/2, etc. Bosons (e.g. photons) have spin values of 0, 1, 2, etc.; Spintronics is the name for a new and still (2007) immature field of electronics based on the exploitation of spin, mainly using electrons in quantum dots (single-electron traps). It promises low-power devices that are smaller and much faster. The big hurdle is developing materials that are sufficiently cheap and reliable, and that work at room temperature. It has one modest commercial success, MRAM. Exploiting spin states for quantum computing (see qubit) is proving even more difficult.
SPM: Scanning Probe Microscope (or Microscopy). An instrument (or technique) for atomic-level imaging of surfaces. AFM and STM are types of SPM.
spoofing: Falsifying the source of an online message. E-mail spoofing, a standard spammer trick, puts a false address in the “From:” field and often in the SMTP message header as well. The IP packets carrying the mail normally retain the IP address of the true originator, unless the spammer also manages the much trickier IP spoofing.; Mail servers don’t normally block spoofed messages because there are legitimate reasons to change the “From” address. However, see SPF.
SPR: Special Purpose Register. Same thing as SFR.
SPS: (1); Standard Positioning Service. See GPS.; (2); Samples Per Second. The rate at which a device can digitally sample an analog input. Can be given in kiloSPS (KSPS), megaSPS (MSPS), even gigaSPS (GSPS).
SPST: Single Pole, Single Throw. See switch.
SPX: Sequenced Packet Exchange. See IPX/SPX.
spyware: See malware.
SQL: Structured Query Language, unofficially pronounced “sequel”. An originally IBM (1974) programming language for talking to databases (e.g. DB2, Oracle, Sybase, Ingress, Access), blessed as a standard by ANSI and ISO, and still very popular. It was designed to implement the relational database model. In applications, SQL is normally used in conjunction with a full-featured programming language.; A Web site with access to a searchable database can fall victim to a SQL injection attack, which exploits user data-input fields to trick the site into running an unauthorized SQL command. For a relatively accessible example of this attack, see https://www.xkcd.com/327.
SQPSK: Staggered Quadrature Phase Shift Keying. Another name for OQPSK.
sr: Square Radian. See steradian.
SR: Software Radio. See SDR.
SRAM: Static Random Access Memory. A type of volatile RAM that uses tiny transistors (6 per bit) as on/off switches, so that, unlike DRAM, it maintains state without refreshes as long as the system has power. This makes it faster and more power-efficient than DRAM, but also bulkier and more expensive, so its main computing application is for small, fast RAM caches. When supplied with a long-life battery to hold charge, it can serve as non-volatile RAM.; Video RAM (VRAM) is sometimes SRAM. Some DRAM technologies are used for SRAM also, such as QDR and DDR.; Decreasing SRAM cell size has made it vulnerable to soft errors caused by ionizing radiation. The traditional fix is to add error-correction cells to each row. A newer approach is to add capacitance to each cell, which has minimal effect on cost and size.
SRB: Solid Rocket Booster. A rocket that burns solid fuel, and is used to propel a multi-stage boost vehicle. It gets discarded once its fuel is used up.
SRT: Smart Response Technology. Intel firmware introduced with its Z68 motherboard chipset for the LGA 1155 socket. It boosts the performance of a PC’s mechanical hard drive by using the SSD (if the PC has one) as a hard drive cache.
SRU: Shop-Replaceable Unit. A hardware module that must be replaced by the manufacturer if it malfunctions or needs an upgrade, as opposed to LRU.
SS: Spread Spectrum. An RF signal that’s spread over a wide spectral range, making it more resistant to interference or deliberate jamming, and giving it a low probability of detection (LPD) and interception (LPI). Either the signal data is XORed with a higher-speed pseudo-random sequence (the spreading code) before modulation, or the carrier frequency-hops as dictated by a pseudo-random sequence (the chipping code). The pseudo-random sequence can be transmitted separately along with the coded signal (transmitted reference, TR), or just stored at both ends (stored reference, SR). SR is preferred because it’s more secure. See DSSS, FHSS.
SS7: Signaling System 7. A packet-switched, out-of-band, common-channel system defined by CCITT as part of the OSI layered network model, used by Ma Bell for telephone trunk signaling. It typically has a 64 kb/s rate. There is a universal communication stream between switches. SS7 has become a standard for mobile phone systems worldwide, but is outdated and insecure.
SSA: Serial Storage Architecture. Obsolete 1990 IBM data bus standard, running at 40 MB/s in each direction.
SSB: Single Sideband. Amplitude-modulating a carrier wave produces a signal in the frequency band above the carrier (upper sideband, or USB), and a mirror-image signal below it (lower sideband, or LSB). This is a dual sideband (DSB) signal. The USB and LSB carry the same information, so it’s a waste of power and bandwidth to transmit both. The carrier contains no information at all (although that’s not apparent in a time-domain view), so it isn’t needed either. Filtering out the carrier and one sideband and sending just the other sideband produces the more efficient SSB signal that FDM and CB radio use. It requires the receiver to generate its own copy of the carrier to demodulate the transmission, but that’s easy to do.
SSD: Solid State Drive. Also called a flash drive. A data-storage drive using non-volatile, solid-state memory technology – like the removable thumb drive, but bigger, and capable of higher read/write speeds. Compared to the standard HDD with its spinning media and magnetic read heads, the SSD has faster random access, uses much less power, and is potentially more durable, but costs more per gigabyte and has lower storage density.; SSDs are often paired with conventional hard drives to create a two-drive PC, with the operating system and certain other programs on the much faster SSD. The HDD must be formatted separately, and it’s a good idea to move default file folders to it. (In Windows 8, do the first with Disk Management, opened from the Control Panel’s Computer Management; and do the second with the Windows System Assessment Tool, run by entering WinSAT.exe formal at the command prompt.); The current (2013) basis for SSDs is NAND-gate flash memory. There are three sub-types: SLC (single-level cell), MLC (multi-level cell), and TLC (three-level cell), in order of decreasing cost, speed, and longevity, and increasing density and power consumption. NAND-gate MLC is most common for SSDs, with budget models using TLC.; Initially, most SSDs were built for the SATA bus, which couldn’t keep up with their speed gains. Alternatives include mSATA (mini-SATA), the slower USB (mainly for external drives), and the speedier PCIe for internal drives mounting in motherboard expansion slots. All of these can use the newer M.2 form factor instead of a bus-specific slot or port.; The hybrid drive, or solid state hard drive (SSHD), is a magnetic-disk drive with a large (several GB) solid-state cache and integrated cache manager. For computers with consistent patterns of use, it can achieve access times close to those of a SSD with much higher capacity and/or lower cost.
SSE: Streaming SIMD Extensions. Intel’s set of 70 new instructions for the Pentium III, circa 1999. Some are extensions to MMX (which is itself an extension), while others are floating-point instructions. SSE2 is for Pentium 4.
SSH: Secure Shell. A member of the TCP/IP protocol suite at the Application layer, this is a network protocol for opening a command shell and issuing commands on remote network nodes. The default TCP port for the SSH daemon (sshd) is 22. Some servers use a different port, to reduce the frequency of brute-force password-guessing attacks.; SSH-2 (1996) is more secure than the original SSH-1 (1995). It has largely replaced Telnet, which doesn’t have encryption.
SSHD: Solid State Hard Drive. See SSD.
SSI: Server Side Include.
SSID: Service Set Identifier. A character string that acts as the name of a wireless network. Network packets carry the string to identify themselves, unless SSID broadcast is turned off as a security measure.
SSK: Slope Shift Keying.
SSL: Secure Sockets Layer. An encryption protocol for Web sites introduced by Netscape in 1993, made an IETF-approved WWW standard, and now considered insecure. The newest version is 3.0 (1996). EVSSL (Extended Validation SSL) is a variant with additional security features. Most sites offering encryption instead use TLS, which operates in the same way as SSL.
SSM: Sign/Status Matrix. See ARINC-429.
SSOP: Shrink Small Outline Package. See JEDEC.
SSPA: Solid State Power Amplifier.
SSRAM: Synchronous Static Random Access Memory.
SSTL: Stub Series Terminated Logic. See logic family.
ST: See fiber.
stack: A block of memory that the OS reserves for an execution thread to store variables & data and perform calculations while it runs. When a thread ends, the OS releases its stack to the system memory pool.; Because its size is set when the code is compiled, unlike the dynamically allocated heap, the stack is a simple, rapidly accessed LIFO structure. This advantage comes with the danger of stack overflow – a thread requiring more memory than the stack has, crashing the program.; Modern computer architectures are either stack-based or register-based, although each type makes use of both structures. The program maintains a register for each thread called the stack pointer (SP), which contains the memory address of the top of that thread’s stack. To manipulate variables, the program pushes them onto and pops them off the top of the stack. For example, an add command would push two numbers onto the stack, pop them off again to add them, and finally push the sum onto the stack. By convention, a stack starts at the highest address in its assigned memory range, and expands into the lower addresses as objects are pushed onto it. The value of the stack pointer therefore decreases as stack use increases.; A register-based architecture transfers values between registers. Since it doesn’t need to adjust stack pointers all the time, it can be faster. On the other hand, the code has to constantly specify which registers to use, slowing it down.
stainless steel: An alloy of steel, at least 10.5% chromium, and one or more other metals, commonly nickel, molybdenum, titanium, or aluminum. The surface forms a metal-oxide film that resists wear and corrosion. Forcing stainless steel surfaces together in a way that scrapes off this film causes the two surfaces to spontaneously cold-weld together.; More than 70% of all stainless steel manufactured is austenitic (chrome-nickel) steel, which contains no more than 0.15% carbon, and is not ferromagnetic due to the influence of nickel on formation of the crystal lattice – even though nickel is itself ferromagnetic. The most common grade of austenitic steel, 18-8, is 18% chromium and 8% nickel by weight. About 25% of all manufacture is ferritic steel, which contains molybdenum (usually), up to 1% carbon, and little or no nickel, and is ferromagnetic.
standard deviation: A statistical measure of the variability of a data set or function. A low standard deviation indicates the data points don’t vary much. It’s based on data with a normal probability distribution, better known as a bell curve, and is less meaningful for data sets with other probability distributions. To find the standard deviation of a set of n values:; First find the mean (average) value, which is the sum of all the values divided by n.; Then calculate the variance – subtract the mean from each of the n values, square each result, sum all of the squares, and divide the total by n-1. (If the data set is the entire population being studied rather than a sample, the divisor is n rather than n-1, but this is a rare case, and biases the result unless n is large.); Finally, to get the standard deviation, take the square root of the variance.; For a normal distribution, about 68% of all values fall within one standard variation about the mean, above or below. Two standard deviations about the mean covers roughly 95% of the data, and three includes 99.73% of cases. Six sigma, a buzzphrase for talking about reliability, means six standard deviations from the mean, and includes 99.99999988% of the data. The “sigma” shows up because the Greek lower-case sigma (σ) is commonly used to represent standard deviation.
STAPL: Standard Test And Programming Language.
state machine: A system with memory of the past that constrains its future. A computer can be viewed as a very complex set of states, with each program being a state machine. State machines are used as system design models to describe specific device or program interactions. Each consists of 1) an initial state or record of something, 2) a set of possible input events, and 3) a set of new states that can result from the combination of previous states and inputs.; A finite state machine (FSM) is one that has a limited or finite number of possible states. An infinite state machine can be conceived, but isn’t practical. See Turing machine.
static electricity: The opposing charge that can develop between certain non-conducting materials in contact, as valence electrons migrate from one material to the other. The static voltage potential between common objects can be surprisingly high, in the thousands or tens of thousands of volts, making it a real threat to electronics (see ESD). The non-conducting nature of the charged materials means the current they can discharge is very seldom large enough to harm living things. There are dangerous exceptions, such as the charge developed by aircraft in flight – and lightning, which is essentially static discharge on a huge scale.; Whether a material gives up electrons or attracts them depends on the other material. Rubber and plastic are almost always attractors, developing a negative charge. Glass, skin, and hair tend to be electron donors, hence positive. Paper and most fabrics are ranged in the middle.
STD: Simple To Design. A simple, rugged, 8-bit data bus & expansion card standard also known as IEEE 961, introduced in 1978 for industrial applications. The cards are 4.5" × 6.5", and have a slotted 56-contact PC edge connector. STD32 is a backward-compatible, 16- and 32-bit expansion with a 136-pin connector. The original 8-bit STD is sometimes called STD-80.
STE: (1); Secure Terminal Equipment. The US government’s 1990s successor to the STU, with which it’s compatible. It was designed to connect to ISDN as well as POTS, and handles data rates up to 128 kb/s. A removable PC Card holds its encryption key and algorithm.; (2); Specialized Test Equipment.
steganography: Traditionally, concealing digital data in a file representing some analog source, typically a sound or image file, by substituting the data for some of the least significant bits of the file. The change in sound or image quality is usually negligible. Compare hyperencryption.; The technique can also be used to embed malicious behavior in a neural network.
STEM: Science, Technology, Engineering, & Math.
steradian: Also called a square radian, abbreviated sr. A measure of the angular area about the origin of a sphere subtended by a two-dimensional circle on the surface of that sphere. Just as a circle of unit radius 1 subtends 2π radians about the origin, a sphere of unit radius 1 subtends 4π steradians about the origin, since the surface area of a sphere is 4πr². Steradians are used in physics and antenna modeling when flux through a 3-D surface is involved. See fluence, isotropic.
stereo: See sound.
stigmergy: Complex behavior emerging from the actions of large numbers of units following simple rules. This sort of behavior, also known as an emergent property, comes from the world of social insects, but has powerful AI applications.
stingray: A trademark of Harris Corp., the first company to market the technology. Used generically, “stingray” – a.k.a. IMSI catcher, pen register/trap-and-trace device, or cell site simulator – means a device that can function as a cell tower for 2G mobile phones but is run by someone other than the mobile wireless provider. In the US, stingrays are widely used by law enforcement, normally but not necessarily with the wireless provider’s co-operation. Acting as connection points for signals from the mobile phones of nearby suspects (and anyone else in the vicinity), they’re able to record conversations, texts, and other data. With time and triangulation, they can also determine where suspects are. Compare dirtbox.; The large ones with greater range are commonly used from vehicles or aircraft. There are smaller, easily portable versions that run on batteries but have shorter range, and are more appealing to spies and criminals.; 2G connections are easier to hijack because they don’t require end-user verification. Where the network is 3G or 4G, stingray operators have been known to jam those bands, forcing subscriber handsets to fall back to 2G signals. A follow-on system called Hailstorm is designed for 3G and 4G.; There are mobile phones with their own encryption specifically to defeat this kind of man-in-the-middle surveillance, although they can’t conceal the user’s location. There are also mobile phone apps that claim the ability to detect stingrays, but as of 2017 they’re relatively easy for stingray operators to defeat.
STL: Standard Template Library. A set of generic C++ class templates and algorithms that supports common data structures. See ATL, MFC, .NET, WTL.
STM: Scanning Tunneling Microscope. Like the SEM, it uses a focused electron beam for imaging, but this beam forms a quantum tunneling current across a vacuum barrier to the sample. It can resolve objects as small as 0.1 nm, so it can show individual atoms. It can also manipulate single atoms and molecules.
STM-#: Synchronous Transport Module-#. SDH’s carrier-multiplexing standard. The fundamental SDH carrier, STM-1, employs 2430-byte TDM frames at 8000 frames/second (155.52 Mb/s). A frame is itself divided into nine 270-byte blocks, each containing 9 bytes of transport overhead followed by 261 bytes of payload. The resulting payload throughput, 150.336 Mb/s, has the capacity for 2016 DS-0 voice-grade channels. Higher-rate carriers (STM-3, STM-4, STM-6, etc.) are integer multiples of the STM-1 capacity.
STP: (1); Shielded Twisted Pair. A transmission line consisting of two insulated wires twisted together and shielded from EMI by a conductive sheath of foil or braided wire. It provides better noise protection than UTP, and is used for noisier environments. Most STP cables have multiple pairs.; The ISO and IEC have created more detailed terminology: F rather than S to indicate foil rather than the more effective and expensive braided shield, and a leading S/ or F/ or U/ to indicate the degree of shielding around all the twisted pairs in a multi-pair cable, as distinct from the shielding of individual pairs. For example, F/UTP means a cable with a foil shield around all of its pairs, and no shielding of individual pairs.; (2); Standard Temperature and Pressure. A temperature of 0° C and pressure of 100 kPa, used as a common starting point for experiments.
strap: Any of several hardware configurations that set a digital value to be read by software or firmware. This can include DIP switches, jumpers on header pins, pull-up/pull-down resistors to a DC voltage or ground, or wires that a user can connect to a DC source.
STS-#: Synchronous Transport Signal-#. SONET’s carrier-multiplexing standard for wire-line signals. The fundamental SONET carrier, STS-1, employs 810-byte TDM frames at 8000 frames/second (51.84 Mb/s). A frame is itself divided into nine 90-byte blocks, each containing 3 bytes of transport overhead followed by 87 bytes of payload. The resulting payload throughput, 50.112 Mb/s, has the capacity for 672 DS-0 voice-grade channels, i.e. a DS-3. Higher-rate carriers (STS-3, STS-9, STS-12, etc.) are integer multiples of the STS-1 capacity, extending at least as high as STS-192 (9.953 Gb/s). OC-# refers to the optical equivalent of an STS-# carrier.
STTL: Schottky Transistor-Transistor Logic. See logic family.
STU: Secure Telephone Unit. A US government telephone with voice and data encryption capability. Each STU is programmed with the 6-digit department/agency/organization (DAO) code of the organization to which it belongs. It also has an access control list (ACL) of other DAO codes, and will support encrypted communications with any STU whose code is on this list. Changing the list requires the STU’s master crypto-ignition key (CIK), which contains an EEPROM chip that can be updated by remote dial-up. The third-generation STU-III was introduced in 1987. See STE.
stub function: An intermediary between a compiled library and software being developed for a specific platform. Rather than having to modify and rebuild the library for each new platform, the developer modifies just the stub, and launders all library calls through it.
Sturgeon’s Law: 90% of everything is crap. A 1958 comment by sci-fi author Theodore Sturgeon, although he was quoted as saying “crud”.
stylometry: Identifying the author of a document by comparing its patterns of word choice, sentence structure, etc. to those of documents by known authors. There are computer algorithms to automate and greatly speed up the process.
Subversion: See version control.
Sun: Originally “Stanford University Network”. Sun Microsystems, a big computer company, part of Oracle Corp. as of 1/2010.
supercap: See ultracap.
super-heterodyning: See heterodyning.
superscalar: Having multiple execution pipelines.
SUS: Single Unix Standard. A set of standards for Unix system interfaces. It includes POSIX.
susceptance: See admittance.
SVC: (1); Switched Virtual Circuit. A temporary path through a packet-switched network. Compare PVC.; (2); Static VAR Compensator. A solid-state device that power grids use to maintain voltage levels by generating or absorbing reactive power (volt-ampereres reactive).
SVG: Scalable Vector Graphics. Graphics with rendering that can be optimized to the display device.
SVGA: Super Video Graphics Array. See graphics.
S-video: Super (or Separated) Video. Any of several mini-DIN connectors used for NTSC, PAL, or SECAM analog video signals. A 4-pin version (2 signal, 2 ground) was common on TVs, VCRs, and DVD players. 7- and 9-pin versions exist.
SVM: Secure Virtual Machine. Early name for AMD-V.
SVN: Subversion. See version control.
SW: Switch. A common abbreviation in diagrams. See switch.
SWAG: Scientific Wild-Ass Guess. A WAG in a lab coat.
swatting: See troll.
SWE: Software Engineering.
SWER: Single Wire Earth Return. The original telegraph and telephone wire. Still used for rural power because it’s cheaper and more reliable than the conventional two- or three-phase power distribution line. It’s also less efficient, and the power quality is often less.
SWIR: Short-Wave Infra-Red. See IR.
switch: (1); As opposed to the more complex, cross-network role of the router, the switch makes simple switching decisions for traffic within a network, operating at layer 2 (data link) of the OSI network model. This means it tracks local nodes by their MAC numbers. However, it learns which MAC addresses correspond with which IP addresses, and routes packets accordingly. Because of these virtual connections a switch establishes between its ports, it avoids the collision delays and wasted packets associated with a hub.; A multi-layer switch can route traffic at higher OSI layers – layer 3, the domain of IP, and sometimes layers 4-7 as well. This makes it very much like a router, except that it can’t be reprogrammed.; (2); A mechanical or electromechanical device that makes and breaks connections between input and output terminals. Common types are listed below. Other, more complex switches exist.; SPST – Single Pole, Single Throw. Makes & breaks a connection between one input and one output.; SPDT – Single Pole, Double Throw. Switches one input between two outputs.; DPST – Double Pole, Single Throw. Essentially, two SPST switches that operate simultaneously. The DPST therefore has two separate input/output pairs.; DPDT – Double Pole, Double Throw. Two SPDT switches that operate simultaneously. The DPDT has two inputs, each of which is switched between two outputs (four possible outputs in all).
SWR: Standing Wave Ratio. See VSWR.
SX: Under Intel, see 80386, 80387, 80486, 80487.
SXGA: See graphics.
SYN: From “synchronization”. The first packet sent by a server setting up a TCP connection. The receiving server should respond with a SYN/ACK packet, or a RST (reset) packet if the TCP/IP port receiving the SYN is closed. DDoS attacks frequently use a barrage of SYN packets from many sources to tie up the target’s ports. Malware uses SYN packets to scan a target’s ports for openings.
synchronous: When referring to data protocols, this means one that requires bits to be transmitted at a fixed rate with a synchronized start time. Compare asynchronous.

SATA Pin #	Function	SATA Power Pin #	Function
1	GND	1, 2, 3	+3.3 V
2	A+	4, 5, 6	GND
3	A-	7, 8, 9	+5 V
4	GND	10	GND
5	B-	11	multi-function
6	B+	12	GND
7	GND	13, 14, 15	+12 V

lime green	stereo output to speakers or headphones
orange	center speaker/subwoofer output
black	surround-sound (5.1 or 7.1) rear speakers stereo output
silver	surround-sound (7.1) side speakers stereo output
pink	microphone input
light blue	stereo input