Obsolescing the Omnipresent (Why, when, how?)

Mar 1, 2007 12:00 PM
By Dhaval Dalal, Systems Engineering Director for Power Supplies, ON Semiconductor, Phoenix

We engineers are creatures of habit (even though we are loathe to admit it). Take the engineering schematics of your last three or four designs and go through them to look for the least common denominators. What you will find is what I'd like to call omnipresent signature blocks. These are your tried-and-tested solutions from the bag of tricks that you have built over the years.

However, if you take an additional few minutes to figure out the number of times you could have done without one or more of these tricks in a particular circuit, you might be amazed at the answer. “But,” you might protest, “using these tricks has allowed me to go to production in a hassle-free manner!”

If we repeat the exercise by taking a superset of all the designs in a certain functional area, say power supplies, we will very likely come up with a set of circuits or design methodologies that are repeated all over. Do we call these the de facto “standards” since they are everywhere? Yes, is a very tempting answer, but let's dwell on it a little further.

The essence of engineering is to build an optimal solution. The optimality has many dimensions such as cost, performance, time to market, quality and reliability. If we persist with the omnipresent for too long, we are not pushing the envelope of achieving optimum solutions. In many cases, this path starts a slow but inevitable drift toward mediocrity, which has dire implications in this global age. So you have to look for ways to make your pet tricks obsolete before someone else does.

Let's consider an example close to a power-supply designer's heart. If you have to do a generic power-supply design, you will more than likely start with the omnipresent UC384x PWM controller. If asked your reasons for this choice, your answers would probably be familiarity and cost. However, if you look at the cross-section of many new designs, the UC384x is not the controller of choice. So maybe it's time to look for alternatives. (One can substitute the other common omnipresent circuits such as a line transformer, or the TL431 or a bleed resistor for output regulation or a certain gate-drive technique.)

When is a good time to make this painful transition? The answer is before your competition does. A more helpful approach would be to use your antennae and probes to find out what is going on in the adjacent engineering world and markets and sense that impending change in the wind. The identification of advances in adjoining fields is an engineer's biggest trump card.

In the previous example, it might take a cursory browse through various IC manufacturers' websites to identify the more optimum alternatives. If you can get past the engineer-unfriendly home pages of many websites, you will find several controllers that are more optimized and better suited for today's requirements.

How often do you read the online journals or print magazines to sample the winds of change? How often do you contemplate potential changes to your tried-and-tested approach prior to starting a design? If the answer to these questions is never, you are likely already behind the eight ball.

How to implement these changes is another tough one. In this interconnected world, you have to look for partners who have a stake in helping you make the transition rather than doing it by yourself. If the required change is industrywide, you may need to drive collaboration by using forums such as IEEE working groups. Of course, expectations have to be realistic and any risk has to be shared proportionally between parties.

If you go to an IC vendor and say, “Give me something that works like a UC384x, pins out like that and, by the way, costs a nickel less,” you are unlikely to get what you want. On the other hand, going to the vendor's applications person and discussing your constraints and expectations may lead to a more optimal solution.

In many ways, the task of obsolescing your omnipresent solution is akin to getting the timing right in handing over the baton in a relay race. If you start to do it too soon or too late, you will be the loser.

Dhaval Dalal has been responsible for defining the strategy, roadmaps and technical content of many products for power-supply applications since joining ON Semiconductor in 2002. Prior to that, Dalal held various positions in marketing, engineering and research at TI/Unitrode, Digital Equipment and Philips Laboratories. Dalal holds a B.Tech. (EE) degree from IIT-Bombay, an MSEE from Virginia Tech and a master's degree in management of technology from NTU.

Acceptable Use Policy blog comments powered by Disqus