| Episodes | 122 |
| Mentions | 188 |
| Cited here | 68 |
| First — last | #3 — #729 |
| Top guests | Jeff Keyzer, Shahriar Shahramian, Ian Danaher |
| Related | oscilloscope · multimeter · open source hardware · automated testing · calibration |
Test equipment is any instrument used to measure other equipment, a definition broad enough to include inexpensive educational instruments when they are used to debug hardware.[87] The category spans a very wide price range, from constrained low-cost tools that cover roughly 80 percent of what most users need to RF and microwave instruments where individual units are rarely worth less than 100,000 dollars.[199][29] Instruments are a major non-recurring cost on a hardware programme, and the market that supplies them — new, used, rented and surplus — is a distinct subject from the measurements the instruments perform.[104] Practical questions of what to buy, when to rent, how to verify a used unit and how to keep a bench calibrated therefore recur across every branch of the field.[104][449][132]
Industry structure
Hewlett-Packard’s founding product, made in the Palo Alto garage that anchors the Silicon Valley startup story, was a piece of test equipment rather than a computing product.[369] When the company split in 1999 the computer and printer business retained the HP name and the test and measurement business was renamed Agilent, so instruments carrying decades of brand equity had to be rebranded.[3] Agilent later separated its electronic measurement business, which became Keysight, following from test and measurement being a lower-margin segment than its rapidly growing life sciences division.[164] A standalone test and measurement company can serve instrument users differently from a division inside a conglomerate, because its chief executive’s attention is on measurement rather than on a larger unrelated business.[164] Tektronix in its classic era was vertically integrated to an unusual degree, manufacturing its own cathode ray tubes and supplying the silver solder needed to work on the ceramic terminal strips inside its instruments.[421]
Incumbent instrument makers repeat a pattern in which they retreat toward ever more expensive high-margin products and abandon the low end, leaving that segment to cheaper entrants.[134] Deliberate feature withholding sustains the tiering: performance that would cannibalise a flagship line is kept out of lower tiers, which receive secondary conveniences instead.[127] Entering the low end is correspondingly difficult, because a new product must ship with a constrained feature set while competing against established Chinese manufacturers who have already reduced the function to an ASIC.[554] Good Will Instek is a Taiwanese instrument maker whose mid-range oscilloscopes and function generators are sold under the Goodwill or Instek brand depending on the country of sale; such makers found themselves undercut by as much as 50 percent by mainland Chinese competitors selling copied products, despite already having a low-cost manufacturing base.[57] Hameg is a European brand with little presence in the United States but a near-leading position in Europe, particularly in education and in low-cost affordable instruments.[205] Entire categories of instruments are made and sold only inside domestic markets such as China and South Korea, including meters that cannot be bought elsewhere.[593]
Where the value sits inside an instrument has shifted. In a modern logic analyser it lies in the software rather than the FPGA and front-end silicon, which is why a vendor differentiating on user interface can move up market and raise prices.[554] Hardware security and protocol testing has likewise moved from proprietary add-on modules bought for oscilloscopes to open-source tools with Python interfaces.[442]
Capability, bandwidth and price
The cost of an instrument is usually set by the frequency range required rather than by any general need for the most expensive model, so specifying the frequency span honestly is the main lever on price.[470] Contrary to the assumption that rising signal speeds would push everyone upward, the bulk of instrument users remain below 100 MHz while a smaller group needs six-and-a-half, seven-and-a-half and eight-and-a-half digit multimeters, so vendors serve a broadening base and a rising top end at once.[199] Measurement above roughly 3.2 GHz sits beyond the price most engineers will pay, which confines characterisation at those frequencies to laboratories holding vector network analysers.[474]
At the low end, reaching 50 MHz of analog bandwidth with off-the-shelf parts is difficult at a low price point, and this is the central engineering constraint on inexpensive USB and student instruments; later work in that segment targeted 250 MHz of analog bandwidth while holding roughly the same price bracket.[302] A survey of three decades of electronics magazine advertising shows the entry-level oscilloscope price point holding steady at roughly 700 to 800 dollars, a level maintained so that cheap models do not undercut the tiers above.[186] Even so, the market changed sharply over the late 2000s and early 2010s, to the point where a 500-dollar instrument offers capability that previously required far more expensive equipment.[199]
Headline specifications can mislead, because performance is limited by the weakest subsystem: an expensive high-bandwidth front end delivers little if the trigger system has excessive jitter, since rare events cannot then be reliably captured.[178] High-end instruments are built in low volume and therefore receive far less user-interface engineering than mass-market low-cost instruments, whose large user base generates constant usability feedback.[104] The feature set of a flagship follows from its purpose: a high-end oscilloscope is designed primarily for automated compliance analysis of serial standards such as SATA rather than simply displaying a waveform, which explains both its capabilities and its interface complexity.[104] Instrument user interfaces have historically lagged behind other product categories, which is part of why engineers still favour physical knobs over menu-driven front panels.[382]
Precision instruments carry their specifications in their component selection. A single precision resistor can cost four dollars in volume against roughly twelve dollars at single-piece price, so unit economics at low volume consume most of the margin.[554] Designers of high-end instruments routinely reject integrated single-chip signal-chain solutions in favour of discrete implementations, because instrument specifications are hard requirements that a general-purpose integrated part cannot be trimmed to meet.[348] Open-source precision designs are frequently rebuilt with lower-precision components, and the resulting units do not meet the original specification even though the schematic is identical.[554] Selling to a stated specification requires custom-built test gear to verify each individual unit, and traceability back to a national standard adds further cost.[554]
Buying, renting and the used market
Instruments needed only occasionally are better rented than bought, and companies developing high-speed products often find test equipment consuming a large share of the research and development budget.[104] The same reasoning applies to a large modern oscilloscope wanted for a single project.[613] Well-equipped labs typically contain many instruments used only a small fraction of the time, which is the standing argument for weighing occasional-use purchases against rental or borrowing.[216] Rental has its own failure mode: a unit taken for a few days stays on the books for years, because nobody revisits the arrangement once the immediate need has passed.[104] In smaller markets distributors carry little or no stock of obscure or high-end instruments and their demonstration loan units are often already out with another customer, which forces buyers into rental in the first place.[104]
Purchases are best tied to a specific project, since buying instruments can become a substitute for building anything with them.[328] Buying a multi-thousand-dollar instrument is reasonably deferred until a job actually requires it, given how much constrained low-cost tools cover.[199] Against that, persisting with an inadequate low-end instrument can consume far more effort than reaching for the right one at the outset, even though the struggle has some teaching value.[203] Non-recurring engineering costs such as tooling and test equipment are significant on a hardware programme, but staff cost usually dominates, which is why firms under pressure shed people before they shed instruments.[104]
Used and surplus
Access to surplus is strongly regional: the United States has a deep surplus market where a lab of nominally high list-price instruments can be assembled for a small fraction of that, while Australia has almost no equivalent supply.[22] The regional gap sustains a durable arbitrage, since buyers in some countries will not risk overseas purchases; an importer who takes that risk, repairs and cleans the unit and resells locally can realise around three times the purchase price.[490] It also imposes costs, with imported faulty and used instruments carrying shipping charges that can reach around 500 dollars per unit, which changes what is worth repairing.[643] A related hobbyist pattern is to buy instruments cheaply on eBay, refurbish them and resell, cycling through a large number of units over the years.[18]
Surplus dealers differ in what they stock: some are large undifferentiated piles of mixed junk, while smaller curated dealers concentrate on industrial gear and instruments that stand a reasonable chance of still working.[81] All Electronics, a United States catalogue component supplier, also operates a walk-in retail store in Van Nuys, California.[81] When an established test and measurement service business closes, its warehouse stock of older instruments typically cannot be sold individually and is disposed of by the pallet load or scrapped as electronic waste; such liquidations release warehouses of equipment, and local enthusiasts strip the highest-value items first so that later buyers see picked-over stock.[727]
Screening a used unit rests on a few reliable tests. Because bench instruments carry comprehensive built-in self-test, a unit that powers up and passes self-test can be judged 80 to 90 percent likely to be usable.[449] A seller’s claim that a unit was pulled from a working environment is unreliable when it comes from an individual seller, whereas instruments liquidated in bulk by an auction house after a lab or factory closure are near-certain to have been in service.[449] Refurbished dealers test what they sell and typically charge about double the price of an as-is private seller, which is the price of removing the risk that the unit is faulty.[449] Instruments bought from another mains region are often not switchable between supply voltages and require modification at the printed circuit board level, a hidden cost of importing.[419] Some vendors police the secondhand and clone market by searching eBay listings and filing reports to have them removed, which affects what used buyers can find.[273] Prices also respond to visibility: public attention to a particular vintage model raises its price on the used market.[613] Even faulty high-end instruments hold substantial value, with broken examples routinely selling for thousands of dollars, so repair-for-profit at the top of the market requires real capital.[470]
Some brands resell better than others. Stanford Research Systems instruments hold their value unusually well because the company still builds the same designs decades later, uses through-hole construction, and publishes service manuals, all of which keep old units repairable.[613]
Composition of a bench
A typical beginner’s progression runs from multimeters and a simple kit oscilloscope to a two-channel oscilloscope and a signal analyser, driven by work such as PWM debugging that a single-channel instrument cannot resolve.[278] Modern embedded development often needs only a debugger and a known-good board rather than a large setup, because contemporary instruments are small, comparatively inexpensive and very capable.[489] Paying for instruments whose problem is already solved in software, such as a logic analyser with a well-designed capture and decode interface, buys back engineering time even though the hardware itself is expensive.[373] Instrument design has trended toward consolidation, with mixed-domain products absorbing functions that were once separate boxes.[244] All-in-one instruments illustrate the price of that: National Instruments’ VirtualBench combines several instruments in one box at a price around two thousand dollars, above the reach of most hobby and educational buyers, and at around 5,000 dollars targets academic benches that need adequate acquisition rather than standards-class accuracy and cannot house a full set of separate instruments.[242][354]
Engineers who travel frequently need instrumentation that is not heavy and bench-based, which is the design driver behind small USB-host-powered oscilloscopes and analysers.[302] Small self-contained instruments such as thermocouple data loggers now make continuous retroactive measurement of physical variables cheap, replacing the guesswork that previously accompanied intermittent faults.[442]
Physical arrangement matters on a working bench. A common convention places less frequently used instruments on higher shelves and keeps daily-use instruments at eye or hand level.[676] Suspending an oscilloscope beneath a shelf, alongside the bench lighting, recovers bench area in a small workspace while keeping the display in view, and mounting a source measure unit on a bracket at the back of the bench so its terminals sit at board level shortens the lead path to the device under test.[676] In shared lab space, marking an instrument with a prominent fault label is an effective way to stop it being borrowed and to keep a known-good reference available.[404] A working collection accumulates duplicates and project-specific instruments that outlive their purpose, and periodically moving those on keeps the bench usable.[655] Individual collections can nonetheless reach industrial scale: one documented inventory listed 88 multimeters, 38 oscilloscopes, 11 logic analysers and 58 function and pulse generators, catalogued down to counts of knobs, switches and BNC jacks.[688]
Practising RF engineering requires a large quantity of expensive instrumentation, which is a structural reason engineers in the field gravitate to employers who already own the gear rather than working from home labs; home RF benches tend to be assembled from vintage instruments whose frequency coverage and noise performance are inadequate for modern work on Bluetooth or Wi-Fi.[93] Work that depends on shared instrumentation argues for co-locating a team rather than distributing it, because the equipment cannot be duplicated at every remote desk.[101] Engineers who produce public measurement content while employed by a research laboratory can keep the two strictly separate by owning all the instruments used for the public work.[228]
Calibration and upkeep
Calibration laboratories must hold environmental conditions inside defined limits to issue certification, typically a temperature window of about 5 degrees Celsius and relative humidity in the region of 30 to 80 percent.[132] Environment of use is itself a factor in condition and drift: instruments returned for calibration from harsh industrial sites are often found coated internally with contamination.[132] Stored instruments should be powered up periodically, because equipment left unused for long stretches is more likely to fail when it is finally switched on.[688] Owning a second identical instrument gives a benchmark to compare against while modifying or repairing the first, which is the practical way to confirm that a change improved rather than degraded performance.[643]
Repair and vintage instruments
Repairing test equipment is an effective way to learn both electronics design and instrument design, because earlier equipment solves the same measurement problems from a different angle than a modern instrument that hides its method behind software.[655][613] Instruments from the 1960s and 1970s were simpler in construction than modern equivalents, so opening one exposes the design decisions and manufacturing process in a way a current instrument does not.[421] Studying category-defining instruments reveals the measurement philosophy behind them; the HP 8753 occupies that position among comparatively low-cost vector network analysers.[613]
The work has become genuinely hard, because instruments now around forty years old have no factory parts support left, a much larger technology gap than earlier restoration work spanned.[655] When a repair requires fabricating an unobtainable part, cutting a batch of ten to thirty costs little more than cutting one on a laser cutter, and the surplus can be sold to other restorers.[655] Routine tasks include replacing the perished foam filter behind the rear cooling fan of a vintage oscilloscope, which keeps dust, dirt and hair out of the instrument.[655] The TekScopes groups.io mailing list is the principal community for vintage Tektronix repair and is notably more active than the equivalent HP and Agilent list, and the Tektronix community also maintains a museum for its vintage equipment.[655] A first oscilloscope should be a known-working unit, even a low-end one, because repairing a second, broken instrument requires a trusted reference to measure against.[449]
Vintage instruments carry their own conventions. Equipment predating the 1960s is marked in kilocycles and megacycles rather than kilohertz and megahertz, which can confuse users.[407] Analog circuit designers of the 1960s and 1970s worked without simulators, computers or even pocket calculators, relying on slide rules and on far more limited instrumentation than is now routine.[47]
Production and factory test
In a mature production line the marginal cost of testing can approach zero once the automated test equipment is fully depreciated, at which point line uptime rather than test cost governs the economics.[71] Throughput is capped by the number of test stands available, since each unit occupies a stand for the duration of its test, which is the economic argument for built-in self-test.[393] High-speed production test equipment is often designed in-house to test subassemblies on the factory floor before final assembly, as with audio modules verified before being fitted into greeting cards.[424] Calibration and functional test equipment is not part of a normal contract manufacturer’s capability, so a product requiring per-unit calibration obliges its designer to supply and maintain that equipment at the CM.[607]
Production test systems commonly remain on obsolete operating systems such as Windows 3.11 or DOS for decades, because the equipment still produces working product and there is no forcing reason to revalidate a new platform.[263] Knowledge of how the gear works and how to maintain it often ends up concentrated in a single employee, which is a maintenance risk for the organisation.[108] Experienced production technicians carry empirical fault knowledge about specific boards, so consulting them before building simulations saves time that theorising alone will waste.[108]
Analysing the resulting data has its own discipline. Grouping production measurement data by fixture, chamber or operator exposes systematic offsets, such as one chamber consistently reading higher, that aggregate pass or fail rates would never reveal.[328] Gauge repeatability and reproducibility studies are worth doing religiously on factory fixtures as process control, but must not replace first checking that the equipment is connected, in calibration and physically sound, or weeks of data collection will be wasted.[328] Test equipment is also a determinant of part lifetime upstream: semiconductor parts are usually obsoleted only when something physical forces it, such as a fab closing or the production test equipment for that part becoming unusable, at which point customers are offered a last-time buy.[270]
Independent product testing at credible scale requires major capital infrastructure, including RF anechoic chambers, acoustic chambers and dedicated power supply test benches, plus engineers hired specifically to run them.[569]
Building and automating instruments
Measurements for which no commercial instrument exists require building custom gear, as in a purpose-built drift counter used to track oscillator frequency drift over time; the absence of any commercial instrument capable of a measurement is a useful signal that the work has crossed from engineering into science.[127] Historically this was the normal case, and electronics magazines of earlier decades were filled with test equipment construction projects because affordable commercial instruments did not exist — a digital storage oscilloscope kit sold in the thousands simply because a ready-made DSO could not be bought.[229] That has reversed for common bench instruments: a conventional 30 volt 3 amp linear bench supply can be bought for around 50 dollars, leaving DIY justified only for niche requirements.[229] A standardised open hardware platform with interchangeable analog front ends can act as a general-purpose instrument chassis, letting analog designers contribute front ends while software work is shared.[198] Low-cost exploratory instruments intended for learning are not substitutes for production instrumentation, which must do things like curve-trace transistors coming off a line to confirm they meet specification.[87]
Regulation treats instruments as a special case: radio rules in the United States and across Europe contain exceptions for test equipment, which is how a broadband transmitter covering roughly 1 MHz to 6 GHz can legally be sold when it would not be permitted as a consumer device.[265]
Automation has become broadly available. Many mainstream instruments expose Python control libraries, including Rigol oscilloscopes and the Saleae logic analyser, which allows instrumentation to be driven from a continuous integration harness rather than by hand.[461] sigrok is an open-source driver layer providing a common interface to a very wide range of oscilloscopes, multimeters and logic analysers, including Bluetooth-connected handheld meters, with separate graphical front ends built on top of it.[665]
Instruments and skill
Acquiring an oscilloscope is a turning point for a self-taught engineer because seeing signals in the time domain converts abstract circuit behaviour into something observable.[127] Hands-on use builds intuition and mental models that theory alone does not produce, because the operator can vary a parameter and immediately observe the effect.[470] Bringing an instrument into a lecture and showing the mathematics applied to a real measurement, such as an FFT displayed on a spectrum instrument, makes the theory concrete.[465] Being able to walk up to an unfamiliar instrument and navigate its menus is a distinct professional skill, complementary to having a small set of personally owned tools used for every job.[551]
Instruments also shape what work is possible on a given target. Ease of hardware modification depends heavily on whether a board exposes labelled test points; some game controller boards provide clearly labelled test pads while others route button contacts as carbon pads on a printed silkscreen circuit that offers nothing to probe or solder to, in which case a third-party equivalent product may be far more tractable.[490]
The high-frequency extreme
At the top of the frequency range the instruments themselves become the limiting infrastructure. On-wafer probing extends to about 750 GHz in dedicated probe rooms, and probes for those frequencies use MEMS contactors that are fragile and can only be repaired by returning them to the manufacturer.[729] The most expensive instruments and probes tend to be the most fragile, because performance at extreme frequencies is achieved through delicate mechanical structures with no margin for mishandling.[729] At very high data rates small impedance discontinuities accumulate quickly and erode link performance, so installed high-speed infrastructure must be measured rather than assumed to meet the rate it was specified for.[533] The International Microwave Symposium carries a large industry exhibition showing state-of-the-art instruments from every manufacturer, whereas circuit and ASIC conferences have little physical hardware to demonstrate.[228]