Is a Ph.D necessary for success in the analog IC industry?

[Swzpae]

Im trying to pursue a M.S degree on analog IC design and want to know how good is the market for analog IC designers these days in North America and/or Europe. Also, is a M.S enough to do well in this area or should I think about getting a PhD ?

 

[JakeBrodskyPE]

I don't care how fancy the digital designs get. At some point it will have to acquire data and control real things. This will require good low noise, high bandwidth, high dynamic range analog designs. Someone will need to design those devices.

Other examples are low noise oscillators. As digital performance gets better and better, analog designs will have to accommodate them. Do not overlook servo motor designs, RF transmitters, and the like.

Again, digital circuitry is really cool, but eventually it will need to interface with the analog world. People have been predicting the demise of such technology for many years. It has changed, but it's not going away any time soon.

And yes, an MS is good. You could pursue it with just a BS alone, but an MS will look better to the bureaucrats in Human Resources.

 

[leright]

I agree with Jake completely. We live in an analog world. Real, physical quantities, are analog in nature. There will always be a need for analog and mixed-signal IC designers, and there is currently a SHORTAGE of them.

 

[analogdesign]

I'm a professional analog IC design engineer working in the Bay Area. I disagree with Jake and leright in principle. Just because something is needed it doesn't mean you can find good paying work doing it.

That said, right now the market for analog designers is quite good but it is very cyclical. It was tough getting a job five years ago, but there is a shortage now. Apple is scooping up every analog designer they can find right now (four of my former coworkers are there) and this is creating a bit of a shortage for everyone else.

An MS is generally considered the entry level degree. Analog design is one of the few sub-areas of engineering where a Ph.D. can help you get a job in industry. It takes a long time to get up to speed in analog design (compared to software or digital) so companies like pre-trained students. You get that doing a Ph.D. since you'll have to do a complex design yourself from concept through tapeout and test.

It's hard to get an entry-level job with a BS unless you have some amazing internship experience. Not saying it's right, but it is the way it is.

 

[leright]

The entire semiconductor industry and the demand for IC engineers is highly cyclical. Not just in analog design. But your point well-taken.

 

[Shaun_W]

Isn't everything cyclical?

 

[JakeBrodskyPE]

Isn't everything cyclical?

Some industries have stronger boom/bust cycles than others. Some (such as utilities, government, academia and the like) tend to be very sedate places to work. Few, if any become wealthy at such employment, but they can make a comfortable middle class living.

Conversely, others such as aerospace and energy exploration industries have strong to even extreme boom/bust cycles.

I was not aware that the analog IC industry had such cyclical behavior, but I can see how that might be the case. As for NEEDING an MS or a Ph.D --that's a function of the hiring processes. Like many other fields, you can learn a lot about this field informally (and many from my generation did just that) or you can get a formal degree which looks fabulous to a bureaucrat driving a desk.

They don't hand out secret decoder rings in the university when you graduate. If anything, it's just the opposite. How well you understand and apply the things they taught is all about who YOU are as a person, not the education you received.

 

[analogdesign]

The semiconductor business in general has brutal boom/bust cycles due primarily to the large investments required to increase capacity. The cycles have somewhat smoothed out in the last 10 - 12 years or so (since the industry recovered from the epic dot bomb) but particularly in the 70s and 80s they were legendary.

I agree no one "NEEDS" a degree for anything. People can be mentored and learn anything on their own. One of the best intuitive analog designers I ever met, Jim Williams, didn't have a college degree. There is more to it than a bureaucrat driving a desk.

I'll tell you two reasons why analog IC design companies like to hire Ph.D.s. First, it's because they can hit the ground running and cost less to train. Advanced analog development is difficult and requires more expert knowledge than equivalent software or digital design because less knowledge is embedded in the tools and frameworks used in development. This is primarily because an analog design has more degrees of freedom than a digital design and can't be endless tweaked and updated like an Agile software development.

Now, to become a useful analog designer you really need to go through a full design cycle. You need to design a circuit in schematic, do the layout, do all the verification, tape it out, bring it up in the lab and characterize it, and write or contribute to the datasheet. If you hire a fresh BS student, he or she might have a bit of background, but you're going to have to hold their hand through this entire process and it could take over a year.

On the other hand, if you hire a Ph.D or an MS (from select universities), they've gone through this process in school and actually going through the process and experiencing it is where the real learning takes place.

So it comes down to companies wanting to hire someone to help RIGHT NOW and not being interested in training and developing their workforce. This is just another consequence of that general trend.

The second reason is the level of sophistication required from analog chips and macros is going through the roof as more and more of the system architecture is subsumed into the chips that comprise the system. I work on systems that contain analog adaptive equalizers (for serial links) and self-calibrated ADCs. Designing these systems requires a great deal of math and I would have a hard time understanding them if I didn't have a strong background in random signals, time series analysis, and discrete math. These are not typical things you learn as a BS EE focusing on circuits and 20 or 30 years ago analog ICs were more components of systems and now they are systems unto themselves.