Is there any design left in engineering?
By Bill Schweber, Executive Editor -- EDN, March 4, 2004
Design and engineering aren't what they used to be, as EDN's audience well knows. Years ago, before processors and software ruled the world, designers would build up a circuit from relatively low-level building blocks, such as resistors, capacitors, small-scale ICs, transistors, and even vacuum tubes. They could configure a set of such components to provide functions that were defined primarily by final circuit topology; looking at the BOM (bill of materials), you would have only a hint concerning the design's final function.
Today's reality is that system-level ICs—whether we called them ASICs, ASSPs (application-specific standard products), SOCs (systems on a chip), or some other name—dominate in the circuit-versus-system debate. These ICs usually target a specific, carefully defined application, serve well in that application, and are ill-suited for almost anything else; graphics accelerator and 802.11 ICs are obvious examples. When you look at the product's BOM, the names of the key ICs give you a good idea of what the final system does. Although the design still contains plenty of passives and a few active, discrete devices, those devices no longer define circuit function; their role is to ensure that the IC can properly do its job with requisite bypassing or fill-in for functions that the designers could not integrate.
Reference 1 highlights the implications of this changed situation. It shows how nontraditional vendors are getting into the TV and multimedia business, because being a "vendor" now means specifying the required ICs and chip sets and then having an outsourced manufacturer buy them and put them on a pc board for you. The IC vendor handles the bulk of the detailed engineering work, supporting it with reference designs, application notes, and application-specific software and drivers. Most of the engineering design effort, therefore, concentrates not on design itself, but on making the prescribed design manufacturable at the right cost; integrating the enclosure, screen and keyboard, power supply, and other noncore portions of the product; and lining up vendors for the BOM.
No one knows the long-term implications of this change on the role of the design engineers. Will it cause them to lose sight of and comfort with low-level, design-related issues, which only a few practitioners will actually understand? Meanwhile, will the rest of them happily combine chip sets that were, in turn, designed by a handful of other engineers? Will the industry attract more software-only jockeys, for whom parameter passing and calling the RTOS define design life? Will it stop attracting engineers who know how to get into a design's dirty real-world details? Will design become a contradiction in terms for many who call themselves engineers? Will we need fewer "real" engineers?
Or, will engineers and their design challenge take on a new shape that we don't yet understand, as they concentrate on the struggle to get that final 5% of their product's design working reliably and ready for manufacture?
I don't know the answers, but I know what I'm going to do: build that four-transistor, software-free lightning detector I saw—the one that had no critical component values or layout constraints (Reference 2). It looks like pure engineering fun to me.
Contact me at bschweber@edn.com.
| References |
|
-
Isn't it ironic that there is an article in EDN that seems to oppose the shift to using other engineers' designs, yet the same magazine publishes predesigned circuits for everyone to use. Just some food for thought.
Kraig Ritmanich - 2004-29-3 09:32:00 PST -
Having read your comment some weeks after its publication, I see that it has stuck a chord with many engineers, including this one. I guess that what has attracted many of us to the field of electronic engineering is the opportunity for creative problem-solving, and then the satisfaction of putting a few simple parts together and getting it working.
Unfortunately, creativity and grappling with the real world details of making a circuit work reliably, require an investment in time that fewer companies seem able to afford. 'Getting to market' is the over-riding commercial necessity, and there is an increasing choice of pre-packaged solutions available. In one form or another, at one level or another, design re-use is preferable to design innovation.
I recall seeing a comment somewhere by James Dyson (of vacuum cleaner fame) to the effect that it is becoming harder and harder to find something to invent, quite simply because the possibilities are being used up. How many ways are there to connect three or four transistors together? We now have the Darlington connection, the long-tailed pair, the emitter-coupled astable, two/three/four transistor current mirrors, and so on. Of course, many of these familiar configurations pre-date the transistor! Only the more arcane possibilities remain at that level. The same argument applies to the development of computing algorithms.
These, I think, are the factors underlying the changes we are seeing. Transistor-level design seems increasingly to be the preserve of fewer engineers, because, economically, the world prefers it that way. No doubt there are a few corners, here and there (in academia, or blue-skies research that only the largest companies can afford) where the kind of design work we'd enjoy still goes on.
David Ashby - 2004-23-3 04:19:00 PST -
Over the years the "design engineer" has seen his/her role take on additional tasks such as program management and pre-production engineering. Compared to 10 years ago I imagine the actual engineering work has decreased by a third.
It is true that IC reference designs are an important factor engineers consider in selecting final solutions. They are very important in Asia and Japan.
David Gillooly - 2004-15-3 14:14:00 PST -
Dear Mr. Schweber:
Your comments resonate with this retired engineer: I graduated BSEE 2 years after the transistor was invented. I read EDN today and, while I find it interesting and informative (much better than Design News, which I also still get), I have been struck with much the same thoughts as you express. It often seems that the only actual detailed circuits I find in EDN are those in Design Ideas.
Brian Dipert's seemingly-excellent article is a case in point: not a single mention of a resistor or a capacitor, never mind a soldering iron. I say "seemingly" only because a lot of the details went right over my head.
You ask whether hands-on engineers will fade away. Of course, each and every item employed by, for instance, Brian Dipert, had to have been meticulously engineer-designed, so that likely will not happen in the forseeable future. But I wonder what publications are going to be around for such engineers to read and take nitty-gritty value from. Not EDN or
Design News, I'd think.
And what language might they be written in? Urdu, perhaps? Chinese, surely. And perhaps not English...
Regards,
Ken Herrick
Oakland, CA
Ken Herrick - 2004-15-3 07:33:00 PST -
Bill,
I just wanted to congratulate you an a well written and timely article. I very much identified with the situation as you described it. I graduated from college in '97 and have been working as a hardware engineer at a series of startup companies since then. As a board designer, I have become increasingly frustrated with the lack of creativity and actual design required of my job. My job has increasingly become an exercise in integrating a processor with a set of ASICs and support chips. The actual implementation is often spelled out by the ASIC manufacturer and there is little room for creativity or even what I would consider design. Instead the devil is in the details and then debugging the boards.
It seems to me that, unless you are pushing the technological edge, hardware design has become more of an exercise of details.
I believe that there are some fields of "hardware" design such as FPGA or ASIC design where you have a chance to be creative, but these design processes are increasingly become more software like. In fact I would advise most aspiring engineers today to pursue degrees in computer science rather that electrical engineering for this reason. The other alternative would be to remain an academic pursueing hardware or semiconductor research. In my case, I will be starting an MBA program in the fall. I love technology and startups, but I don't want to spend the next 20 years plugging chipsets together!
Regards,
Gary
Gary Gill - 2004-12-3 07:21:00 PST
