Summary
- U.S. Patent 5,436,567 names Donald J. Wexler and Jeffrey L. Smith as inventors and Automated Test Engineering, Inc. as assignee. Filed in 1993 and issued in 1995, it describes a clamshell fixture that uses vacuum below a printed circuit board and mechanically driven probes above it.
- The design record exposes the practical problem behind automated board testing: a fixture must align many spring-loaded probes, apply force without damaging them, reach test points on both sides, and return to the same contact geometry over repeated cycles.
- A 1993 network record independently shows Automated Test Engineering in San Jose, while later public biographical references must remain secondary and attributed. The available evidence does not establish company scale, commercial outcome, customers, market position or a complete executive history.
The hidden layer between a circuit board and a test result
Electronics manufacturing is often described through the devices being assembled or the instruments that decide whether those devices pass. Between those two sits a less visible layer: the fixture that physically presents a device-under-test to an automatic test system. It is the fixture that has to place a field of probes onto selected electrical points, hold the board in a controlled position, survive repeated actuation and release the board for the next cycle. Without that interface, even a capable test instrument cannot make a dependable measurement.
The public record around Donald J. Wexler is unusually useful because one document makes that interface concrete. U.S. Patent 5,436,567, issued on July 25, 1995, names Wexler and Jeffrey L. Smith as co-inventors. Automated Test Engineering, Inc. of San Jose is identified as the assignee. The application was filed on February 8, 1993, and the title describes the mechanism directly: a double-sided automatic-test-equipment probe clamshell with vacuum-actuated bottom contacts and mechanically actuated top contacts.
That title is more than a catalog label. It lays out a division of mechanical work. Vacuum draws the lower plate and frame toward the base, allowing base-mounted probes to reach through the plate to the printed circuit board. Pushrods carry the resulting relative movement into mechanisms mounted in the upper frame. Rack-and-pinion elements reverse the direction of that force, moving a top plate and its probes toward the board. The same actuation therefore brings spring-loaded contacts to test points on both faces of the device-under-test.
The design belongs to manufacturing operations because its value depends on repeated physical execution. A probe that reaches the correct pad once but arrives at an angle on the next cycle is not a stable interface. A fixture that closes but blocks every top-side adjustment limits what can be done during a test. A structure that applies enough force but concentrates it poorly can bend probes, misregister contacts or stress the board. The patent is organized around those constraints, not around a claim of abstract automation.
This narrow evidence also gives Wexler a defensible place in the history of test engineering. It does not require an expansive company story or a retrospective claim about business success. The inventor and assignee fields establish who is named and where the invention was assigned. The detailed description shows the manufacturing problem they chose to solve. That combination is enough to examine a consequential but easily overlooked part of electronics production: making electrical contact repeatable.
What the record establishes, and what it leaves open
The strongest person-level evidence is the patent itself. It identifies Donald J. Wexler and Jeffrey L. Smith, locates both in San Jose at the time represented by the document, and assigns the invention to Automated Test Engineering. It also provides fixed dates, a technical title, six claims and three drawing sheets. Those are formal facts within the patent record. They establish inventorship and assignment for this specific mechanism; they do not establish who conceived every feature, how responsibilities were divided or whether the design was produced at scale.
A separate 1993 archive of network registrations supplies a modest corporate cross-check. In a long list of connected organizations, it includes Automated Test Engineering, Inc. in San Jose, associates it with NETCOM and shows the network identifier 198.177.149 under the short label ATE. This entry does not describe products or leadership. Its importance is narrower: Automated Test Engineering had a public technical footprint under its own name during the same period in which the patent application was filed.
Later biographical material is a different class of evidence. A public obituary supplies a wider career account, including company leadership and later technology work. Those statements are useful only as attributed obituary claims. The patent and 1993 network entry do not independently verify the corporate title, transaction chronology, buyer, financial terms or later employment account, so none of those claims carries the technical thesis here.
A 2016 report from an Apple shareholder meeting identifies a Don Wexler of Lake Tahoe as a second-time attendee and shareholder who reacted favorably to the event. The location and name are consistent with the obituary's public account, but the report says nothing about an employment role. Attendance at a shareholder meeting cannot be converted into evidence of work on an Apple platform, management authority or a product assignment.
The resulting profile is intentionally constrained. No available document establishes Automated Test Engineering's customer count, revenue, production volume, market share or leadership in the fixture business. No evidence here identifies the buyer of the company or the commercial outcome of the patented design. The article can describe a documented invention and a contemporaneous company footprint. It cannot turn those facts into an unverified success narrative.
Why populated boards made probing harder
The patent begins with a short history of the testing problem. Early printed circuit boards, it explains, commonly placed through-hole components on one side and an etched copper interconnection pattern on the other. With relatively little crowding and a lower degree of integration, automatic loaded-board testing could use an array of spring-loaded probe pins against strategic points on the accessible face. The familiar term for that array was a bed of nails.
That arrangement worked because board construction and test access were aligned. Component leads passed through the board, and the interconnection pattern could be reached from the opposite side. A fixture needed to support and locate the board, but the electrical access problem was predominantly one-sided. The patent describes test-equipment development in that period as being concerned more with signal simulation, interpretation and faster computing than with the mechanics of reaching dense points on two faces.
Surface-mount devices changed the geometry described in the patent. Their pins could not be expected to continue through vias to the bottom of the board. Useful signals and contact points could therefore remain on the component side. Testing a populated board increasingly required access to both faces, not simply a larger array beneath it. A one-sided fixture could no longer assume that every electrically significant location had a convenient counterpart on the underside.
The immediate answer, double-sided probing, created another constraint. The patent notes that vacuum fixtures had become popular because vacuum could draw a board evenly onto a matching bed of nails. Atmospheric pressure spread over the board area could overcome the combined spring force of a hundred or more probes without relying on a small number of concentrated clamps. For a one-sided arrangement, this also left the top surface comparatively accessible.
Applying vacuum on both sides solved the force problem in one way but enclosed the board. The patent's background says such fixtures obstructed access: switches could not readily be set, option jumpers changed or variable resistors adjusted while the board was under test. It also characterizes double-sided vacuum fixtures as costly to construct and as a source of slower production time. Those statements are the inventors' description of the prior problem, not quantified performance findings, but they define the target clearly.
The design challenge was therefore not merely to add another bed of nails. Both arrays had to approach their test points in a controlled direction. The lower half still needed the distributed force and seal offered by vacuum. The upper half needed to move without requiring its own vacuum chamber over the board. The complete structure also had to open, accept a device-under-test, close without the probes interfering with placement, actuate both arrays and then release cleanly.
This is the repeatability problem in its basic form. Board geometry, test-point location, probe travel, fixture deflection and actuation force all meet at the instant of electrical contact. The patent does not claim that mechanics replace electrical test design. It shows that the electrical decision depends on a mechanical interface capable of returning many contacts to predetermined locations.
A bed of nails is a positioning system
The phrase “bed of nails” can make the fixture sound passive, as though it were only a plate filled with pins. The patent describes something more demanding. Each contact pin is a spring probe placed according to the pattern of test points on a particular board. The upper and lower arrays must remain registered to that pattern while moving far enough to clear the board during loading and far enough in the other direction to make reliable electrical contact during the test.
Spring loading lets an individual probe accommodate controlled travel once it reaches a pad. It does not remove the need for accurate alignment. If the fixture shifts laterally, the probe can meet the edge of a pad or miss it. If the approach is not perpendicular, a probe can be exposed to a side load rather than compression along its intended axis. The patent expressly associates perpendicular engagement with a reduced risk of bending or breaking probes and with lower risk of misregistration.
The document even records how application-specific the contact interface could be. It describes probes made with plated tubing, springs and plungers, and lists possible tip forms including crown, pyramid, serrated, cup and pointed shapes. Those alternatives indicate that the final point of contact was selected for the application. Yet every tip choice still depends on the larger fixture holding its approach path and travel within the intended geometry.
The board itself forms part of the lower vacuum arrangement. It rests on the bottom plate and completes the seal bounded by the bottom frame, gasket and base. That means board location is not incidental to actuation. The same placement that brings the board into registration also closes the pressure system that moves the lower plate relative to the base. A poor fit would be both a location problem and a vacuum problem.
The fixture therefore coordinates three systems at once. It is a mechanical carrier that receives the board. It is a pressure-driven actuator that converts vacuum into relative movement. It is an electrical interface that brings a patterned set of probes to selected points. Repeatability comes from keeping those systems coupled in the same sequence every time: locate, close, actuate, make contact, release and open.
This view helps explain why fixture engineering is not secondary to the test program. The instrument can apply stimuli and interpret responses only after the interface has established the intended electrical paths. A changing interface can make a stable board look inconsistent, while a mechanically repeatable interface gives the test system a sound basis for comparing one cycle with another. The patent does not provide false-failure rates or lifetime data, so no numerical benefit can be assigned. It does, however, describe the structural choices intended to control the main sources of contact variation.
The clamshell divides loading from actuation
The preferred embodiment is a hinged clamshell. A top frame with a latch opens away from a bottom frame with a catch, leaving the lower plate available to receive the printed circuit board. A base beneath the bottom frame connects to an external vacuum mechanism through a port. A gasket joins the base and lower frame, allowing their spacing to change when vacuum is applied while preserving the seal.
This open state is central to the loading sequence. The patent's drawings and description show the upper and lower contacts retracted while the board is introduced. The probes do not yet touch or interfere with it. The board settles into the bottom plate, where its position completes the vacuum boundary. Separating loading from probing reduces the need to slide a board across extended pins or force it into an already engaged contact field.
When the top frame closes and latches, the system enters a second state. The board is enclosed between the frame assemblies, but both probe arrays remain retracted. The upper and lower pins are aligned perpendicular to the predetermined points they will eventually touch. Four pushrods attached to the base are now in mechanical relation with four rack-and-pinion actuators mounted around the upper plate, but they have not yet advanced those mechanisms.
Only the application of vacuum creates the third state. Vacuum draws the bottom plate and frame toward the base and compresses the gasket. Because the lower contact pins are attached to the base, that relative movement exposes them through the bottom plate and brings their tips to the underside of the board. At the same time, the pushrods fixed to the base advance relative to the lower structure. They press the input elements of the actuators in the top frame. Those mechanisms reverse the direction of movement, driving the top plate and its probes toward the board from above.
The arrangement links two-sided engagement to one actuation event without placing a vacuum enclosure on the top face. The device-under-test is clamped between arrays that approach from opposite directions. The bottom probes are moved through the vacuum-driven relationship between the base and lower plate. The top probes are moved mechanically by force transmitted from that same relationship through the pushrods and gear assemblies.
Four actuators are shown at the corners of the top plate, but the patent does not make that count universal. It says fewer may be used for a smaller plate and more for a larger plate. That qualification reveals the function of the distributed mechanisms: they support movement of the probe plate over its area. Scaling the plate changes the support requirement, so actuator count is treated as a design parameter rather than a symbolic feature.
The hinged structure also organizes fixture setup. The top plate carries the upper probe pattern; the bottom plate locates the board and guides the lower probes; the frames maintain their relationship through the hinge and latch. Each part has a defined role in returning the arrays to the same face-to-face arrangement. The clamshell is thus not only a cover. It is the reference structure through which the two probe fields are opened, aligned and brought back together.
No production throughput is stated in the patent, and the document does not compare cycle time against a measured alternative. The operational contribution is instead visible in the sequence. Loading occurs with contacts clear. Closing establishes alignment without immediate probing. Vacuum then actuates both sides. Removing vacuum allows the mechanism to retract before the fixture is opened. A repeatable test cycle is built from those discrete states.
Turning vacuum motion into opposite probe motion
The most distinctive mechanical step is the transfer of force to the top plate. Vacuum acts below the board, yet the upper probes must move downward toward it. A direct rod moving upward from the lower base would push in the wrong direction for that job. The patent's rack-and-pinion actuator converts the incoming pushrod motion into movement in the opposite direction.
In the described actuator, one rack receives the pushrod force. A pinion gear couples that rack to a second rack attached to the upper probe plate. As the input rack advances, it rotates the pinion; the second rack travels oppositely and carries the plate toward the top face of the board. A compression spring biases the mechanism back toward its relaxed position so the plate can retract when the pushrod force is removed.
This is a compact form of mechanical logic. The mechanism does not need a separate powered actuator in the lid. It derives the timing and magnitude of upper movement from the lower vacuum stroke. When the lower assembly has not moved, the pushrods have not driven the gears and the upper probes remain clear. When vacuum produces the lower stroke, the same event advances the pushrods and engages the upper field.
The coupling also makes mechanical setup consequential. The patent says any gap between the pushrods and actuators in the closed, unactuated state subtracts from the available motion at the upper plate. It calls the reach of the pushrods and the mounting positions of the actuators critical. The preferred description has the pushrods press-fitted into the base and not adjustable by the user.
That detail is a direct statement about repeatability. If the pushrod height were casually adjustable, upper probe travel could vary with setup or intervention. Fixing the rods and controlling the actuator positions makes the conversion geometry a property of the fixture. The patent does not supply tolerances, but it identifies where dimensional variation would enter: the initial gap, pushrod reach, actuator location, gasket compression and resulting rack travel.
Alternative gear-box embodiments in the document preserve the same principle. They use paired racks, a pin hub or pinion relationship, bearings and mounting features to transmit the opposed movement. The exact construction can vary while the functional chain remains stable: lower vacuum motion produces pushrod force; the geared element reverses direction; the upper plate advances; the spring supports return when force disappears.
The claims generalize that chain. They describe a top, door-mounted probe array, a bottom bed-of-nails array, a compressing structure around the device-under-test, a lever arrangement that receives force in one perpendicular direction and drives the upper probes in the opposite direction, and a vacuum-actuated element that locks down the board while driving lower contact. Additional claims cover the hinged door, pushrod generation, rack-and-pinion coupling, spring return and fastening of the upper array.
Those claims should not be read as evidence that every later fixture used this design or that the patent dominated a market. Their value for this profile is more grounded. They show how Wexler and Smith framed the invention: not as a loose collection of parts, but as a coordinated way to obtain opposed, perpendicular probe engagement from one vacuum-driven cycle.
Repeatability begins before electrical measurement
The patent never reduces repeatability to a single specification. Instead, repeatability appears across the sequence. The board must nest in the lower plate in a known position. The top frame must close against the bottom frame. The probe arrays must be aligned before actuation. The vacuum stroke must compress the gasket and move the lower structure. The pushrods must meet the upper actuators with controlled spacing. The gears must translate that stroke into upper-plate movement. Finally, the probes must contact their assigned points along their intended axes.
Each dependency can influence whether the electrical system sees the same interface again. Board placement defines the relationship between test pads and both probe patterns. Frame closure defines the relationship between the upper and lower assemblies. Pushrod reach influences the available top-side travel. Probe spring compression influences contact force at each point. Retraction determines whether the next board can be loaded without interference.
The design's use of perpendicular contact is especially important in this chain. The patent depicts both sets of probes aligned to the board before they extend. They then move toward the board rather than sweeping across it as the lid closes. Closing and probing are separate actions. This reduces the tendency of hinge motion to drag upper contacts along an arc into their targets.
That distinction is easy to miss. A hinged cover naturally rotates, but a probe is intended to compress longitudinally. Mounting fixed probes directly in a rotating lid could expose them to lateral movement at the point of engagement. The patent instead mounts the probes in a plate that translates after the lid is closed. The hinge handles access; the rack-driven plate handles the final probe approach.
Vacuum contributes another kind of control. The patent's background favors it for distributing force across the board area and overcoming the accumulated spring force of many probes. The lower arrangement uses the board as part of the seal and draws the fixture elements together. That does not mean every location receives an identical force, and the patent gives no force map. It means the design uses a broad pressure mechanism rather than relying only on a few point clamps to oppose the probe field.
Repeatability also depends on release. Once vacuum is removed, the lower structure can return and the springs in the upper actuators can retract the top plate. Contacts clear before the clamshell opens. A mechanism that engaged precisely but released unpredictably would still be unsuitable for repeated handling. The patent's relaxed, closed and vacuum-actuated states account for both directions of the cycle.
The useful lesson is not that mechanics guarantee a correct test result. Electrical coverage, stimulus design, limits and interpretation remain separate questions. The fixture's role is to reduce variation in the physical connection so those questions can be answered against a more consistent interface. The patent documents a design effort aimed at that precondition.
Perpendicular engagement protects the interface
Probe pins are compliant in a particular direction. Their spring action is intended to absorb travel along the shaft while maintaining pressure at the tip. Side loading asks the same slender component to behave as a lever. The patent explicitly presents perpendicular engagement as a way to reduce bending, breakage and misregistration, making the direction of motion part of the reliability strategy.
The clamshell separates gross movement from fine engagement to preserve that direction. The lid can rotate through a large arc while open because the upper probes are retracted. After the lid is closed, the top plate translates toward the board. The lower probes likewise advance through the lower plate as vacuum draws the assembly together. At the contact stage, both arrays move generally normal to the board plane.
This division also protects the patterned relationship. Test points are not interchangeable; each probe is routed for a particular electrical purpose. A small lateral error can place a tip on solder mask, a neighboring feature or an edge rather than the intended location. By aligning before extension, the fixture makes engagement an axial travel problem instead of asking the contacts to find their points during closure.
The patent's list of different tip shapes reinforces this dependence. A crown, serrated form or point interacts with a surface differently, but none can compensate for a fixture that approaches the wrong location. Contact design and fixture design are therefore nested decisions. The selected tip handles the local interface; the plate, frames and actuators deliver it to that interface.
The same reasoning applies to probe preservation over repeated cycles. The document does not report lifetime tests, so it cannot support a claim about how many actuations the mechanism achieved. It does identify the load condition it was designed to avoid. Reducing lateral force and misregistration is a credible mechanical objective because those conditions are named directly in the patent. Any further claim about durability or maintenance savings would require evidence not present here.
Keeping the top of the board available
The invention's other stated objective is access. A double-sided vacuum fixture could enclose the top surface to maintain a second seal. The patent argues that this made it difficult to operate switches, change jumpers or adjust variable resistors during a test. Its proposed clamshell uses vacuum only in the lower assembly and transfers force mechanically to the upper probes, so the top does not need its own vacuum chamber.
Top-side probes still occupy selected locations, and the closed frame still surrounds the board. “Access” therefore should not be exaggerated into a claim that every component is unobstructed. The supported point is comparative and structural: the top probe action does not depend on preserving a vacuum over the entire upper face. Openings can be provided around the top plate and probe pattern for the adjustments contemplated by the patent.
That matters in loaded-board testing because some procedures are interactive. The patent specifically envisages setting switches, changing option jumpers and adjusting variable resistors while the device-under-test remains connected. These actions require stable electrical contact and some physical reach at the same time. The clamshell attempts to provide both by separating the means of upper actuation from a top-side pressure seal.
The design also claims rapid adaptation for customized applications. The probe arrays are patterned for the board, and actuator count can change with upper-plate size. Contact-tip form can vary with the target. Those elements indicate a fixture architecture meant to be configured around a device-under-test rather than a universal claim that one plate fits every board.
Customization, however, is not the same as effortless changeover. The patent itself makes pushrod reach and actuator placement critical, and every probe pattern must correspond to specific test points. No setup-time data or cost comparison is given. The defensible interpretation is that the architecture exposes adjustable design choices within a common mechanism while fixing sensitive motion relationships once the fixture is built.
This combination of access and controlled customization returns to the central operational problem. A useful fixture must do more than close around a board. It must let engineers build the required electrical interface, preserve the geometry of that interface and support the actions a test procedure requires. The patent's solution is physical: a hinged structure for access, translating plates for perpendicular engagement and one vacuum-derived motion chain for coordinated contact.
Automated Test Engineering's 1990s public footprint
The network archive dated November 1993 is not a product brochure, company registry or financial filing. It is a technical list concerned with network policy and connected organizations. Within that context, Automated Test Engineering appears under its full corporate name in San Jose, with a NETCOM association and the identifier 198.177.149 labeled ATE in the United States.
This is a small but useful piece of independent chronology. The patent application had been filed earlier that year and named Automated Test Engineering as assignee. The network record shows an organization of the same name and city maintaining a visible network entry during the same period. Together, the documents place the patented fixture and the company footprint in a shared early-1990s context.
The network entry should not be asked to prove more. Possession of a network identifier does not establish employee count, technical sophistication across the business, sales reach or manufacturing volume. The NETCOM notation describes connectivity in the record, not a commercial endorsement. The label ATE links the listed network to the organization, but it supplies no organization chart and names no individual.
Its value lies precisely in that restraint. Company histories often become dependent on later recollections or promotional summaries. Here, a contemporaneous technical artifact confirms that Automated Test Engineering existed publicly under that name while the patent process was underway. It supports the corporate setting without validating every later biographical statement about formation, leadership or sale.
The two records also reveal different parts of a small technology company's public surface. The patent shows intellectual property assigned to the company and describes a fixture in detail. The network list shows the company as a connected organization. Neither records customers or financial results. What survives is an engineering and infrastructure trace rather than a complete business archive.
For Wexler's profile, the patent remains primary because it names him. The network entry strengthens the existence and period context of the assignee, not his title. Keeping those roles separate prevents corporate presence from being mistaken for proof of executive status while still showing that the invention was situated inside an identifiable operating organization.
The biographical boundary around Don Wexler
The obituary for Donald Joseph Wexler supplies the broadest career narrative in the available material. It says he completed an electrical-engineering degree at San Diego State University, served in an electronics group in the U.S. Air Force, began a technology career with Hewlett-Packard and later started Automated Test Engineering. It further says he ran the company as chief executive until a 1998 sale and later worked at Apple.
Those statements can orient the public record, but they remain claims published in an obituary. The patent independently verifies a Donald J. Wexler as co-inventor and Automated Test Engineering as assignee. The 1993 network list independently verifies the company's contemporary public footprint. Neither independently confirms the asserted founder title, chief-executive tenure, sale chronology, buyer or Apple work. This article therefore does not use those parts of the obituary to establish authority or commercial achievement.
The 2016 shareholder-meeting report adds only a narrow later reference. It identifies Don Wexler of Lake Tahoe as a shareholder attending the meeting for a second time and records his positive reaction. That report is consistent with the location in the obituary, but it does not identify him as an Apple employee. A shareholder's presence and an obituary's employment account are different facts from different kinds of publication; combining them would not create independent verification of the employment claim.
Private-life details in the obituary are outside the manufacturing-test subject and need not be repeated. They do not help explain the fixture, the assignee record or the repeatability problem. Excluding them also keeps the profile from borrowing emotional weight to compensate for gaps in the business evidence.
The bounded conclusion is stronger than an enlarged biography. Wexler is formally named on a patent that describes a technically coherent solution to double-sided board probing. His name appears with Smith's, which means the invention should not be narrated as a solitary achievement. Automated Test Engineering is the assignee, which anchors the work in a company context. Beyond those facts, titles and later roles require explicit attribution and do not change what the mechanism demonstrates.
What the fixture record contributes to manufacturing history
The patent captures a transition in the physical demands of board test. As components moved toward surface mounting and useful electrical points appeared on both faces of populated boards, the test interface had to move beyond simple underside access. The response documented here was not merely denser probing. It was a coordinated fixture that managed board loading, sealing, force distribution, opposed motion, probe alignment, user access and retraction.
Its engineering argument is built from dependencies. Vacuum supplies the lower stroke and distributed force. The device-under-test completes the lower seal. Pushrods carry movement upward. Rack-and-pinion actuators reverse its direction. The upper plate translates only after the clamshell is closed. Spring-loaded probes meet patterned test points from both sides. Return springs and vacuum release clear the contacts for unloading.
No single component explains the mechanism's relevance. The operational result comes from timing them together. If the probes extend during loading, the board can interfere with them. If the upper plate follows the hinge arc, the contacts can approach laterally. If the pushrod gap varies, upper travel changes. If top access requires a vacuum cover, interactive adjustments become difficult. The patent addresses these issues through the architecture of the entire fixture.
That is why repeatability is the most useful lens for understanding Wexler's documented work. It avoids claiming a market outcome the record cannot show, yet it does not reduce the patent to a historical curiosity. Manufacturing test depends on a repeatable relationship among the board, probes and instrument. The clamshell design is evidence of engineers treating that relationship as a system worthy of invention.
The surviving documents cannot say how many fixtures were built, which products they tested or whether customers adopted the exact embodiment. They cannot quantify yield improvement, reduced stoppage or probe life. Those are meaningful unanswered questions, and they should remain unanswered rather than being inferred from patent issuance.
What can be said is precise. In 1993, Wexler and Smith filed an application assigned to Automated Test Engineering for a two-sided fixture that coupled vacuum-driven lower contact with mechanically driven upper contact. In 1995, the patent was issued. Its description identifies the board-technology change that made two-sided access necessary and explains how perpendicular, coordinated motion could protect alignment while leaving the top side comparatively accessible.
This record places Don Wexler in the hidden manufacturing layer between an assembled circuit board and the decision made by automatic test equipment. The evidence is narrow, but the problem is fundamental. Before a system can judge a device-under-test, a fixture has to meet it in the same way, at the right points, cycle after cycle.