Q&A: Power plays

Saul Kupferberg is a second-generation member of the founding family of Kepco, which brothers Jack, Jesse, Max, and Ken Kupferberg founded in 1946. The company grew out of the involvement of three of the four founders in one of World War II's "Big Science" projects—the Manhattan Project. Much of the work they did at Los Alamos remains classified to this day. The work they were involved with required them to help invent and build electronic instruments. They had to build their own instruments because there really were no electronics businesses and no commercially available electronic instruments at that time. Indeed, the word "electronic" had not yet been coined.

Where do users of products such as yours have the greatest misunderstandings?

People get confused on how to use remote error sensing to compensate for voltage drops. We always find we are re-educating engineers about it. We even publish a nomograph of how to size the wires for the situation. It usually starts with a call from someone saying "Your power supply is not working."

What's your impression of the state of engineering and science education?

We find that US-educated engineers and others give little attention to analog and power design. Engineering graduates looking for work have never had a hands-on components or circuits experience or ever debugged a circuit. That's a big problem for our own talent needs and for our customers. The course they take has been simulation only. The general level of education is good, but the courses haven't focused on these aspects. In Europe, the analog-design and hands-on-experience portion of education is stronger.

What changes in specifications have you seen?

Switching supplies have become more and more dominant, combining low noise, a rise in PFC [power-factor correction], and harmonics regulation. Increasingly, we have to provide switching supplies that meet both conducted- and radiated-noise specifications. We saw the first switchers with PFC about 10 years ago. Now, everything above 50W has PFC and must meet the conducted- and radiated-emissions standards. Aerospace, sensor testing, and sensitive applications still need linear supplies.

Do you find that customers are increasingly asking you to do more?

Customers are asking us to do more of their work for them, especially in software applications. We want to help them, but at what point do we cross the line and are doing their job, as consultants? And helping them to this extent means there are other customers we can't help.

Why does the power-supply industry have so many vendors?

About 60 to 75% of the market is custom, so there are always those who do a custom unit and then hope to sell it as a standard product. Also, everybody thinks they can make a power supply, and, to some extent, it's true. It's make-versus-buy: They can make one, but can they make it over and over? It's an opportunity for companies such as ours. It takes a lot of time and energy to purchase components, and then you have to spend money to get past UL and other approvals. The BOM [bill-of-materials] cost is less than buying from us or others, but the total cost is not. There are so many hidden costs and other things that you could be doing with your engineering time. Lots of good engineers can build one supply, but to build tens or hundreds—that's another story.

Have any incidents caused a chuckle?

I was at a prestigious educational institution where we had delivered three 0 to 20V, 0 to 20A supplies. The postdoctoral student couldn't understand why he couldn't get 0 to 60V, 0 to 60A from the supply trio at the same time. I had to politely explain to him that he could connect them in series for the higher voltage or in parallel for the higher current—but not at the same time.