Connect with us

Career in Analog IC design

Discussion in 'Electronic Design' started by Steve, Dec 15, 2005.

Scroll to continue with content
  1. Steve

    Steve Guest

    Hi,

    I am looking to persue a career in analog design, but, before I commit
    myself to this decision, I would like to get the opinions of the people
    here to have a (hopefully unbiased) look on what my options are. I've
    heard a lot of hand-wavey arguments about how
    "everything-is-really-analog" on some level, and how there are 10
    digital designers to every 1 analog designer, but is this really true?

    I like the fact that analog design is multi-faceted, i.e. one has to
    know some mathematics, some design, some physics etc.. I also like the
    fact that there is a lot of "design" in the process, one has to create
    a layout matching the specification of the design, one has to think
    about what the circuit will do out of a variety of possible options. If
    fact, I think that it's good to specialise in such a field because the
    knowledge that one holds is far more difficult to learn than C or HDL,
    and therefore of greater value to a company. These are the issues that
    are drawning me to this discipline. I would therefore like to know what
    the career paths are for those starting out in analog design. Will
    analog design become less previlent in years to come as more and more
    is moved into the digitial domain? Are analog designers really in high
    demand within the industry?

    I have read Grey and Meyer's book on Analog design, and Razavi's, but
    I'm not entirely clear what the cutting-edge issues in analog design
    actually are. The books in question concentrate mainly on simple
    differential op-amps and current sources, analysing their stability,
    frequency response and what not. To me, it doesn't appear that the
    design of op-amps (signal goes in, larger signal goes out) is really an
    interesting career choice, and more so, I don't understand why someone
    who designs them would be in command of such a high salary such as they
    are. Surely there is much more to analog design than just making an
    op-amp that is -say- stable at 1Mhz and can work at a voltage of 2v. I
    would like to know what industrial designers work on. Do they work on
    op-amps such as these, or is this just the simple kids stuff,
    equivalent to the -say- state machine design in digital design books.

    What are the issues in analog design these days, (Sigma Delta
    converters, PLL, high speed front ends, low-noise, low-power etc...)?
    I understand that people are often limited by the affect of deep
    sub-micron processes but is this considered an analog question, or a
    physics question? What would constitute a PhD. thesis in the area of
    analog design: the design of a high-performance op-amp, a DAC/ADC, an
    RF front-end, a new "current mirror" design? I really need to get a
    feel for the levels of complexity involved.

    Thanks,

    Stephen
     
  2. Jim Thompson

    Jim Thompson Guest

    If you actually want to design, go do it, forget the PhD. If you want
    the PhD, go teach and pretend you know how to design ;-)

    If you have fun doing analog, do it... otherwise go sell real estate.

    ...Jim Thompson
     
  3. My view, as an analogue designer of er.. some years? If you have to ask
    people whether or not to commit to a career to analogue design, or for
    that matter, any career, then probably analogue design (or that career)
    aint for you.

    Many of us started at like, 11 years old. It simply wasn't a choice that
    was made after graduating, or even in selecting what degree we should
    take. We have *always* been analogue designers. Its what we are. Its in
    our blood.

    I personally find it rather strange that people have to ask others as to
    what they should be doing as a career. We are not you, and cant possible
    know what you *like* doing. We can only advise on objective matters
    like, salary, and job demand. So, as far as salary goes, yeah its quite
    good when there are jobs, as far as number of jobs go, analogue design
    is a very poor choice, especially i.c. design. There are so few
    companies. Try software, they are *millions* of C++ jobs, hence its
    easier to find just the one for you.

    Kevin Aylward

    http://www.anasoft.co.uk
    SuperSpice, a very affordable Mixed-Mode
    Windows Simulator with Schematic Capture,
    Waveform Display, FFT's and Filter Design.
     
  4. linnix

    linnix Guest

    ....
    Not that there are fewer companies, but they are moving closer to the
    factories. Many crtical issues (PCB manufacturings, components,
    wirings and shieldings) are resolved at the factories. Last time I
    checked, your country (UK) and mine (US) are not interested in
    productions. So, the best choice of analog career is in Asia.
     
  5. Dan Hollands

    Dan Hollands Guest

    Unless you are designing ICs there I don't believe that there is much call
    for pure analog designer

    Every product that I have designed in the past 10 or 15 years has been a
    mixture of analog and digital circuits. Since I worked for fairly small
    companies we didn't have dedicated digital or analog designers but had to
    deal with analog, digital, embedded microcs, micro programming in ass'y and
    in C, and then finally make the complete product pass all of the
    environmental and safety requirements. In a smaller company you can also be
    involved with production, sales and marketing so that you see and influence
    all phases of the business

    I would find doing only analog design boring. Much more interesting to
    design a complete product. For this kind of design practical experience is
    more important than advanced degrees. Most companies want someone who will
    be hands on and who knows the art of design as well as the mathamatical
    theory

    Dan

    --
    Dan Hollands
    1120 S Creek Dr
    Webster NY 14580
    585-872-2606

    www.QuickScoreRace.com
     
  6. Yes it is basically true as digital solutions can replace analog soutions 9
    out of 10 times.
    Depends on the company. RF oriented companies still value analog.

    These are the issues that
    No. Analog is indeed the mode of the physical universe but 0s and 1s are
    easier to minupulate.
    Digital signal processing is replacing analog IC designs in many cases.
    Let's say you need a stable feedback system with variable loads ... DSP
    shines. Sure, you need an analog gain stage here and there, but the core is
    a DSP computer.
    If you can find a deep specialty in A/D or RF, then you will be OK.
    Otherwise, cave in to digital. Just my opinion!
     
  7. Chris Jones

    Chris Jones Guest

    You might like to read Tom Lee's book on RF CMOS, that has some reasonably
    up-to-date discussion of what people were working on about 5 years ago (as
    well as some ancient history in the front of the book). Another source of
    information about what companies are interested is the IEEE Journal of
    Solid State Circuits (JSSC), which you should be able to find in a
    university library. The more processing is done in the digital domain, the
    better ADCs will be needed to make it possible, therefore there is always
    analogue work to be done. Similarly there are plenty of signal processing
    jobs that are not sensible to do in the digital domain for power and
    performance reasons (e.g. a 2GHz amplifier could theoretically be
    implemented by digitising and multiplying by ten then putting through a
    DAC, but the performance, cost, size and power dissipation would each be
    1000 or more times worse than the all analogue solution. Maybe in 10 years
    time it would be only 100 times worse in each respect, but the gap will
    never close.

    Chris
     
  8. Yes Please,
    at PHY and baseband level there will always be need for analog. Most
    of this is moving to mixed-signal design (analog and digital on same
    chip) this means CMOS analog design in a standard CMOS process. This
    is very difficult! Very high-end analog will still be done in separate
    chips.
    and worked the problems? hmmm If you can read Razavi's book and
    understand it, you are much smarter than I am. I think Gray and Meyer
    is a better book. Allen and Holberg is another book to add to your
    library if you are serious.
    not good. Pick up a copy of the red rag (IEEE Journal of Solid State
    Circuits). If you are an undergrad, reading it you can find what you
    like and who you might want to work for as a grad student that is
    cutting-edge enough to get put into a journal. Picking the right
    program where right=what you want to do is critical.
     
  9. Ingeniur

    Ingeniur Guest

    Good on you Stephen - we need more like you. Unfortunately many of the not
    so good Universities neglect Analogue nowadays but it is still very much
    needed!

    Good luck

    Ing
     
  10. Ingeniur

    Ingeniur Guest

    Sounds like a guy with no Ph.D! (He's right though)

    Ing
     
  11. Ingeniur

    Ingeniur Guest

    That's a bit daft - the world has changed since 1940!

    Ing
     
  12. Jim Thompson

    Jim Thompson Guest

    I have an MSEE. But, indeed, theory certainly doesn't make
    application.

    ...Jim Thompson
     
  13. theJackal

    theJackal Guest

    Analog designers deserve every cent/euro/pound /etc they earn and
    probably more. Analog circuit design is a mind blowing process and
    you hardly have time to get bored as you'll most of the times be
    wiping the sweat of your forehead.
    It involves taking time to anticipate all the quirks or misbehaviours
    of your circuit. If you think you can make a circuit printed in a
    magazine to work or another that worked by making a small change you
    are inviting disaster. And even when it does , place it on a PCB and
    it might not. Attention to detail is everything .
    Opamps easy to design ?If you look at IC design ... things are even
    worse as neither does the SPICE simulation nor does the breadboard
    tell you what is actually going to happen. Breadboard measurements are
    influenced by stray capacitances from the scope and by connections
    within the circuit and SPICE has well known limitations. The real
    actual circuit depends on Physical parameters of the IC and layout of
    devices not to mention all the parasitics associated with the
    interconnects. So you are actually stuck with inventing every
    possible scenario.
    Its not that circuits don't obey the laws of physics but that the
    analog designer has to understand his circuit better then anyone else
    in the world!

    "Go easy on the whisky"

    theJackal
     
  14. PeteS

    PeteS Guest

    I inherited a non-working design from an external contractor. The
    switching power supply was not operating amongst other problems.
    Turns out he has/had virtually no analog experience, and had routed his
    sense and control signals back the controller *right under the
    inductor*, and had nice fat vias under there and the FETs too which
    added to the fun.

    Without a thorough understanding (and that's made up of education,
    blood sweat and tears) of analog effects, such things will bite very
    hard. If you think you'll enjoy figuring out major layout issues along
    with the design, analog may be for you. I admit I do a mixture, but an
    analog background makes doing digital much easier.
    (I remember a time when electronics was considered a subset of radio ;)

    As noted above, most of us started at 11 (or earlier - I had my first
    radio kit on sprung connectors when I was 6) and it's either in your
    blood or it ain't.
    Cheers

    PeteS
     
  15. PeteS wrote...
    One piece of advice for Steve, the market for analog designers
    who *use* analog ICs is much larger than the market for analog
    IC designers. Experienced, skilled analog circuit designers can
    do very well financially, especially if they develop expertise
    in a few niche areas, such as miniaturized switching supplies,
    high-voltage stuff, low-noise techniques, high-frequency digital
    radio, high-power systems, scientific instruments, etc.

    Reading Grey and Meyer is still a good idea, because a designer
    who uses analog ICs benefits greatly from an understanding of
    the design details and operating principles of those ICs. But
    as can be seen by glancing at my sample list of niche areas,
    analog circuit design in the real world consists of much more
    than understanding how to design a classic opamp IC. :)
     
  16. Fred Bloggs

    Fred Bloggs Guest

    It does not- all you ever have to do is look stuff up in books....
     
  17. Oh? I could probably name them *ALL*, like Analogue Devices, TI,
    National Semi, LT, Maximum, CSR, Apex, etc, like maybe 100 tops. For
    software there are literally 100,000's of companies. IC design costs BIG
    money. Even a fabless may be spending like, for Cadence tools, $50k+ per
    seat per year.


    Kevin Aylward

    http://www.anasoft.co.uk
    SuperSpice, a very affordable Mixed-Mode
    Windows Simulator with Schematic Capture,
    Waveform Display, FFT's and Filter Design.
     
  18. OTOH, hardware designers can work from their homes. With the tool-
    set upwards of six-figures and high-speed Internet, it's not much
    of a financial stretch to put the workstation in the home,
    anywhere.

    A couple of years ago most of the analog designers at this site
    were laid off (dumb move, but...). AFAIK, none had to move out of
    the area (there was only one company employing such designers in
    the state). One of the usual analog design houses built a design
    center where the engineers were.

    Hardware designers make a tad more than C++ programmers too. ;-)
     
  19. Jim Thompson

    Jim Thompson Guest

    Doesn't usually take that much money.
    That seems to be one of the growth industries here in Arizona...
    "design centers".
    Yes ;-)

    ...Jim Thompson
     
  20. I haven't looked at what our Cadence seats cost, but they're upt
    here. Of course they are floating licenses, so are used by more
    than one person. At one time I had almost six figures (over $80K)
    in FPGA development software. At about %15 of that for annual
    maintenance we dropped it after the project was done. SOme of this
    stuff is expensive!
    No surprise. With the Internet, one doesn't have to live where the
    boss does.
     
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day

-