Do you believe that the real analog engineering shortage
is at the PhD level?
Ken Connor (RPI): Not exactly. There is indeed a shortage of people who can do real analog design. This has been the case for years. Students get jobs as BS grads, but only if they already have some good practical experience. I see the trend as companies hiring grads who know something. We have been very lucky and recently hired two analog-circuit designers for our faculty. One came from Ohio State and worked at RF Micro Devices for a while, and the other came directly from Stanford. The master's students \[under the tutelage\] of the one from OSU are being hired by Analog because they know something. Analog also just hired a PhD student who was being advised by one of our emeritus faculty. We have a very good relationship with Doug Mercer from ADI, who is a spectacular analog person. He has given a couple of talks here in the last two years. The last time he decided to ask the audience a few questions that he thought were fairly simple. Usually, the answers came from grad students, mostly doctoral students. A couple of undergrads also did okay. (One of them is working on an ADI-funded project and worked for them last summer.)
Dave Robertson (ADI): The "analog community" is a relatively small one. There tends to be a handful of schools with strong analog electronics programs. Alumni from those schools often maintain strong ties to former professors and stay close to the programs. Analog does some sponsored research; in some cases sponsors a chair, gives guest lectures, works co-op programs, and generally works to "stay connected." One thing to keep in mind—this approach is applied worldwide, not just in North America. The vast majority of the hiring continues to be at the bachelor's and master's level, and there is a great deal of value in a great breadth of training—analog, digital, semiconductor physics, system simulation, signal processing, etc.
Sam Fuller (ADI): Over the past several years, our new college hires have been about 40% BSEEs, about 50% MSEEs, and 10% PhDs. I would also add that an engineer doing leading-edge design at ADI learns the equivalent of an MS degree every three to five years. Those new hires that come to us directly after earning their BS degree have some additional on-the-job learning to get to a fully productive level. As a result, the continuing trend for more of our new college hires is to come with an MS degree. We also recognize that it is important to have a consistent relationship with the EE departments where we hope to hire new graduates. Fresh BS and MS engineers are graduating every year, both in the upcycle and in the downcycle of the semiconductor business. Companies foolish enough to cancel campus recruiting during the downcycle find they are at a significant disadvantage at colleges when business turns positive again and they need new hires. Analog Devices has made a point to recruit at the colleges and universities with the leading EE departments every year, irrespective of the phase of the business.
Do undergrads have to make up their mind about pursuing an advanced degree
before they start upper-division studies?
Ken Connor (RPI): That is correct. They need to take every available circuit design course.
Dave Robertson (ADI): I would agree... but would point out that for every analog function that digital replaces, three new "analog" problems may crop up. The DDS part you used as an example \[the DDS chip eliminates the need for many engineer-weeks on the bench\] replaced an analog modulator with a digital multiplier, but the overall clock rate is now very fast, with some rather steep jitter requirements on the clock PLL.
At the same time, the DAC dynamic performance is much more demanding, and significant portions of the digital logic may now be running so fast that parasitic back-extraction and simulation must be done at the transistor level—all "analog" class problems. Does that mean we need more PhDs? Not necessarily.
You need very smart people with a rigorous base training (at the bachelor's and master's level) and then experience and teamwork to learn how to piece things together—essentially great problem-solving techniques. That experience could come from a PhD thesis, but it can also come from working on a couple of advanced projects alongside strong mentors who are experts in their field. Consider the Analog Devices Fellows. Sure, some started with PhDs. But many came out of school at the bachelor's or master's level.
One implication of larger levels of integration is the need to work effectively in teams—the skill and art of "gang tackling"—which makes technical communication and documentation very important, not to mention relatively new skills like " verification strategy."
We're certainly all in favor of strong programs that offer challenging courses. But when you really look at the top talent in the industry, the training didn't come from an "exhaustively complete" set of courses, but often from a handful of good courses delivered by great teachers who could simultaneously infuse technical rigor, insight, and creative problem solving.
Sam Fuller (ADI): In large part, I agree with your thinking. What I might add is that we continue to see increasing levels of integration. More of the signal chain, and more of the overall value chain, is moving onchip. This means all the chip design teams, even the teams doing " digital" chips, must address analog design issues. Maybe to say this another way, the system design challenges of yesterday (analog and digital) are relentlessly moving on-chip. Does this mean we need fewer analog-circuit designers? No. We will probably need fewer analog-circuit designers working with soldering irons, but we will need more IC designers with strong analog circuit design knowledge.
How do you reach undergrads?
Ken Connor (RPI): We have two activities, both associated with Analog. I already mentioned Doug's talks here. He has done a good job of publicizing ADI with our students. The other is an instructional electronics project also funded by ADI and HP. (Soon one other company will probably be added.) This helps students know about such opportunities. I have a new program beginning where we will be creating a community of interest in circuit design. I hope to get some people from ADI, Linear, and a few others to join with students and faculty who share this interest.
Also, our digital design activities are largely analog, since they deal with SiGe and are thus very fast. This has traditionally been a good area for us, and the students are told over and over that what they are doing is really analog.