Acronym Addiction

When you live on the cutting edge of technology, there are, literally, no words to describe it. Instead we have acronyms. Lots and lots of acronyms.

And for good reason. Imagine a discussion of high-density-technique metal-oxide semiconductors, integrated database application programming interfaces, or—let’s go out on a limb here—separate absorption-graded multiplication avalanche photo diodes. Without acronyms, by the time you got halfway through the conversation, the technology in question would be obsolete.

Here we’ve compiled a list of our favorites to help you through your day. It’s not meant to be comprehensive—there are plenty of more thorough sources, including an IEEE dictionary—and it’s certainly not meant to be offensive. But we hope it shows the interesting and sometimes really weird ways new acronyms come about.

Don’t feel bad if many of these terms are new to you. It has gotten to the point where even the luminaries are in the dark, so to speak. We contacted several newly minted IEEE Fellows, whose experiences prove the point.

Sandra Johnson, chief technology officer for IBM’s global small and medium businesses, recalls attending a presentation that was so chock full of esoteric acronyms that she “leaned over to the people next to [her] and asked if they knew what the presenter was talking about, and they didn’t,” she says. Johnson’s question got all the way around the room, but no one was familiar with all of the acronyms the presenter was using. “It was amusing,” she says. “This guy was going to town, and no one knew what he was talking about.”

“Certainly I’ve been lost,” says Charles J. Alpert, technical lead for design tools, at IBM’s Austin (Texas) Research Laboratory. “Especially the first year I was at IBM. I’m embarrassed. I’m new. But I realized I might as well interrupt people when they’d use acronyms I didn’t know and ask.”

That’s Waguih Ishak’s philosophy, too. “I think it’s actually sometimes insulting to assume people know these acronyms,” says the vice president and chief technology officer of chip maker Avago Technologies, in San Jose. He recalls a corporate technology review at HP in 1996. Intrigued by the volume of acronyms he heard during the very first presentations, he began writing them down. Ishak was scheduled as the penultimate speaker, but after the barrage of scholarly papers, he realized everyone was exhausted. So he scrapped his presentation and instead announced: “I’m going straight to the glossary.” He’d compiled seven slides’ worth of acronyms in just two days.

The vast majority of what we commonly call acronyms are really another type of abbreviation: an initialism. Technically an initialism becomes an acronym only if pronounceable as a word—radar (radio detection and ranging) or BASIC (Beginner’s All-purpose Symbolic Instruction Code), for example. On the other hand, people have found ways to pronounce the ostensibly unpronounceable, and thus SCSI (small computer system interface) became “scuzzy.” More recently, some of the folks involved in the wireless metropolitan area network (WMAN) field have publicly wondered if it was a good idea to pronounce that acronym ”woman.”

Engineers, who tend to be adamant about technical standards and specifications, are pretty ambivalent about consistent capitalization. Thus we end up with the indecisive VoIP (voice over Internet Protocol) and QoS (quality of service). To say nothing of MIPS and MOSFET, which, rendered entirely in capital letters, are like little printed screams.

Some acronyms become like talismans—kept and frequently used long after the exact meaning has faded. IBM’s Alpert says he once attended a “Jeopardy”-like game held for a gathering of 300 design engineers. “One category was acronyms. We all recognized them, but nobody knew what they were. We’d used them for so long we’d forgotten what they stood for,” he says.

The classic example is laser. Though countless people use lasers every day, most nontechies have no idea the word is an acronym for light amplification by stimulated emission of radiation. Laser’s place in language has so evolved that it has even spawned a verb: “to lase.”

Some lexicographic wit coined a term for what’s happened to laser, radar, and their ilk: they’ve become anacronyms, a neologism that smooshes the sounds (and the meanings) of acronym and anachronism. The product of smooshing two words together, by the way, is a portmanteau.

When an acronym becomes an anacronym, funny things can happen to it. For one, people sometimes start saying the acronym coupled with the verbalization of one of its constituent elements. Hence in “SCSI interface,” the word “interface” is completely redundant, because that’s what the “I” is for.

Another linguistic mind-bender is the creation of matryoshka acronyms—acronyms that, like Russian nesting dolls, when opened are found to contain other abbreviations inside. Our favorite: ABT, Advanced BiCMOS Technology, with the acronym BiCMOS right in the middle. True, BiCMOS isn’t exactly an acronym; it’s more of a portmanteau of bipolar and CMOS (complementary metal oxide semiconductor), but you get the idea.

Using the 26-letter English alphabet, the number of possible three-letter acronyms is 17 576. And yet the potential afforded by this sizable number is apparently insufficient for engineers and technology hawkers who can’t seem to avoid reusing abbreviations.

Take, for example, ATM. It is asynchronous transfer mode, automated teller machine, and Adobe Type Manager. What we love about ATM is that even engineers don’t know which ATM anyone is talking about without contextual clues. It is the same with DLL (dynamic link library and delay-locked loop) and SPI (SCSI parallel interface, serial peripheral interface, stateful packet inspection, and system packet interface). And then, of course, there’s PC: printed circuit, personal computer, program counter, and, oddly enough, “carrier power” of a radio transmitter.

Though we can’t help you tell whether someone is talking about CMOS or sea moss, we think we can broaden your knowledge of the industry lingo. So for your delectation, we’ve compiled a banquet of some of our favorite recent electronics acronyms and initialisms. Find your favorites or see how many you know. If you’ve had an amusing encounter with an abbreviation or know the odd secret history of an acronym, drop us a line.

The Glossary

ABT: Advanced BiCMOS technology. Building BiCMOS chips, which combine bipolar transistors and field-effect transistors, started out as a fairly complicated process; apparently it’s become even more so.

AJAX: Asynchronous JavaScript and XML. A programming approach combining the constituent technologies in the acronym, intended to make Web pages feel more responsive. It lets a server load new content onto an open Web page without the browser’s having to reload the whole page.

BEOL: Back end of line. The BEOL and end-all of acronyms. (It’s a pun. No, it’s not really funny. Puns never are.) Refers to latter-stage processes in IC production, including interconnecting the devices on the chip. It’s the opposite of front end of line (FEOL), where sexier stuff, like making transistors, happens.

CSP: Chip-scale package. A sure sign that some people in the electronics industry are underemployed is that they keep inventing new abbreviations for things that already have a perfectly functional one. Case in point: CSP is for all intents and purposes just BGA (ball-grid array). That’s where you place a chip onto a substrate of the same size that has an array of solder balls beneath it. Then you put that assemblage onto a printed circuit board and melt the balls to link the chip to the circuit board’s wiring. W-CSP (wafer-level CSP) is actually something new. In W-CSP you build the package with its solder balls onto the chip even before it’s been cut out of the wafer.

DFM: Design for manufacturability. A design approach that recognizes the seemingly obvious idea that if it’s hard to build, it’s not a good design. In IC manufacturing, DFM takes into account things like whether your newly designed circuit is in danger of being accidentally scrubbed away during important wafer-smoothing steps.

DMD: Digital micromirror device. An array of tiny movable mirrors on a chip. Developers at Texas Instruments (TI) originally thought DMDs might usher in an age of highfalutin optical computing, but instead DMDs are found mostly in rear-projection TVs (RPTVs) showing low-brow programming and in projectors showing boring PowerPoint presentations. In both cases a DMD steers light onto the screen to form an image.ï»'

ESL: Electronic system level. Modifies the word “tools.” ESL tools, still in the process of being defined, are circuit design-automation software that handles things like the integration of hardware and software—a problem barely acknowledged by current design-automation systems.

FBAR: Film bulk acoustic resonator. A type of piezoelectric filter that has allowed cellphone makers to greatly shrink the size of handsets. It replaced another type of filter made from ceramic materials, which was, for a time, one of the bulkiest components in a phone.

Image: John Ueland

FeFET: Ferroelectric field-effect transistor. Imagine a nonvolatile memory that operates at low voltages, stores and disgorges data in mere nanoseconds, and doesn’t destroy the data in order to read it, as some other memory technologies do. You just got all tingly, didn’t you? We thought so. You’d need a functional FeFET to make such a memory. It’s a type of transistor that stores data in a layer of ferroelectric material—stuff that, once polarized by a voltage, stays polarized even after the voltage is gone. For a number of reasons, these transistors aren’t ready for commercialization.

GaN-on-SOD: Gallium nitride–on–silicon on diamond. It’s really hard to make a decent-size wafer of gallium nitride; so people are always trying to build GaN devices on top of other stuff. GaN-on-SOD takes that approach to the extreme. SOD is not to be confused with its evil twin, DOS (diamond on silicon), which has absolutely nothing to do with a PC’s disk operating system.

GMR: Giant magnetoresistor. A spintronic device using the giant magnetoresistive effect. Spintronics is a relatively new endeavor involving nanoelectronic devices that make use of electron spin rather than charge. In a GMR device, a magnetic field produces a change in the resistance of a nanometers-thick conductive layer. The devices make good magnetic switches that are so reliable some implanted defibrillators now incorporate them. The effect is commonly used by hard drive read heads.

IP: Internet Protocol; intellectual property. The first is the communications protocol that makes the World Wide Web go round; the second describes a reason to end up in court. IP is distinguishable only in context and is a particularly pernicious reuse, in that you could easily have a single conversation that includes references to both.

LCOS: Liquid crystal on silicon. A display technology used in some RPTVs (see DMD), in which a layer of liquid crystals is bonded to and controlled by a dense array of transistors on a silicon chip. The liquid crystal can either reflect or block light depending on the voltage at the transistor, thus forming an image. In an RPTV, a bright light reflects off one or more of the chips and projects a magnified image onto the TV screen.

LOC: Lab on a chip. A miniature chemical analysis system built on silicon chips or at least built using chip-manufacturing techniques. Often they include micrometer-scale pumps, valves, and flow sensors. They should be good for on-the-spot chemical analysis, environmental monitoring, medical diagnostics, and other things. Fans of Greek lettering call it a µTAS (micro total analysis system).

LSA: Laser spike annealing. A step in building advanced microchips that prevents current from leaking in nanometer-scale transistors. Basically, it zaps the chip, while it’s still part of a silicon wafer, with a laser, heating it to 1300 °C and fixing defects in its transistors. LSA is not to be confused with latent semantic analysis, a very different technology that lets a computer study a textbook and then pass a multiple-choice exam about the material in it.

LTPS: Low-temperature polycrystalline silicon. A material found at the edges of some new liquid-crystal display (LCD) TVs. The transistors that control the individual pixels of an LCD are made of amorphous silicon, a version with no crystalline structure. Baking amorphous silicon—the high-temperature approach—makes it polycrystalline silicon, or polysilicon, a more crystalline version that makes for faster transistors. LTPS is a more practical idea wherein you crystallize the amorphous silicon by blasting it with a laser. LCD makers use LTPS to form circuits at the edges of the display to better control the pixels. Thus, LTPS is destined to become a term misused by marketers to confuse consumers shopping for large-screen high-definition TVs.

MHEMT: Metamorphic high-electron-mobility transistor. An MHEMT is a variation of a high-electron-mobility transistor (HEMT), which is a type of really fast switch. It’s made by putting a layer of material between two semiconductors whose crystal structures are different enough that they otherwise would not get along. Among other applications, MHEMTs are found in adaptive cruise-control radar in cars you probably cannot afford.

MIPS: Millions of instructions per second. A measure of computer performance—aka meaningless information provided by salesperson. Also microprocessor without interlocked pipeline stages, a chip architecture developed by MIPS Technologies. You might conceivably ask a salesperson from MIPS how many MIPS his MIPS chip does.

MPSOC: Multi-processor system on a chip. An SOC (system on a chip) integrates all the components of an electronic gizmo on a single chip. Reflecting a general trend in the microprocessor business, an MPSOC ups the ante by integrating multiple processors onto the same chip.

MRAM: Magnetoresistive (or magnetic) random access memory. A newly commercialized nonvolatile memory device that uses electron spin, which is related to magnetism, to store data. It combines some of the best attributes of other types of commercial memory technologies, except, at the moment, their low price.

NEMS: Nanoelectromechanical system. A chip with moving parts, or one with parts through which something other than electric current moves. They used to be microelectromechanical systems (MEMS), but they’ve been on a diet.

PDS: Placement-driven synthesis. A design method that lets you specify the behavior of a block of circuitry and translates that into a detailed circuit plan, right down to the size of the parts of the individual transistors. It uses a bunch of nifty techniques, including placement, which figures out the best route for the circuit’s wiring to minimize how long it takes a signal to get where it needs to go on the chip.

PECL: Positive emitter-coupled logic. A way of constructing logic circuits so that they operate faster, but at the expense of continuously draining power. PECL is just ECL operated between positive voltage and ground, in contrast to negative voltage and ground. Another ECL variation is LVPECL (which we dearly wish were pronounced “love peckle”), or low-voltage PECL, a PECL powered from a 3.3-volt power supply. Though CMOS is the logic circuit technology of choice these days, PECLs can still be found transmitting clock signals on chips and sending data down the bus in computers.

QDIP: Quantum-dot infrared photodetector. An IR detector noted for its high temporal resolution and low noise-equivalent temperature difference (NETD), which is a measure of sensitivity. That would seem to make QDIPs good for use in night-vision goggles and other cool stuff, except that they have to be frozen to work well.

RAS: Reusable asset specification. The recycling rules for software engineering. It defines a standard way to package pieces of code so that they can easily be reused. RAS is part of a movement in software design called asset-based development (ABD), in which you keep software reuse in mind as you build a system. Elsewhere RAS means reliability, availability, serviceability, a perfectly sensible set of system performance parameters.

SED: Surface-conduction electron-emitter display. That should be SCEED, right? We thought so. Normally we disdain cheating, but on the other hand we admire the chutzpah required to simply discard 40 percent of your word count to get to a marketable acronym. SEDs are flat-panel displays where each pixel in the display is like a miniature cathode-ray tube. They’re said to have the brightness and contrast of CRTs but to use one-third less power than plasma TVs. The developers are having a devil of a time commercializing SEDs, by the way. SEDs are similar to field-emission displays (FEDs), which you can’t buy yet either.

SIMOX: Separation by implantation of oxygen. A means of producing SOI (silicon-on-insulator) wafers. First you implant oxygen beneath the surface of a silicon wafer. Then you cook the wafer until an insulating oxide layer forms beneath the surface of the silicon. Besides making attractive trivets, SOI wafers let transistors run faster and consume less power, because the insulation layer reduces the amount of charge the transistor needs to move in order to switch and it blocks leaking current.

STRIFE: Stress plus life (testing). A portmanteau posing as an acronym—there’s no reason for this word to be in all capital letters other than the perversity of whoever minted it. We’re fond of it anyway, because it actually means what it says.

TCAM: Ternary content-addressable memory. Say you are looking for the letter Q in a computer memory. If the memory is plain old random-access memory (RAM), you have to check each address one at a time to see if the binary equivalent, 01010001, is hiding there. The same process in a CAM is much faster because it lets you look for a match by reading the entire contents of the memory all at once. CAM is great for things like Internet packet routing and pattern recognition. TCAM takes it to another level by letting you search for 010100XX—or P, Q, R, and S—at the same time.

TLA: What do the following have in common? ATM, VCR, IBM, LSI, GPS, SOI, BER, DVD, RCA, QAM, SDI. Yup. Every last one of them is a three-letter acronym. Two letters are rarely enough. And four letters or more give people the inexplicable urge to try to pronounce them, even if they shouldn’t. Thus, the electronics industry’s penchant for acronyms is so powerful it has its own acronym.

TTL: This used to mean transistor-transistor logic, a way of designing logic ICs without diodes, which took up too much space on the chip. Now that most logic chips are built using CMOS technology, TTL is more likely to refer to time to live, a measure of how long data are allowed to bounce around in a computer network before they get deleted. TTL keeps networks from becoming clogged up with undeliverable data packets.

UTOPIA: Universal test and operations physical interface for asynchronous transfer mode. Basically, an interface that gives test equipment access to a particular kind of data network. Kudos for reaching past five letters, and applause for the matryoshka action of squeezing ATM (asynchronous transfer mode) into the final A. But seriously, if this is your idea of utopia, you really, really need to get out more.

VCSEL: Vertical-cavity surface-emitting laser. A semiconductor laser diode that emits light from the top surface of a chip. That’s in contrast to most laser chips, which are edge-emitting and have a horizontal cavity. Because the latter were invented first, they are usually just called “lasers” instead of HCEELs.

WAF: Wife acceptance factor, wife approval factor. A product feature or modification sufficiently appealing to women that they will permit their husbands to buy the product. An example is the ability of Sony’s PlayStation Portable to play children’s movies as well as Grand Theft Auto.

The SO Family:

Image: John Ueland

The letters S and O have lately become a dizzyingly fecund breeding ground of jargon. Here’s the family tree.

SOI: Silicon on insulator. A type of silicon wafer that improves the speed and power consumption of CMOS circuits. It involves making a thin layer of insulator, usually silicon dioxide, buried up to 100 micrometers below the surface. The insulation layer reduces the amount of charge the transistor needs to move in order to switch, and it blocks leaking current. AMD uses lots of SOI; Intel, not so much.

SOS: Silicon on sapphire. Silicon on insulator (see SOI) by other means. In SOS you deposit a thin layer of crystalline silicon atop a wafer of sapphire—otherwise known by its less bejeweled name, aluminum oxide. SOS circuits are usually found in space and military systems, because the sapphire keeps stray currents caused by radiation from messing with the silicon circuits.

SOAN: Silicon on aluminum nitride. A new version of silicon on insulator (see SOI) where you build a silicon wafer with a thin, electrically insulating layer of aluminum nitride buried just below the surface. The aluminum nitride gives electrical benefits similar to those of SOI’s usual insulator, silicon dioxide, but it’s better at drawing heat away from the circuits.

SOG: Silicon on glass, spin-on glass, sea of gates. In general, SOG is some form of silicon on a substrate of glass. From there, things get complicated. The silicon in question could be polycrystalline, the type found at the edges of newer LCD panels, where manufacturers are building an increasing amount of circuitry right onto the panel. Or SOG could refer to crystalline silicon that is patterned into a circuit and then bonded to glass. SOG can also mean spin-on glass, a way of applying a thin layer of silicon dioxide insulation to chips using a liquid chemical (see SOD for how that works). And finally, SOG is sometimes sea of gates, a way of minimizing the work of chip construction. A sea-of-gates approach assumes lots of different chips can start out as just a vast grid, or “sea,” of logic gates. The only difference between the chips is in how you wire the gates to each other.

SOA: Silicon on anything. In SOA, a favorite of Philips Electronics, you build a circuit on a silicon wafer, then chemically dissolve away almost the entire wafer, starting from the back side, so that only the part containing the circuitry is left. Finally, you glue this sliver of circuits to “anything.” In practice, “anything” seems to mean glass (see SOG). The glass-silicon combination lets you integrate RF components such as inductors onto the chip. On ordinary silicon chips, integrating inductors is difficult to do, because the silicon tends to soak up the inductor’s signal.

SON: Silicon on nothing. Taking the silicon-on-insulator concept to the extreme, researchers—mostly at STMicroelectronics—are experimenting with transistors suspended over small pockets of air, one of the best insulators around.

SOIC: Small-outline IC. A rectangular plastic chip package that has connecting pins that stick out from the side in a so-called gull-wing configuration. That is, the pins stick out, then turn down, and then go out again.

SOD: Spin-on deposition. A method of applying a layer of material to a semiconductor wafer. The wafer is spun like a record, and a liquid is poured onto the center. The wafer’s rotation draws the liquid out from the center, so it forms a layer of uniform thickness across the surface. SOD is the method you’d use to apply a spin-on dielectric, a material used to insulate the on-chip wiring on some ICs.

SOM: Sulfuric acid–ozone mixture. We bet you thought this was silicon on manganese oxide or some other crystal. But no! It’s actually a nasty chemical brew used to scrub silicon wafers clean in between chip processing steps.

SOC System on a chip. An IC that packs a microprocessor, memory, timers, voltage regulators, and all the other stuff you’d need to build a computer or another electronic system. The reconfigurable version of the chip is, predictably, called a system on a programmable chip, or SOPC.

SOP: System on package. Can’t fit your system on a chip? Make it on separate chips and put them in the same package. And integrate all the capacitors and other components into the package while you’re at it. It’s often confused with system in package (SIP). SIP predates SOP and tends not to squeeze in quite so much non-IC stuff.

Scoring

Fewer than 25 correct: Welcome to the club.

25–29 correct: Ph.D.

30–34 correct: Potential game show contestant.

35–39 correct: Übergeek.

40–48 correct: Nobel Prize candidate.

49–50 correct: Liar! (There were only 48.)