Simulation gets real

Cambridge, MA --No plan survives first contact with the enemy. This military maxim applies equally well to engineering, where a corollary might read: No design survives first contact with reality. In either case, planners strive to incorporate more real-life variables into their models, if only to avoid uttering those famous last words, "It looked good on paper."

The connection between the arts of war and engineering is not lost on Warren Katz, co-founder of MAK Technologies, Cambridge, MA. A veteran developer of interactive simulations for the Department of Defense, Katz is working to incorporate computer-aided design and engineering technologies into simulations of unprecedented detail and realism.

"Current simulations resolve the interaction of entities with rules," Katz says. "We want to resolve them with physics."

For example, if a tank fires at another tank in a conventional simulation, the round has a certain probability of hitting the target based on programmed variables, such as terrain, relative motion, visibility and weapon accuracy. If a hit is obtained, rules determine what the results of that hit are by calculating the round's penetration value against the target's armor rating. Generally, the target unit can be eliminated, forced to operate at some pre-determined reduced capacity, or is unaffected.

Such outcomes are fine for wargames, where the goal is to approximate battlefield results. However, Katz hopes to be able to move simulations into the realm of design tools by substituting appropriate engineering software for rules-based outcomes. In the engagement above, the tanks would be CAD assemblies, with each component modeled individually. The hit would be resolved using a shock-impact physics code, such as MSC/Dyna. Boundary conditions would be determined from parameters extracted from the simulation engine. Actual damage would be calculated, and the assembly model of the target updated to reflect the damage with new kinematics based on component deformation.

"This way, you would know what happens to the turbine blades of an engine when a depleted-uranium anti-tank round hits hull encasing the engine compartment," Katz says. "You might want to beef up the armor in that area or select different materials or layouts."

Game, not play. The modern military simulation is descended from the "kriegspiel" of the Prussian general staff in the 19th century. That innovative and efficient body determined much of the fog of war could be dispelled with markers representing military units that officers maneuver on a map of a hypothetical battlefield. Rules defined the constraints of unit actions, such as how far a battalion in column can travel down a road in the rain, and the outcome of events, such as an infantry rush at enemy breastworks.

Despite the ironic association of war with game, planners used the kriegspiel to design campaigns and engineer strategies and tactics. The value of the Prussian general staff system was aptly demonstrated in the wars against Austria in 1866 and France in 1871. However, the danger in allowing the simulation too much gravity was revealed when the timetables of the infamous Von Schlieffen Plan drove Germany, and Europe, to ruin in the First World War.

A simulation is only as good as its rules. If the rules do not reflect reality, then the simulation is diminished as a design tool. Worse, the simulation may produce spurious results that are misinterpreted as realistic outcomes. These "artifacts," as they are called, can lull designers into a false sense of security and are a reason skilled analysts are required to scrutinize the results of complex simulations.

One way to remove artifacts is to take the job of simulating human behavior away from the computer. Computer opponents in military simulations are notoriously predictable, once an experienced human participant divines their strategies. Simulator developers often found that trainees were playing against the simulation's artificial intelligence (AI) rather than operating their equipment in a realistic manner. Such situations can actually produce negative training effects, as a real enemy in the field cannot be counted on to act as dumb as the computer.

MAK has, somewhat stealthily, created the standard method the DoD uses to link its simulators together, enabling mock-engagements with thousands of participants. The Distributed Interactive Simulation (DIS) evolved out of the Defense Advanced Research Projects Agency (DARPA) SIMNET (SIMulation NETworking) program expanding single task trainers into networked team trainers. SIMNET has produced over 300 networked simulators with the technology that was to develop into DIS.

DIS the enemy. As Katz explains, the foundation of DIS is a standard set of messages and rules, called Protocol Data Units (PDUs), used for sending and receiving information across a computer network. The most common message is the Entity State PDU which contains data about an entity's position and velocity. By using the position, velocity, acceleration, and rotational velocity data, a receiver is able to determine a vehicle's position before the arrival of the next PDU, thereby reducing consumption of network bandwidth. This enables most of the processor-intensive functions, such as model rendering, to be handled locally with only behavior data passing over the network.

Using this technique, DIS is able to limit the amount of data an average simulator transmits to approximately 250 bytes per second. Optimizations, such as dead reckoning, permit very large virtual battles to take place. The largest DIS exercise, part of DARPA's Warbreaker program, had 5,400 simulated entities interacting in a single DIS virtual world.

The DoD's Defense Modeling and Simulation Office is seeking to establish a common technical framework to improve interoperability between the wide spectrum of modeling and simulation applications. This common technical framework includes MAK's High Level Architecture (HLA) networking standard that will be a follow-on to DIS.

The HLA defines a set of rules governing how simulations, now referred to as federates, interact with one another. The federates communicate via a data distribution mechanism called the Runtime Infrastructure (RTI) that lets different types of systems interact. These systems can include high-end engineering models that run much slower than real time and simulate individual subsystems with very high accuracy.

According to Katz, the updated HLA standards will permit engineering tools to be incorporated into complex simulations with thousands of accurately modeled entities. The fallout for engineers is that such simulations may soon provide a much more realistic framework in which to analyze designs.

It's not all serious business. Inevitably, perhaps, MAK's simulation networking technology is leading to more sophisticated methods of killing time. MAK is working with partners in the computer gaming industry to develop a new generation of realistic games that can be played alone or over the Internet. These games promise accurate shell trajectories, vehicle dynamics, and explosions. One game, in fact, is being developed in concert with the DoD as a training aid for the Marines. No doubt, the Prussians would have loved it.

Collision course. Networking is just one part of the simulator problem. Rendering complex assemblies in anything approaching real time is challenging for the largest computers. Current practitioners of simulation-based design tend to be restricted to fly-through-type evaluations of static virtual mock-ups in dedicated locations due to the processing resources required.

However, competing graphics standards initiatives spearheaded by Silicon Graphics Inc. and Hewlett Packard Co. promise to speed rendering operations dramatically. Both architectures, SGI's Open GL Optimizer and HP's Direct Model, feature occlusion culling techniques for streamlining graphics processing. In occlusion culling, the rendering engine does not process those components in an assembly that are hidden from the viewer's perspective.

Another important relevant trend is the growing availability of multi-processor systems for the engineering desktop. Such systems enable applications that support multi-threading to farm out operations to whichever processor has available capacity. Even conventional applications can be made to run faster. All major Windows- and Unix-based workstation vendors have multiprocessor computers on the market or in development.

Collision course

Networking is just one part of the simulator problem. Rendering complex assemblies in anything approaching real time is challenging for the largest computers. Current practitioners of simulation-based design tend to be restricted to fly-through-type evaluations of static virtual mock-ups in dedicated locations due to the processing resources required.

However, competing graphics standards initiatives spearheaded by Silicon Graphics Inc. and Hewlett Packard Co. promise to speed rendering operations dramatically. Both architectures, SGI's Open GL Optimizer and HP's Direct Model, feature occlusion culling techniques for streamlining graphics processing. In occlusion culling, the rendering engine does not process those components in an assembly that are hidden from the viewer's perspective.

Another important relevant trend is the growing availability of multi-processor systems for the engineering desktop. Such systems enable applications that support multi-threading to farm out operations to whichever processor has available capacity. Even conventional applications can be made to run faster. All major Windows- and Unix-based workstation vendors have multiprocessor computers on the market or in development.

See a video of this technolgoy by clicking here.

Set-top box to convert HDTV signals

Fremont, CA --AITech is developing an HDTV set-top box receiver that will meet new broadcast standards established by the Federal Communications Commission (FCC). Known as ATSC Digital Television Standard (ADTS), it is part of the FCC's approval this spring of digital broadcasting.

AITech plans to deliver the HDTV receiver by the end of next year to meet the FCC's schedule for ADTS broadcast. The company estimates the price at around $400--a fraction of the expected $5,000 cost of a new HDTV set.

"AITech's receiver represents a product category that will be the highest volume digital receiver in the consumer market," says Jonathan Dassell, an industry analyst with San Jose-based Dataquest.

Virtually every home in the U.S. has one or more analog televisions. When HDTV sets are available, they will display at a resolution of 1,92031,080, but, at $2,000 to $5,000 will likely be out of the average consumer's price range.

AITech's HDTV set-top box will receive high-definition digital broadcast signals from the airwaves or from cable, decode the signal, and convert it to a format that can be displayed on any analog TV or standard VGA PC monitor. The unit will also include an audio decoder to perform digital AC-3 decoding into 6 channels of multi-lingual or surround-sound audio.

If the consumer already owns a high-quality TV with S-VHS inputs, the receiver will provide up to 30% more resolution than analog cable or terrestrial broadcast programming.

"The digital broadcast industry has enormous growth potential," says AITech's Casey Ng. "AITech's HDTV receiver will make digital TV affordable for all, which will connect broadcasters to the viewing audience they need."

Plans for the set-top box draw on many of AITech's existing technologies including: FlicFree^TM filter and patented VSPro^TM PC/TV scan conversion.

Faster chip handles complex applications

Palo Alto, CA --Echelon Corp.'s new 20-MHz version of its Neuron chip doubles the performance of previous Neuron chips, say company officials. It supports faster response times of 3 to 4 msec across a LonWorks control network and doubles the I/O performance over previous versions.

The new chips further expand the range of high-speed industrial applications for LonWorks networks in variable-speed drives, motor controls, conveyor control, valve supervision, and motion control. Engineers can use the faster chip as an off-the-shelf component to control complex manufacturing processes in environments ranging from clean rooms to textile factories to breweries.

Motorola and Toshiba are offering two different versions of the 20-MHz part. The Motorola part contains 10 kbytes of ROM, 2 kbytes of EEPROM, and 2 kbytes of RAM. The Toshiba part has 10 kbytes of ROM, 1 kbyte of EEPROM, and 1 kbyte of RAM.

Standard features include: three 8-bit pipelined microprocessors, 11 programmable I/O pins, two 16-bit timer/counters, sleep mode, media-independent network communications port, 7-layer LonTalk(R) protocol, unique 48-bit internal Neuron ID, and built-in low-voltage detection.

Both parts are sampling now and are pin-compatible with existing parts in the Neuron 3120 family.

Display design hinges on hinge

Stirling, NJ --Retail is a competitive environment, and companies that put their products on display want potential customers to have a clear view of those products. That's why Thermoplastic Processes Inc. improved the design of its familiar display boxes and bins with the addition of a clear, non-mechanical integrated hinge.

"Clear materials are critical to the display business," says Wes Wheeler, executive vice president of Thermoplastic Processes. "Our objective was to develop an alternative to conventional opaque hinges."

Thermoplastic Processes found that alternative in a coextruded elastomeric substrate sandwiched between two PETG (glycol-modified PET) profiles. Wheeler calls the solution the Excelon^TM hinge. The design is based on SPECTAR copolyester supplied by the Eastman Chemical Co. (Kingsport, TN).

Typically, the display industry uses sheet stock and profiles made of the copolyester for a majority of its display components. "We recognized that the combination of clarity, toughness, and flexibility that made SPECTAR perform so well in other parts of the displays could serve us well for the hinge design as well," Wheeler explains. "Lids get a lot of rough treatment, so the hinges have to be as strong and durable as the unit itself."

Using the one-material system enabled Thermoplastic Processes to offer its customers a completely compatible, lighter, more streamlined, and better-looking product that provides a clear view of the product on display. Moreover, the Excelon hinges can be mechanically or solvent bonded to the sheet components. "These crystal-clear hinges look so good you don't even know they are there," Wheeler adds.

Simulation helps work out design of exercise equipment

Franklin Park, IL --By using Working Model 2D motion simulation software (Working Model Inc., San Mateo, CA) to analyze exercise equipment on a computer, Life Fitness is reducing the costs of physical prototyping, shortening design cycles, and bringing high-quality equipment to market more quickly.

In the past, it took more than a year to finish a series of costly physical prototypes. Now, with Working Model, engineers can visualize and analyze dozens of different variations of mechanisms in two weeks.

Recently, mechanical engineer Chuck Rosenow used Working Model to design the Cross-Trainer, exercise equipment that gives a total cardiovascular body workout through simultaneous movement of the arms and legs.

He focused on two elements an exerciser manipulates: the handlebars and pedals. Rosenow first used Working Model to generate rough shapes for the major parts of the Cross-Trainer. Then he produced linkages for these parts, established bolted and welded joints, and applied forces to simulate the exerciser. He also input the density for each of the links.

Once the model took shape, Rosenow used AutoCAD software (Autodesk Inc., San Rafael, CA) to draw the actual parts and create a detailed 2-D drawing. He employed AutoCAD with a transparent interface to Working Model by using AutoMotion (Working Model Inc.) to export the CAD file into Working Model for simulation.

While running the simulation of the model, Rosenow traced the motions of the handlebars and pedals and quickly checked to see if they were within the appropriate dimensions to provide effective, comfortable exercise.

To analyze stresses at the joints, he took force data from the simulation model, combined it with the geometry from the AutoCAD drawing, and imported the information into Cosmos/M finite element analysis software (Structural Research & Analysis Corporation, Los Angeles). He then ran an analysis to ensure that the links could indeed handle the stresses placed on them by the exerciser. Once the overall design gained departmental approval, Rosenow used AutoCAD to generate detailed engineering drawings for manufacturing.

"We couldn't have achieved the design for the Cross-Trainer without Working Model," remarks Rosenow. "It enabled me to look quickly at over 100 linkage simulations without needing to build physical prototypes, greatly increasing our speed to market."

Miniature motors make mincemeat of Mars

by Mark A. Gottschalk, Western Technical Editor

Seal Beach, CA --Space vehicles are typically thought to push the limits of design like few other engineered objects. So it might come as a surprise to discover that this summer's media darling, Sojourner, perambulated about the red, dusty surface of Mars under the power of eleven nearly stock, commercially available electric motors.

Supplied by Maxon Precision Motors (Burlingame, CA), the identical model RE016 DC graphite-brush motors drive each of the rover's six wheels, provide steering control to the four corner wheels, and power the deployment mechanism for the onboard alpha-proton X-ray spectrometer. Each weighs 38 grams and is 16mm in diameter and 41mm long. Rated at 3.2 watts output, the motors use rare-earth neodymium magnets to attain 86% efficiency.

The motors drive through five-stage, 2000:1 ratio, planetary gearheads manufactured by Globe Motors (Dayton, OH). Engineers at NASA's Jet Propulsion Laboratory (JPL) in Pasadena, CA, modified the gearheads to cut weight and friction by slimming some of the gears and replacing sleeve bearings with ball bearings. The four steering mechanisms receive additional 2:1 ratio right-angle gearsets.

Torque per unit mass was a big design driver, say JPL engineers, but the biggest was survivability in the Martian environment--which makes the selection of a brush-type motor all the more interesting. "Normally you couldn't use brushed motors in the Martian environment," says Howard Jay Eisen, chief mechanical engineer for Sojourner at JPL. "The graphite brushes would just fall apart in the near vacuum due to arcing."

The RE016 motors, however, incorporate a proprietary Maxon technology called Capacitor Long Life (CLL) that "dramatically improves brush life," says Eisen. It consists of a capacitor placed in parallel with the winding segments of the motor. Energy stored by the natural induction of the windings is then dissipated by the capacitor rather than in an arc at the brushes.

Brushless servo motors were considered as well. But their extra electronics and wiring made them less desirable. The rover is insulated, and it couldn't afford the conductive heat loss down additional wiring.

Though nearly stock, the RE016's were slightly modified to address the Martian environment. Changes include:

Castrol Braykote grease--good to -80C--was used on the commutator segments. Engineers specified a 10% Braykote/Freon solution for the motor bearings, followed by a mild bake out.

JPL engineers selected a single motor type partly to hold down development costs. In all, they ordered 110 of the roughly $100 motors, and spent just $220,000 to adapt them for the application. They proved so spaceworthy that Pathfinder lander's airbag-retraction mechanism used RE016s as well.

"The RE016 is a fantastic motor," says Eisen, "and with some minor changes we got a motor that really did the job for us."

Light switch has its say

San Jose --A light switch that talks--that's the latest application of ISD's ChipCorder(R) voice record and playback chip. The MemoSwitch^TM light-switch plate cover from Logic Labs (Leesburg, VA) records and plays back a 10-second message whenever the switch is turned on or off.

Pressing a button on the MemoSwitch activates the ISD circuit to record a message. Because the chip integrates automatic gain control and filters, it is sensitive enough to let people record messages just above a whisper.

"Perhaps your grandfather doesn't like to use answering machines or your child can't read a hand-written note yet," says Bob Smallwood, Logic Labs' president.

One of the biggest applications may be a kind of electronic string on a finger. "Sometimes it's even hard to remember to check the answering machine or a message board," says Smallwood. Since turning lights on and off is an involuntary habit, Memo-Switch is a simple solution for sending quick messages to yourself, coworkers, or members of the household."

ISD's patented ChipCorder technology is a single-chip solution for voice recording and playback in a variety of consumer, communications, and industrial applications. It uses a proprietary "multilevel" storage methodology in which one of more than 250 distinct voltage levels is precisely stored per memory cell, providing approximately eight times more storage space for any given memory size than the alternative two-level, digitized signal storage technology.

ChipCorder requires no external D/A or A/D circuits or software, and stores voice data directly--that is, with-out using compression or encoding.

16-bit controller gains DSP

San Jose --SGS Thomson Microelectronics has expanded its ST10 microcontroller family with two new chips that handle digital signal processing (DSP). The chips suit embedded applications that require high computational capability but without the costs associated with using dedicated DSP chips.

The ST10R262 and ST10F262 both feature a built-in multiply-accumulate unit (MAC) and a pulse-width-modulation (PWM) circuit. These added features make the chips useful for high-performance servo control operations in hard-disk drives and DVD-ROM players, which typically use more-expensive DSP-based circuits, says the company. The ST10F262 also offers 64 kbytes of on-chip flash memory.

All ST10 family members are based on a 10-MIPS core that features a 4-stage pipeline and a register-based architecture. The two new chips will be available by the end of the year.

Pliers won't corrode or conduct

Pearland, TX --Ever open your fishing tackle box and reach for those trusty pliers to extract a fish hook from the jaws of a "keeper," only to find that they have rusted shut? Plastix Technix says you won't face that problem in the future if you buy a pair of their pliers made of a long-glass-fiber-reinforced nylon 6/6 structural composite. And, if you drop them in the water, they float.

"The environment for fish pliers can be highly corrosive,'' says Plastix Technix's Allen Groseth. "Anywhere you have water, you're going to have corrosion. Even stainless steel will corrode around salt water.''

What Groseth needed was a material that would be durable, yet not affected by salt or fresh water, or fuels. He found that material in Verton RF supplied by LNP Engineering Plastics (Exton, PA).

Groseth looked at other materials, including polycarbonates, before deciding on the structural composite. What he needed was a material that would not only solve the corrosion problem for his 10-inch, needle-nose, fish-hook pliers, but would also do double duty in other industries.

"We wanted a product that was versatile enough to serve the electrical and aviation industries," Groseth explains. "This meant we had to have a structural composite that combined strength and durability with non-sparking and non-conductive electrical properties.

When a technician reaches into a confined space to access wires or components, he needs to feel confident that his grasp has no potential for conductance or sparking. "This uncompromising requirement could only be achieved with Verton RF structural composites," says Groseth.

Ergonomic chair no 'shell' game

Grand Rapids, MI --Kimball International's new Purpose office chair features a shell that consolidates a back rest, seat, and flexible support beam into one gracefully contoured molded part. The shell's ribbed-channel design not only increases the chair's long-term load-bearing capabilities, but lets the user stretch, turn, and recline in comfort.

By incorporating the flat, ribbed channel into the shell's midsection, the shell does not need a metal support beam, adding to the chair's sleek profile. Moreover, the channel distributes stress loads equally across the shell, while allowing controlled flexibility. Ribbing in the upper back-rest area provides added back support and minimizes deflection. And, by consolidating the back rest, seat, and support beam, the design eliminates a complicated assembly operation, as well as many of the metal bolts and connectors needed in conventional chair construction.

Making the design possible is the use of a black, glass-fiber-reinforced, impact-modified nylon 6. The Ultramid(R) B3ZG6 Q660 nylon, supplied by BASF Plastics (Mt. Olive, NJ), combines high flexural strength with elongation and impact properties. The material meets American National Standards Institute/Business and Institutional Furniture Manufacturer's Association performance standards.

Key to Kimball's selecting Ultramid is the material's ability to mold shells that meet the industry's Back Pull Test. Here, a minimum load resistance of 300 lbs is required, and the chair must withstand a 300-lb drop impact test. Kimball reports that Ultramid's high tensile strength (14,500-21,000 psi) and elongation properties outperformed other materials it tested.

Ultramid also contributes some other benefits to the chair's design. For instance, the material provides the high melt flow characteristics (520-555F) needed to efficiently fill the long, thin mold cavity used to make the shell. Further, the as-molded black shell's surface requires no painting or other secondary treatment.

The back and seat cushions attach to the shell by a snap fit and keylock fastener technique. The chair arms are mechanically fastened to a metal arm strap, which also secures the shell to the base and chair control.

BASF provided the office furniture maker with close technical support throughout the chair's development. Assistance included: structural analysis work in designing the shell; flow analysis to optimize the physical properties; processing aid during molding trails; and product modifications to optimize the material's elongation properties and surface finish.

Embedded apps go desktop

Austin, TX --For the first time, Motorola has optimized a Power-PC microprocessor for embedded applications.

The EC603e RISC microprocessor offers clock speeds of 100 to 200 MHz and dissipates 3.2W at 166 MHz.Starting price is $20.69 for 10,000-unit quantities of the 100-MHz version. The chip is functionally identical to the Power PC 603e, except that the floating-point function is not available.

Motorola designers say the 32-bit EC603e suits such applications as internetworking, telecommunications, industrial control, imaging, multimedia, and consumer goods.

"The high-end embedded processor market is growing at a compound annual growth rate of 30.6%," says Tony Massimini, chief of technology at Semico Research. "It is a market that demands ever-increasing performance but with lower power consumption and at lower cost. Motorola's EC603e is meeting these requirements."

Motorola plans to modify other PowerPC microprocessors introduced in desktop computers and migrate them to embedded systems.