Pentium flas could spark scramble
for bug-coping strategies
Experts advise cross-checking critical designs using different processors and software. The tradeoff? Time to market.
Newton, MA--By now, engineers worried about potential Pentium division errors have contacted Intel and swapped their flawed chips for new ones. But while that problem may be "solved," the Pentium flap has raised a broader question: Are other chips or software also making errors that some mathematics professor may not catch until next year? Should engineers consider strategies for overcoming potential flaws, including delaying time to market while checking a design one more time?
"Microprocessors have become so complex that it is no longer possible to completely debug them, or even to determine every bug that exists in one," according to Dr. Thomas R. Nicely, the mathematics professor at Lynchburg College, VA, who discovered the Pentium flaw. His advice: Perform "mission-critical" computations multiple times-preferably with different CPUs, operating systems, and software algorithms.
"This is a direct assault on speed to market," says Michael Schrage, a research associate for the Sloan School at MIT who also writes the Innovation column for the Los Angeles Times. "But you retain no market-share advantage if you're being sued for $250 million by 10 of your largest customers for a flawed product."
The potential for hardware flaws shouldn't be overstated. After all, chips and software go through multiple quality-control checks before hitting the market. Still, caution is advisable. "There's probably a greater chance of error in formulating the problem and writing the code than there is in a microprocessor," says Rich Partridge, a microprocessor market analyst with D.H. Brown Associates, Port Chester, NY. "But they're all tied together. If you can't trust anything, what do you do? You do cross checking."
Pandora's box. The catalyst for this uneasiness is Intel Corp.'s Pentium microprocessor. Its floating-point unit had a flaw that caused it to return reduced-precision results for division involving certain numerators and denominators. The flaw was due to five missing entries in a 4,000-entry look-up table the iterative algorithm uses to perform the divide instruction.
Intel officials say that for 1 in 9 billion possible divides, any digit from the fifth digit on could be incorrect. However, the error could be magnified by multiplying the result by a large number or subtracting two numbers that have been divided.
The impact of the flaw varied by the rate of use of floating-point instructions, the input data fed to them, the use of the results in further calculations, and the accuracy needed. One of the most intensive uses of floating-point math is in engineering software.
"CAE users need to be jolted into the realization that computers are not infallible," says Charles Foundyller, president of CAE market research firm Daratech in Cambridge, MA. He doubts that engineers will be doing CAE on multiple platforms because of the time involved, and advises moving to the latest computer technology with caution.
Schrage predicts that companies will be sharing more hardware testing data and publishing the results. Disclosures from such companies as Boeing, Ford, GM, and HP on their own tests would help speed the discovery of hardware and software bugs.
Validating designs. Testing is an increasingly critical issue for microprocessor designers and manufacturers. Partridge says that as chip densities climb, designers will be devoting more internal hardware to detecting and correcting errors. In 18 months to two years, we may see more chips with error-correcting caches, residue checking, or instruction retry.
Such features would catch errors due to manufacturing problems, but not design problems. For example, the Pentium had a basic design problem that required extensive testing to catch. "Our ability to successfully test microprocessors has been outstripped by the product's inherent complexity," notes Schrage. Adds Partridge: "You won't find every problem using brute-force testing of trillions of combinations. You have to have a more rigorous approach where the engineer who designed the chip needs to defend the mathematics involved in how he or she has implemented the logic."
As a result of the Pentium embarrassment, companies may take the time to do such design walkthroughs-even though that may delay microprocessor introductions. Walkthroughs will be especially crucial with superscalar processors, such as RISC chips, which can execute more than one instruction at a time. These chips are ripe for instruction sequencing problems that would be much more difficult to find-no matter what the method-than the Pentium flaw.
It's not just the chip. Dr. Julian Palmore, principal investigator at the U.S. Army Construction Engineering Research Laboratories (CERL) in Champaign, IL, and professor of mathematics at the University of Illinois, points out that errors caused by a processor chip are only one way computer simulations can become flawed. "In any discipline using computer models, especially where safety is concerned, it's absolutely essential to verify and validate the model using exact arithmetic."
That means writing software that is machine-independent for calculations. For example: In statistics, researchers use exact arithmetic by employing an integer-based random-number generator, so there's no rounding.
Computer hardware and software can introduce design errors.
For critical work, devise a strategy to offset the chance that your design work will be compromised.
You'll have to make tradeoffs between design cross-checking and speed to market.
Palmore's work at CERL on battlefield simulations has revealed many specific opportunities for errors in timing, interfacing, and computations. "Depending on how sensitive a process is, such as in an aircraft controlled by very small signals, these errors can have devastating consequences."
A real-world example of such impact occurred during the Persian Gulf war, when a Patriot system failed to engage a SCUD missile. A U.S. General Accounting Office report found that the computer's binary arithmetic, which introduced small errors over about 100 hours, resulted in a timing mismatch between the computer program and the radar system. The 1/3-second calculation error caused a shift of about 700 meters in the radar range gate, which was enough to make the system fail to recognize the incoming SCUD as a missile to engage.
Warns Palmore: "Errors occur and accumulate when you have any sort of algorithm that's being used iteratively, with the output from one calculation feeding the next. You have to watch out for and understand how errors affect the calculation."
Of course, the ultimate quality-control check is engineering training and common sense. "You can analyze things to death," says Bob Riccomini, lead engineer on Apple Computers' PowerBook 270c. "You have to go with your engineering judgment."
-Deana Colucci, New Products Editor
Hannover aids search for European products, customers
As competition in U.S. industries increases, more and more companies are taking their products to the international marketplace. And with good reason: The European Community's single-market economy represents more than 345 million consumers and $6 trillion of purchasing power. But for would-be customers, finding the right products can be both confusing and time-consuming.
One place to turn: The Hannover Fair tradeshow in Hannover, Germany. Europe's leading industrial trade show, it covers nine sectors ranging from factory equipment and automation to environmental technology.
Last year's show hosted some 6,800 exhibitors from 60 countries and filled more than 12 million square feet of space. Nearly 400,000 attendees traveled from 110 countries in Europe, the Americas, Asia, and Africa.
In addition to the individual company exhibits, attendees can visit special USA Pavilions and group exhibits organized by region or association.
"The show's organization and planning are first-rate," says Bill Taylor, vice president of sales and marketing for Jeffrey Chain Corp., Morristown, TN. "We made good contacts, and we're very happy we went."
Snorkel solves aerial balancing act
St. Joseph, MI--Balance is critically important to Snorkel Economy, a maker of self-propelled aerial work platforms. Their devices move workers in the air up to four-stories high-without using outriggers, special extensions, or braces. "Our machines have to be stable in any position of articulation," engineer Dave Engvall explains.
During one recent project, Engvall analyzed an articulated boom with 27 junctions known as pin joints, where sections of the boom connect and move. Two aspects of the design were crucial: stability of the overall vehicle, and the load on each pin joint.
Engvall imported his CAD files into Working Model, a dynamics/kinematics analysis package from Knowledge Revolution, San Mateo, CA. To find the most unstable aspect of his design, he focused on the lifting structure-the boom, engine, and counterweight, all set on a rotating turntable.
The hardest balancing problem in the design was side tip stability, he says, with the boom extended at a right angle to the vehicle. The key point of balance or "tipping fulcrum' in this position was the tire. In a few seconds, a simulation revealed at what point the mobile lifter was stable: When the force on the separator was less than zero, the machine tipped over; when it was greater than zero, it didn't.
He also used the software to measure the load stress or static force balance on individual pin joints. "If I did the static force balance analysis myself, I would have to describe a free body diagram of each component, write equations for the translations and rotational elements for each link, and solve them all simultaneously."
Instead, he used Working Model to assign the proper weight to each component, put a pin joint at each location so the components could move, and let the simulation run.
The savings? It used to take him more than a week to create models and do static load calculations. Now he finishes those tasks in a few hours.
Composite cuts weight in multi-function printer
Dayton, OH--Fax/telephone/modems. Photocopier/scanner/faxes. Analysts predict a new surge of these "one-man-band" products for 1995, especially in the electronics and computer marketplaces. With that in mind, engineers are struggling to produce multi-function machines that are not large and bulky.
Monarch Marking Systems recently entered this foray with a portable four-in-one printer. The hand-held thermal bar code printer-which includes a label applicator, data collector, and laser scanner-is the only one of its kind on the market, says Tom Keller, senior mechanical design engineer at Monarch.
Keller and his team examined several materials options to reduce the printer's size and weight. They settled on Lubricomp RCL, a carbon-reinforced, internally lubricated nylon 6/6 thermoplastic composite.
"Especially in portable products, weight and ergonomics have become so important, you have to get as much function as you can out of each part," Keller explains. "So, the more characteristics you can squeeze from a given material, the more you can do with that particular part."
Lubricomp's stiffness and strength enabled the printer's drive frame to become the central structural skeleton of the unit, with the outer housing bolted over the top.
The composite also gave the drive frame the rigidity, dimensional stability, and lubricity Monarch sought. The printer, weighing about 38 oz, can survive ten drops from one meter onto concrete in various orientations.
"We also needed good wear resistance," Keller says. "Because two posts in the drive frame support rotating gears, we chose a material with PTFE lubrication."
Using Lubricomp also resulted in a substantial cost savings, according to Keller. "We were able to cut down considerably on weight and volume. After all, those are the two most important criteria when designing a portable unit."
Inhaler gets an upgrade with CAD
Madison, NJ--Pocket inhalers, those small devices used to administer metered medical dosages for treatment of asthma and other bronchial ailments, seem uncomplicated to users. However, assuring their reliability can create a problem for their designers. So, when Schering-Plough Corp. decided to redesign an existing model, it wanted to improve performance as well as embellish the inhaler's look.
Three parts make up the inhaler. They include: a clear housing that holds the canister of medication, an actuator to administer the required dosage, and a dustcap that fits over the nozzle spout to protect it from outside contamination.
"In addition to aesthetics, we wanted to use the redesign as an opportunity to improve the snap-fit between the canister and the actuator," explains Keith Bishop, Schering-Plough's associate principal scientist. "This would eliminate any chance of an inadvertent separation of the two parts-something that had happened with the old design."
To help in the redesign, Schering-Plough called on Phillips Plastics Corp., Prescott, WI. During preliminary design stages, Schering-Plough fed Phillips' CAD designers with concepts used to create a design database. Phillips CAD Engineer Mark Sponsel reports that Pro/ENGINEER (Pro/E) solids modeling software helped model the parts in CAD.
"Pro-E is a perfect fit where no existing design database exists," says Sponsel. "It allowed us to perform interference checks and formed a perfect foundation for stereolithography (SL) work needed to prove out elements of the design."
Improving the fit of the housing and actuator was achieved by incorporating a proven snap-fit design into the inhaler. "Incorporating that snap-fit ultimately saved design time and aided in the SL process," Sponsel adds. It also enabled the use of a two-cavity production tool to build the final eight-cavity production mold.
The redesigned inhaler incorporates several subtle design refinements. Among them: the addition of textured grips in the dustcap, and the placement of the company's logo into the part. The snap-fit design provided the appropriate degree of resistance, alleviating virtually all separation concerns. Moreover, part consistency and tolerances between the canister and actuator were improved. Adds Bishop: "The result is a far better product in today's marketplace."
Intake manifold wins SPE Grand Award
Detroit, MI--A thermoplastic composite intake manifold on the 1995 Cadillac Northstar V-8 engine walked off with top honors in the 1994 Society of Plastics Engineers' (SPE) Grand Award for the "Most Innovative Use of Plastic." The component also won the powertrain award category.
Production of the manifold makes use of the lost-core manufacturing method. It allows several pieces to be integrated into the same molding. The result: elimination of more than 80 components required in the earlier design.
"Along with the performance benefits of increased power, as well as improved idle stability and driving range, the thermoplastic manifold significantly reduces part complexity and mass, which simplifies manufacturing and assembly," reported Sam Winegarden, chief engineer of the Northstar engine, in accepting the award. "Moreover, the nylon 6/6 material can be recycled."
Freudenberg-NOK's Plastic Products Div., Plymouth, MI, designed and produced the manifold. It consists of glass-fiber-reinforced Ultramid nylon from BASF. The 6/6 nylon was specially formulated to resist engine temperatures and attacks from oil, fuel, and underhood fluids.
The manufacturing process required eight metallic intake runner cores die-cast from a soft solder-type material. The cores are locked on a mandrel, set into a plastic injection mold, and the nylon 6/6 injected into the mold.
The mold and cores are immersed in a hot oil bath to melt the cores, creating hollow tubes inside the manifold. This results in very smooth inner walls for optimum air flow and distribution. The low thermal conductivity of the nylon also serves to insulate the air inside the manifold from engine heat to help increase air flow.
"The SPE recognition underscores the great contributions that plastic composites can make in solving the complex challenges automakers now face," says Robert C. Hange, senior vice president & general manager of Freudenberg-NOK's Plastic Products Div. "This means enhanced mileage and performance, while cutting manufacturing cost and reducing automotive noise."
Workstations help patients avoid surgery
San Jose, CA--With medical tools such as laparoscopes, doctors can perform minimally invasive procedures on patients, greatly reducing the need for extensive surgery. Stryker Endoscopy, a designer and manufacturer of medical equipment, knows how important it is to design better and smaller instruments.
To improve its product designs, the company used a computer system that ran 2-D geometry applications, but had poor 3-D wire frame capabilities. Engineers had difficulty detecting interference between parts, and their system was unable to effectively simulate the motions of instrument parts. Often, construction problems could not be detected until the prototype stage. And, Stryker had only one PC for CAD work, while all others were reserved for word-processing tasks. The result was a bottleneck and unacceptably long design cycle.
With a goal of reducing the entire product-design cycle by at least 40%, the company installed a network of SPARCstation workstations and servers. In addition, 18 PCs used for word processing and less demanding design tasks now share files with the SPARCstations through PC-NFS(R). Among the programs running on the system: Pro/ENGINEER, AutoCAD, MECHANICA, and Optical Research Associates' Code V software.
"Since installing the network," says William Chang, vice president of research & development at Stryker, "we have more than met our goals. In addition to reducing the overall design cycle time by approximately 50% and generating prototypes that work the first time, we've increased productivity by almost 70%."
Turbocharger bearing retainers take the heat
Tokyo, Japan--NSK, a large bearing manufacturer, wanted to lower costs and improve productivity in the production of turbocharger bearing retainers. However, machined components made from compression molded polyimide stock shapes proved too expensive. The solution: a switch to an injection-molded polyimide.
Polyimides resist high temperatures, wear, and solvents. Until recently, they also resisted conventional processing. For the most part, they required the fabrication of parts by machining stock shapes, or compression molding powders under high heat and pressure. Enter AURUM JCN 6230 polyimide from Mitsui Toatsu Chemicals, New York, which lays claim to being "the world's first injection-moldable polyimide."
Before making the switch to the injection-molded poly-mide, NSK put the high-performance thermoplastic to a 500-hr simulated under-the-hood test. According to Mitsui officials, the only modification to the processing equipment involves the need for a nozzle that can tolerate temperatures of 75 degrees to 79 degrees F.
Test conditions included: maximum velocity of 135,000 rpm, temperatures of about 300 degrees C, and use of a steel mating material with oil lubrication. Not only did the material pass the test, but it helped NSK lower production costs and increase productivity and moldability. The retainers also have a longer life at high temperatures in oil, according to test results.
Garrett Turbo, Inc., a unit of AlliedSignal, Inc., makes the turbochargers. You will find them in certain models produced by "a large Japanese automaker."
Protective coating keeps superconductors running
Wilmington, DE--Advanced superconductors are moving out of the lab and into practical applications, thanks in part to ultrathin protective coatings of amorphous fluoropolymer.
The coatings, made of Dupont Teflon(R) AF, are applied in a solution only 2 microns (0.08 mils) thick. They protect superconductive metal-oxide films against damage from traces in the atmosphere.
The superconductors, also products of a DuPont new business venture in superconductivity, can function in real-world conditions. Conventional niobium superconductors only work at temperatures near absolute zero. However, the metal-oxide films, pioneered by DuPont, become superconductive at temperatures as "high" as -143 degrees C (-243 degrees F). Cooling can be accomplished with liquid nitrogen or by using mechanical refrigeration.
With no electrical resistance, the "high-temperature" superconductors (HTS) provide the basis for ultra-efficient electronic devices. Among their potential uses: microwave-frequency, broad-band antennas; low-loss filters; and high-quality resonators and oscillators.
The superconductor bases consist of yttrium, barium and thallium, barium, calcium, and copper. They are deposited on wafer substrates of lanthanum aluminate or sapphire. Circuit patterns are etched in the films using the same patterning proccesses employed for semiconductors.
"Because our films are oxides, they can be damaged by traces of acid in the atmosphere, and, in many applications, have to be protected or passivated," says Daniel Laubacher, product development manager for DuPont Superconductivity. "We tried standard polyimide passivation coatings, but they didn't adhere well, and they picked up acidic moisture."
Superconductor devices using the new coating benefit from the low dielectric constant of Teflon AF: 1.89 to 1.93. This is the lowest dielectric constant for any polymer, Laubacher claims. In one HTS design, Teflon AF serves as an adhesive. Circuits now being explored will use the fluoropolymer in a resist layer for reactive-ion etching, and as a coating that makes other resist layers easy to remove.
Oxygen process boosts glass quality
Somerset, KY--Inject pure oxygen into a glass-melting furnace, and what have you got? A whole list of benefits, according to GE Lighting.
By providing better control over glass quality, oxygen fuel firing reduces the incidence of cord (an inclusion of glass of a different composition), and stones (crystalline inclusions). And the method being used by GE Lighting, 92% oxygen fuel firing, promises cleaner flue emissions.
There are other direct benefits. Since GE Lighting converted from regenerative air to the oxy/fuel process, melting capacity has increased by 20%, while gas use has dropped by half on a day-to-day basis. Counting gas and electricity, utility costs have decreased by $45,000 a year.
The inexpensive and accessible oxygen has completely changed the economics of glass melting, according to Dan Cico, GE Lighting's manager, glass marketing. GE had a number of systems to choose from in making its conversion. It finally settled on a vacuum swing adsorption (VSA) system built by Air Products and Chemicals. Because of its success, another GE glass plant switched to the process late last year, with two more plants scheduled to come on line by mid-1995.
Concept cars get serious at '95 auto show
Detroit, MI--What will you be driving in the year 2000? Concept cars at the North American International Auto Show propose future vehicles with interactive electronics and an emphasis on safety.
With climbing attendance rates and an ever-more-international audience, the annual NAIAS is on the grow. In January, representatives from some 30 countries attended the introductions of more than 40 vehicles.
The concept designs-typically a mix of pre-production and fantasy-favored realistic technologies. "In past years the concept cars were almost theatrical," commented one industry analyst. "This year, they're less show and more business."
Witness the Buick XP2000. With interactive electronics and eight airbags, this idea car explores what interior options might be most appealing to future drivers. For example, the XP2000 features a "smart-card" instrument panel and memory chip that would let drivers insert a credit card to automatically pay for tolls, fuel, food, and other services. The card could also store driver preferences to tailor automatic seat settings, transmission shift pattern, and suspension response.
The instrument panel includes a flat-panel color display suitable for future intelligent vehicle highway systems and a color heads-up display. The flat panel is linked to sensors that indicate diagnostics such as low air pressure in the tires.
Anxious to avoid a "showbiz" label, Buick General Manager Edward Mertz described the car as "a realistic preview of Buicks of the future." Some technologies that look promising for the near-term: a single high-intensity discharge lamp to deliver all interior lighting via fiber optics, as well as side and front airbags for front and rear passengers. The car's computers can be programmed to automatically dial local emergency services if an accident activates the airbags, say Buick engineers.
But eight wasn't the record for airbags. The Mercedes X-bag research car uses 17 airbags, including kneebags under the lower dashboard, rear-passenger airbags, side bags in front- and rear-door panels, B-pillar roof bags for side collisions and rollovers, and bags built into the head restraints to protect in rear-impact accidents. Although many are design exercises, some airbags from the X-bag, such as the side bags, will find their way into '96 production Mercedes. They are already offered in '95 Volvos.
In the X-bag concept car, sensors work with a computer chip to inflate whichever airbags are needed to protect the occupants. Engineers hope the car will be a precursor to an "anticipatory crash-analysis system" that would use electronic object recognition to establish the size and weight of an oncoming vehicle based on pre-programmed data. Engineers predict such a system would calculate impact angle and speed, as well as the expected severity of the collision, then trigger the appropriate airbag several centimeters prior to a collision-thus improving airbag performance.
Big ideas. In many stands, the theme was BIG. For example, at just under 17 feet in length, the Chrysler Atlantic concept car unabashedly recalls the heroic proportions and profiles of luxury coupes from the 1930s. Modern touches include an all-steel unitized body and neon stop and taillights.
At Ford, big was to the tune of a 6.0-l, V-12 engine. The design of the GT90 concept car, from the multi-beam, high-intensity discharge headlamps to the stainless-steel exhaust, was completed in six months. Although Ford has "no plans" to build a production version, Vice President of Design Jack Telnack pointed out that "because of extensive use of CAD tools and rapid prototyping, GT90 is very repeatable-unlike most show cars."
The GT90's deliberately un-retro styling includes unconventional lighting such as fluorescent stop lamps and gauge illumination, and ion-charged taillights. Equally unconventional is the mid-ship, turbo-charged, aluminum-block engine with a 720 horsepower rating. The car, which takes its name from the GT40 racer, boasts a zero-to-sixty acceleration of 3.1 seconds and a top speed of 235 mph.