Is there a shortage of analog engineers? Some industry data suggests there is. But when Electronic Design asked its readers earlier this year if their organization is having difficulty finding qualified analog engineers, more than half said they do not. In fact, 1620 of the 2354 readers (69%) said that they perceived no shortage. So while there may be a shortage, it clearly doesn't affect all companies equally. Possibly, that's because most engineers today deal exclusively with the digital domain. They carry their digital hammers, and to them every problem looks like a digital nail. They don't spend much time thinking about analog design.
As a reporter who covers the analog beat, I think there's more to it than that—the analog engineering shortage isn't where most people think it exists. But before we get to the details of my hypothesis, let's look more closely at that survey to see who thought there was a shortage.
Of that "yes, there is a shortage" group, 639 responded to the question: "What are some of the ways your organization is addressing the shortage?" Ignoring those whose responses were off-topic:
- Seventy-eight indicated that their companies have ongoing relations with universities, that they offer continuing education reimbursement, or that they aggressively recruit new grads at the baccalaureate level. Many of these respondents cited college intern and co-op programs. So, a lot of companies seem to be actively engaging colleges and hiring new grads.
- Seventy said they outsourced or used consultants but did not specifically say they were going outside North America for these resources. Eleven more mentioned looking for analog engineers overseas. Maybe, then, analog designers are like medical doctors who specialize. That is, companies only turn to them when their digital engineers (the family-practice physicians) are stumped. Interestingly, nine respondents in this group mentioned going after retirees for their consulting needs. Perhaps those are the all-knowing GPs of an earlier era.
- Fifty-eight described internal training of existing staff. That suggests that the schools could be doing a better job, or that keeping up the state of the art is a continuing challenge and that one can never afford to rest on one's laurels.
- Seventeen said they hire headhunters who specialize in recruiting analog engineers. Maybe headhunting is a job opportunity for analog engineers who can talk the talk but can't quite walk the walk.
- Only three said they relied on vendor (or disti) FAEs to solve their thorny analog problems, but of course, everybody does that. App notes and data sheets are fine, but a lot of the real "how-to" only comes across when you're face-to-face with an apps engineer from the company that makes the parts you're using. This response suggests that apps engineering is a promising career option for analog engineers who can think on their feet.
- Numerous others mentioned hiring fairs, hiring bonuses, referral bonuses, and advertising, but that's just normal recruiting.
- A good number of responses involved some venting. Forty-six frustrated souls said, in effect, that the problem was recognized, but their companies were doing nothing about it. Another disgruntled bunch said their companies just pile on the overtime.
- Only one responder admitted that his company tries to lure away good analog engineers from its competitors. Balancing that, another mentioned his company's efforts to retain good engineers. Based on other conversations I've had, I think a lot of companies would like to poach engineers from their competitors, but aside from friction with management or a spouse who refuses to spend one more " winter in this igloo" or "summer in this sauna," engineers tend to stay put—especially when employers are willing to match competitors' salary or status inducements.
Some of the longer comments about the state of analog engineering today are thought-provoking (see "Comments From ED's Survey" at ED Online 13667 at www.electronicdesign.com).
PhDs only need apply?
The survey, as well as the readers' responses, represent one way of thinking about the analog engineering profession. But let's consider a more radical position: There is indeed a desperate shortage of analog EEs, but it isn't a shortage of analog-circuit designers. As analog chipmakers increase integration levels, the need for deep insight into analog-circuit design at the component level has decreased. For example, a DDS chip eliminates the need for many engineer-weeks on the bench.
I think the shortage of analog engineers isn't so much at the BSEE level, but at the graduate level—analog chip designers who can bend a process technology to their will. The need may be either inside an analog IDM or at an IP house or a fabless startup.
I spoke with Stephen Maneatis, CEO at True Circuits, an analog IP provider that specializes in ultra-stable phase-locked loops (PLLs) and related circuits. His company is presently in a full-court press for analog engineers. The analog engineers True Circuits needs, however, are very specialized.
Catch 'em young
Maneatis' job description fits a very small population of engineers, most of whom already work for companies like Analog Devices, Linear, Maxim, National Semiconductor, and Texas Instruments—companies that would meet any purely monetary offer True Circuits might make to lure engineers away. Moreover, most of those engineers probably already know each other from conferences. Or, at least the ones in North America know each other, so it isn't as if a listing on Monster or Craigslist is going to turn up any surprises.
Also, the fact that Maneatis really needs PhDs (who also have more than five years work experience) limits the talent pool. Realistically, an about-to-be-digital BSEE might come to the realization that spending another year or two pursuing an MS is worth the effort. But for EE students to qualify for analog graduate study, particularly at the doctorate level, they had better start picking out the right upper-division courses when they're sophomores. And who's going to advise them through that process and guide them along the way?
The Maneatis brothers, Stephen and John, know more than just about anybody about building timing circuits in deep nanometer-scale process technologies. However, few are the sophomores who would be tempted away from the glamour of digital by the prospect of creating the world's most stable PLL. Is it possible that analog engineering lacks a certain coolness factor?
Not necessarily. Consider another, younger analog startup, Gtronix. If it can pull off what it's trying to do, it's guaranteed to turn an undergraduate head or two. Gtronix plans to introduce "a new and intelligent programmable analog structure and architecture to completely integrate the signal-processing chain for sensors—Smart Analog for Sensory Processing."
That means the company aims to shift a lot of the DSP associated with processing sensor signals back into the analog domain, ahead of the data-conversion step. Gtronix wants to do it using its floating-gate analog programming concepts. Originally developed at Georgia Tech by CTO Hans Klein and others, these concepts implement functions such as finite impulse-response (FIR) filters with fewer gates and lower power requirements than when they're implemented digitally.
For example, Gtronix designers say they can build an analog multiplier with a dynamic range equivalent to 10 bits of digital resolution using four gates instead of the 3600 required by a digital multiplier. Power-wise, a digital echo-cancellation circuit that would burn 20 mW could be replaced by an analog equivalent that consumed half a milliwatt. This is all based on circuits that can be built using TSMC's Vanguard International Semiconductor Company process technologies. Cool stuff.
Asking the experts
If my strawman has merit, dealing with this kind of shortage means reaching undergrads before they start taking upper-division classes, so they can schedule the right ones. It also means motivating them to follow a program that will keep them at the university for several more years versus pursuing a four-year digital path. Further, it means convincing EE department chairs and faculty senates to make the options available.
To get some real-world feedback, I posed some questions to Kenneth A. Connor, professor and head of electrical, computer, and systems engineering at Rensselaer Polytechnic Institute, and to Dave Robertson and Sam Fuller at Analog Devices. Robertson is ADI's product line director for high-speed signal processing, and Fuller is vice president of research and development. (In his answers, Connor mentions ADI, but that's a coincidence. I picked RPI because I have an MS from there, and I picked ADI because I was talking to them the day I thought of asking.) In my questions, I offered my strawman and asked: first, how realistic they found it; and second, when undergraduates have to decide to pursue coursework that will prepare them for advanced studies in analog EE (suggesting that it must be just before they start upper-division work). I also asked Connor about reaching out to undergrads to make them aware of analog opportunities.
To summarize, all three felt my strawman was a little overstated, but not extravagantly. Professor Connor said that there's always been a shortage of people who can do real analog design, but that BS grads who had practical experience were readily hired. Robertson noted that only a few schools have strong analog programs and that alumni from those schools influence corporate investment in the school and the hiring of new grads.
Fuller provided the concrete data that ADI's hiring runs consistently 40% BSEEs, 50% MSEEs, and 10% PhDs. He added that "an engineer doing leading-edge design at ADI earns the equivalent of an MS degree every three to five years."
All three agreed that PhD or not, it's essential to get undergrads committed to the analog path as early as possible. Each had his own comments to add to that. Finally, Connor described a number of instances of industry involvement that help undergrads become acquainted with real-world opportunities in analog engineering.
The elephant in the room
One last word: Some engineers who responded to Electronic Design's 2006 Reader Survey worry that their jobs, in whatever specialty, are being outsourced overseas. They fret that overseas universities are turning out graduates who can out-engineer their North American counterparts while flourishing at home on half the pay. Additionally, they worry that foreign engineering grad students outnumber their North American counterparts in U.S. and Canadian universities.
This is a serious concern for code developers. It's a little harder to gauge for analog EEs. At this time, Chinese OEMs, especially those pursuing higher-margin, branded product models, seem happy to leave analog and mixed-signal chip development in the hands of companies such as Analog Devices and Wolfson. An Indian VP at a chipmaker with a large analog portfolio confided to me his impression that Indian universities are still concentrating on computer science, because the outlay for capital equipment for training costs less. (Well, that's my oversimplification of what he said.)
Speaking of overseas universities, it is a fact that a combination of high admission standards and early placement of young people on university and non-university tracks tends to select for really bright people. What also happens is that the unconventional thinkers and late-bloomers are selected out.
What about the perception of over-enrollment of foreign students at North American universities, particularly in the grad schools? Opinions vary. The issue came up at an APEC-conference breakout session last spring on the impact of China on the makers of power chips and supplies.
At one point, Fred C. Li, director of Virginia Tech's Center for Power Electronics Systems (CPES), stood up and provided the actual breakdown, by country of origin, of CPES students. His explanation did not support the image of an institute overrun by outsiders. On the other hand, people I talk to continue to insist that this or that program is disproportionately populated by students from abroad.
These issues point to several questions that are worth considering. Is there really a shortage of analog EEs? Is the shortage worldwide or regional? What skill set constitutes an analog EE? How and where does one acquire that skill set? And, how much is an analog EE worth?