“If computer-aided design [CAD] wasn’t so hard to use, took so much space, and sucked so much compute power, you wouldn’t need DMU because you could use CAD. But, you can’t, so that’s why there’s DMU,” sums up Charles Foundyller, CEO of the research firm Daratech Inc. (Cambridge, MA; www.daratech.com), speaking of the relevance of digital mock-up (DMU). Simply put, DMU is the computerized version, the 3D visualization, of parts and assemblies. DMU tools let users visualize, animate, dimension, manipulate, explode, fly through, cut away, and check the clearance and interference of part and assembly designs. And so much more.
With this functionality come changes. One of the more important changes, says John MacKrell, senior consultant at the Ann Arbor-based research firm CIMdata (www.cimdata.com), has been in moving “DMU out of the CAD environment, making it available to users who do not have a CAD system. This decoupling allows us to do a couple of really neat things.” First, whole new classes of users—non-CAD users—now have access to DMU capabilities. Second, it speeds design analysis. “Sometimes it’s good enough just to assign constraints and play with the model completely CAD-neutral environment without having real, honest-to-God analysis going on in the background.” Third, decoupling lets users apply DMU capabilities to part and assembly files created by different people on different systems—in a single environment, namely DMU. It also opens DMU to a wide variety of enterprise systems.
This reach-out-and-touch-someone or -something capability of DMU is well-incorporated in Seemage, the DMU tool from Seemage, Inc. (Newton, MA; www.seemage.com). Seemage is based on eXtensible Markup Language (XML). Everything associated with a part or subassembly (including assembly trees, geometry, part names and layers, and product lifecycle management [PLM] attributes) are extracted automatically from native CAD files and converted to the XML Seemage format. (Contrast this with carrying the whole parametric, feature, history-based view of a 3D world.) Seemage displays are a hybrid representation that mixes tessellated and exact data. Because it’s all based on XML, which is fundamentally simple text, Seemage can automatically compress native 3D CAD models up to 99%. Because of that, explains Jean-Jacques Grimaud, vice president of business development for Seemage, the DMU tool can “probably view up to 100,000 to 200,000 parts [at one time]. We work comfortably with 30,000 to 40,000 parts on a desktop.”
These lightweight XML-based files, continues Alex Neihaus, Seemage’s director of marketing, “are infinitely malleable in the hands of whomever is trying to produce some documentation.” This is because XML is designed for interchange and is easily understood and programmed. Users can, for instance, combine the XML-based 3D data with other part data from other systems, such as enterprise resource planning (ERP), to create documentation beyond that required by engineering, such as marketing brochures, parts catalogs, maintenance procedures, and engineering change documents.
Seemage Sync fully associates the XML data in Seemage with enterprise systems. This way, changes in metadata, geometry, tree structure, bills of material (BOM), or manufacturing information will automatically be updated in the mockup and other Seemage outputs. (This associativity is bidirectional; changes resulting from using Seemage can be reflected back into, for example, core CAD files.) With Seemage Sync, users can create interactive parts catalogs and BOMs, for example, that can be exported in HTML, PDF, and other document formats.
This associativity puts a different slant on engineering management. Explains Neihaus, “XML permits a level of decentralization without chaos because you can give non-technical people ‘point tools’ to solve immediate and pressing problems. Then, because there’s this XML infrastructure under it, you can later integrate those tools in almost any way.” Case in point: Conventional PLM systems are engineering- and server-based; their focus is from engineering out to the rest of the enterprise. Seemage, on the other hand, can be deployed on a desktop first, then later integrated with PLM. “Our secret sauce is being able to do both—be deployed ‘in either direction,’ either as a desktop application or integrated to back-end systems—and end up with a high-fidelity, high-quality, easy-to-use collaborative tool.”
DMU is not just for visualizing mechanical systems. The electronic design automation (EDA) version of AutoVue from Cimmetry Systems Corp. (Montreal, QC, Canada; www.cimmetry.com) combines mechanical parts with electronic components. Among other functions, AutoVue EDA can check violations on printed circuit boards (PCBs), such as pad/trace/component clearances, drill hole sizes, via counts, short circuits, unrouted and empty nets, and unconnected pins by setting multiple tolerances/constraints. The detected violations can be exported for collaboration with other design systems and groups. Better yet, AutoVue can trace the path of a net or signal up and downstream from a specific component on a PCB, as well as through various PCB layers and across multiple schematic pages. This way, engineers can do things like signal matching, visualize what electrical components are supposed to do, and confirm a signal out to something is going to activate a physical action, such as solenoid.
There’s another DMU tool that is “way ahead of everybody,” says Foundyller. “An entire car can be displayed from a computer running Microsoft Windows 98.” This tool is VridgeR from Digital Process Ltd. (Dipro; Kanagawa, Japan; www.dipro.co.jp/english/index.html). “Dipro believes that the two DMU alternatives for customers have always been either low file memory, low accuracy, or high memory, high accuracy. VridgeR provides low file size with high accuracy,” writes Hiromi Mori, Dipro’s general manager and technical director, CAD/CAM/DMU Development. VridgeR attains these twin goals through its “hybrid mesh” technology, which both compacts data structures to 5% of the original CAD file size and controls tolerance representations. A 20-GB CAD model of, say, an entire car, can be compacted and loaded into a laptop computer with less than 1 GB of RAM, and with no loss in data precision.
VridgeR is used by Nissan and Honda R&D, according to Foundyller. The tool has all the major functions one would expect in DMU, including desktop publishing. It lets users add thickness to complicated panels to verify dimensional considerations. The DMU tool also lets users check the draft angle of parts so that users can confirm whether a part can fall out from a mold. Users can even compare imported CAD models with high-density cloud data for quality control.
Vehicle DNA used to be composed of mechanical, hydraulic, pneumatic, and electrical systems. Now add embedded systems. These are the on-board, software-based electronic controls that help improve vehicular performance, safety, and maintenance. As in the mechanical world, engineers want to quickly determine whether newly developed closed-loop control software for the physical systems does what it’s supposed to do. They want to know this quickly. And they don’t want to spend much time ensuring that the software fits into the microcontroller they’re putting together.
To design such systems, engineers typically begin with a paper specification, explains Jon Friedman, automotive industry marketing manager for The MathWorks, Inc. (Natick, MA; www.mathworks.com). “Unfortunately it’s difficult to communicate engineering direction in any natural spoken language, and it’s difficult to read 1,000 pages of engineering specification.” A better approach is to create a “rapid prototype”; that is, essentially simulate the embedded system, including the mechanical hardware, the physical electronic components, and the software control algorithms.
Simulink from The MathWorks, Inc., provides such a rapid prototyping environment for both system-level modeling and algorithm development. By creating both the physical and algorithmic models, explains Friedman, “engineers can create an ‘executable specification’ to determine how both behave together in a closed-loop environment.” Facilitating this process are various MathWorks modules for modeling the physical side of the executable specification. For instance, the SimHydraulics module lets engineers think in terms of pumps and valves, and connect them all together, as opposed to developing all the differential equations for calculating fluid flow, viscosity, and so on. Says Friedman, “You don’t think in terms of the equations; you think in terms of the systems.”
“Model-based design” modeling environment has a number of advantages. One, testing doesn’t have to wait till controller components and software modules show up. Two, it’s unambiguous. Embedded system design teams, regardless of location around the world, can easily see and measure the performance of the simulated, closed-loop, embedded system. Three, lots of alternative and revised embedded system designs and approaches to software logic can be quickly explored. After all of that happens within Simulink, the application can automatically generate the actual embedded code in C or in hardware description language, and target that code for an embedded controller, such as general-purpose microcontroller, digital signal processor (DSP), or application-specific integrated circuit (ASIC). Engineers can also use Simulink to write test scripts to analyze their algorithms.
In April, MathWorks released Fixed-Point Toolbox 2, which provides enhanced floating-to-fixed-point conversion capabilities and accelerated fixed-point algorithms. “Most embedded signal processing and control systems require fixed-point algorithms for microcontrollers, DSPs, ASICs, and field-programmable gate arrays,” explains Friedman. “A major challenge is maintaining the correct behavior of an algorithm when converting it from floating-point to fixed-point representation.” Fixed-Point Toolbox answers that challenge. It also provides an accelerated simulation mode that increases the execution speed of fixed-point Matlab algorithms. (Matlab is the MathWorks’ development environment for creating algorithm, interactively analyzing data, viewing data files, and managing software projects.)
? Engineers can construct a model of an embedded control system, including both its physical and the algorithmic aspects, by assembling design components, each of which could be a separate model.