Advertisement

Blog

Power Management or Financial Management?

Those of us who live in the Bay Area (loosely defined as the area from San Francisco in the north to Silicon Valley in the south) are blessed with the confluence of technology and artistic creativity. The two forces collided in early May in what has become yet another tourist attraction.

A two-mile section of the Bay Bridge connecting San Francisco to Yurba Buena Island was powered up (so to speak) with a Philips Color Kinetics lighting system that rivals the synchronous water show at the Bellagio in Las Vegas, if only by its sheer scale. Before reading further, be sure to have a look.

This remarkable light show will play nightly for two years. The cost for electricity? About $15 a night, thanks to remarkable advances in customized power management. Each dimmable LED — and there are about 25,000 in the nearly 4.5 miles of cabling — will draw only about 1W each at maximum intensity.

A friend of mine said, “Wow, they did an excellent job of power management.” I said, “Maybe, but I think they did a better job of financial management.”

How many times in a day or a week do we hear or read the term “power management?” A lot. Every newsletter I get (and I get several) has multiple references to power management. It really is an overworked cliché. What engineer doesn't make power management (oops, I used it, too) a design priority? Engineers have thousands of options for managing their product power requirements. How can you tell if one option is really any better than the other for a given application?

One solution would be to marry the old cliché with a new term: financial management. Basically, it asks how much it costs you to get the result. If you're an engineer and that sent a shiver down your spine, don't worry. There's no need to cross paths with your finance department. This is easy, especially if you use multiple buck, boost, and/or LDO devices in your product.

Plot your total parts cost and lifetime volumes on the graph below. Depending on where you land, your excellent power management design may or may not be a good financial solution. If not, you should consider a custom Analog iASIC or an integrated ASIC. These are full custom designs embodying all the functions and features you use in your discrete, multi-IC approach. They're surprisingly inexpensive, with total NRE and tooling costs typically between $125K and $350K. Check with your preferred Analog ASIC supplier for details. The chart below includes amortization of these costs.

6 comments on “Power Management or Financial Management?

  1. eafpres
    April 11, 2013

    The shape of the curves makes sense but I note the product of component cost and volume isn't quite constant. Is this due to fixed costs of tooling the ASIC? Are the design costs on the purchasing side also included?

  2. Bob @ JVD Inc.
    April 12, 2013

    Hi eafpres,

    Yes,  your assumption is correct…Generally speaking, a custom integration for perhaps $3 of off the shelf components may cost $130K-$150K range for NRE plus tooling….while a chip that integrates  $20 of components may run $450K… so, yes the curves take into consideration that NRE plus tooling will be more for a more complicated integration. Again these are typical. YMMV

    Bob

  3. Bob @ JVD Inc.
    April 12, 2013

    Hello Scott,

     

    Good questions.

     

    1.     Are economics the reason to do a custom analog IC?  The answer is sometimes. But there are many other reasons that I will address in future blogs. Right now, we have several products in production that were justified purely on cost savings. One, an Automotive Bus Controller for ODB-II interface integrated over $20 of components into a chip that sells for just over $2

    2.     Are there going to be any changes over the production life of the program?  When a standard products company develops a chip, they don't care about this because they are designing for the maximum number of applications. Custom Analog ASICs require very close attention and communications with the customer to attempt to anticipate what if any system changes could occur that might be negatively affected by having a custom ASIC in the design. Sometimes this means deliberately not integrating something that might make sense to allow freedom and flexibility for the customer down the road.

    3.     Also, how does one know, with absolutely certainty, the lifetime dollar volume ? Of course, no one knows this for sure, but a customer who makes, for example, pressure sensors, has good market knowledge and can make reasonable expectations of a products lifetime.  If a customer's analysis places them on the line, the risks are higher, of course. But if in their first pass “best guess” they are up and to the right, then there is lots of margin for error with either future cost reductions to the cost of the off the shelf components or their lifetime volume estimates.

    4.     Large semiconductor companies with hundreds of analog designers…  would never take on a custom IC design with $500K of lifetime volume, probably not even for $5 Million.  You are exactly correct. They cannot afford to consume their resources for such a small ROI.

    5.     Only a much smaller company with substantially less broad experience will take on such a product development. So the development risks go up because of the talent resources. Don't associate smaller companies with talent risk. Our advantage of being in Silicon Valley is the ability to draw upon the large pool of now retired or semi-retired analog engineers who have spent decades at the big analog houses. I am sure you know, that in the analog world, you don't learn from text books… you learn from doing.  Our team averages 40 yrs analog design experience

    6.     I agree that there are certainly applications where a custom device makes sense. But I don't believe those applications are where saving money is the objective.   Exactly…like a senor calibration and conditioner for insulin infusion pumps.

    7.     I say go for it if you can't get it somewhere else or can't solve the problem another way, but be very careful about going down the custom IC route where the decision was based upon saving money . Analog ASICs are not necessarily the last resort, but you are right in your advice…be careful…be very, very careful with whom you engage for Analog ASICs.  I'll have more on this in the future.

    Bob

  4. eafpres
    April 12, 2013

    Regarding rationale, in one past company we had a design for an analog circuit to do active matching of FM antennas to allow small antennas to be inside of cell phones/music players (where the standard design uses a headset wire for the antenna).  For phone platforms, nobody would use it unless it became an ASIC.  So that choice was pretty simple.

  5. DEREK.KOONCE
    April 15, 2013

    Several nice points Bob. A few comments I can agree with are.

    1. A custom IC would probably be the choice these days for certain industries that plan out years of products and large volume – automotive is one. But most likely after full testing of the discrete version has been proven time-after-time.

    2. Small companies my do a custom to convert a large board of discretes into an ASIC for space savings and not necessarily for cost.

    Though the above seems to deviate from the topic of Power Management or Financial Management. Power management is a type of financial since, from a power supply designer's view, one has to weigh the extra cost of a more expensive IC and components relative to gaining that extra 1% efficiency.

  6. Bob @ JVD Inc.
    April 15, 2013

    Hi Derek,

    Thanks for the comments.  Surprisingly, (for us, at least), Industrial and Medical applications out number Automotive applications. ..Probably because the standard product Analog IC houses will do somersaults for the big volumes of the automotive market, making it more difficult for smaller boutique Analog ASIC companies to compete.

    Bob

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.