Originally Published MDDI
August 2004
Design and Engineering
Advancing the Design of Medical Electronics
The advent of visualization and integrated tools has enhanced the
design process and has provided an inherent ability to help document the resulting design.
Robert Kay
 |
| Robert Kay is president of Elite Engineering (Thousand Oaks, CA). He has been a member of the MD&DI editorial advisory board for 11 years. He can be reached via e-mail at e-info@ eliteeng.com. |
The past 25 years have brought many changes to medical design and engineering, and through the years those changes are evident in the pages of
MD&DI. Some changes were predictable and some not. The many forces that shaped the path to today’s state of the art were widely varied and often nontechnical in nature.
Recalling Some Memories
Having now reached an age where one could be classed an old-timer, it is well within my memory to recall design of electronics using vacuum tubes and cat-whisker crystals. College projects involved the use of an Altair computer, which was thought to be an astonishing device all on its own. By the time I launched my engineering company, use of microcontrollers had grown significantly, and the tools for launching a plethora of handheld medical devices had been well established.
A quarter century ago, application-specific integrated circuits (ASICs) were simply called custom
integrated circuits, and commercial CAD programs were few and far between. Simulation tools were in their infancy, and engineers still needed a good calculator and their favorite handbooks to simplify their analysis work. Software tools for thermal, vibration, or fluid-flow analysis usually involved either a mainframe computer or custom-written PC software that had limited graphical capability.
It is not just the technology that has changed significantly in the medical device industry. FDA began to move toward an increasing emphasis on
controls and documentation in the design process. It was becoming obvious to everyone that software was the proverbial double-edged sword: amazing flexibility and what seemed like limitless applications, but with it came an equally limitless number of ways to fail. Out of some dramatic failures of both devices and device companies the message was increasingly clear: software seemed straightforward on its surface, but it was, in fact, a technology that required more than simple programming skills.
Looking Deeper
Market forces—at one time local and now more pervasively global—whether real or perceived, have created a relentless drive toward smaller
devices, lower costs, and faster concept-to-market cycles. These forces have created a need for a wide range of new tools and devices to meet these demands. Yet, while one might point to new components, materials, and processes as things that lead up to our ability to satisfy those needs, just as many underlying changes in design tools enabled the path to today’s technologies.
 |
Two often-overlooked—and now taken for granted—developments stand out as having significantly changed engineering and the design process over the past 25 years. The advancements in both graphical displays and of modeling and simulation software contributed much to medical electronics development. Moreover, the costs of both of these technologies have drastically decreased during that time.
The improvements in performance and reduction in cost of high-quality graphics in PCs allowed software tools to tap into the highest-bandwidth communications channel in humans:
visualization. Slowly but surely, the need to recognize patterns from lists of numbers led to the development of 2-D graphs and, soon thereafter, to
animated 3-D surface displays. Herculean mental focus was no longer required to recognize trending or correlation. With the addition of some computational power, even-more-sophisticated mathematical transformations could be visualized. At a glance, spectral analysis could provide data such as a determination of feedback system stability.
With this trend to present analytical results with a visual display also came increasing sophistication of modeling and simulation tools. The explosion of PCs, and the subsequent transfer of complex simulations from mainframe computers to a desktop, put powerful tools in the hands of more designers and engineers than ever before. No longer limited to simple graphs of circuit simulations, these desktop tools extended into finite-element analysis, as well as optical, thermal, and
electromagnetic analyses.
The expansion of these graphical tools soon took over physical modeling as well. PCB layout truly rocketed away from layouts taped on acetate to become an effort executed in pure electronic form. Design work from concept to analysis to finished PCB
artwork is now executed almost exclusively on a desktop computer.
By this time, gone were the days of the mainframe computer sitting in a darkened room with its own air conditioner and raised floor. But even better improvements were still to come. If one could design PCB artwork on a PC, why not solid parts? Soon, the addition of solid modeling provided the means not only to physically design a mechanical part, but also to fully analyze and simulate that design.
With today’s fully integrated tools, one can design and analyze not only parts, but full assemblies for mechanical, thermal, and optical properties, to name just a few. And, once satisfied with the design, designers now have tools to electronically generate the
necessary files for machining and fabrication of those parts. Rare is the hand-operated machine shop mill, and more common is the multiaxis, tool-changing CNC mill connected directly to a network for the download of
fabrication file data. These tools now encompass a range of knowledge databases and capabilities that far surpasses those of many of the individuals who use them.
Many medical products rely on mathematical modeling and analysis to optimize their performance. In past days, one used handbooks and tables to perform transformations and analyses. A designer no longer needs to know how to execute a Fourier transform integral in order to do such analysis. You need only understand the concept and how to interpret the results. Software tools provide fully automated means for performing these transforms and other sophisticated analyses.
For the medical industry, one of the advantages of the advent of fully integrated tools is the ability to use them to document the design process at each step. And, while it still requires self-discipline to properly arrange and orchestrate that documentation, the tools make it a less-arduous task to create. In some cases, the tools themselves provide change-tracking methods and tools.
The Uncertain Future
To most, this inexorable drive in design technologies is taken for granted. Next year, there will be better tools. One wonders where these improvements will lead.
It is my opinion that graphical tools will not add significantly more value nor play such a dramatic role in improving our productivity in the future. Today, the limiting factor to design is no longer design visualization. Moreover, I don’t believe the future relies on advancements in simulation tools,
either. Rather, one of the most significant limiting factors to design relates to the underlying mathematics and our ability to develop intuitive models for nonlinear processes.
As someone who started his career as an analog engineer with a penchant for closed-loop controls, I have always thought that we have been coasting on the mathematics of the early to mid-1900s. Tools such as LaPlace transforms were developed, which turned calculus into exercises of algebra. Such tools enabled those who struggled at solving differential equations to write closed-form solutions for circuit and mechanical designs.
A closed-form solution has been the most desirable because it is an equation that can be directly applied by plugging in values for each variable to obtain a result. These closed-form equations allowed engineers to develop predictive tools that then provided much insight into parametric response. It was not necessary to run hundreds of simulations to know how to make a control-loop unconditionally stable. By observing the defining equations, one could see intuitively whether specific solution sets existed.
Yet, while these tools have been crucial in the path to where design engineers are today, they have always had one primary limitation. These tools are linear models, but almost all problems of interest involve nonlinear processes. This limitation was overcome by the use of piecewise linear models when a closed-form solution was required.
Then, along came computers and simulation tools. These tools could evaluate nonlinear forms and could provide reasonably accurate simulation results. However, they provided results to specific conditions or parametric settings only as good as the information the person running them had put in. Simulations basically answered only the question asked. They did not give the general insight that closed-form equation solutions would
provide.
It is my contention that the real breakthroughs needed are in the mathematics and tools to allow closed-form solutions to nonlinear differential equations. The advent of a tool or method to solve nonlinear differential equations in closed form would
provide insight and much-improved predictive skills for fluidic devices, software analysis, chemical systems, and information systems. Such a breakthrough would allow the average engineer to better predict performance of designs in these areas and minimize trial-and-error design methodologies.
Basic platforms for enabling such breakthroughs already exist. Software tools are already available to solve equations and to perform transforms in symbolic form. Nonlinear equation– cracking tools could change engineering much as digital circuitry changed electronics. Such a tool, when tied to the graphical and analysis tools already available, would enable almost any properly motivated designer to develop high-end designs where parts can be designed, fabricated, and tested in one integrated process with less testing, fewer errors, and fewer iterations.
Of course, despite the advanced mathematics, and despite how advanced we may make our tools, errors and mistakes will always be a fact of life. Our tools can never be a substitute for the benefits of an understanding of underlying principles and experience.
Copyright ©2004 Medical Device & Diagnostic Industry
