Profiles in Test: Eric Starkloff, NI
EDN: How long have you been working in test?
Eric Starkloff (ES): My whole career, which has been 16 years. Maybe slightly longer if you count some of my prototyping and troubleshooting during the undergraduate research I did developing ASICs for memory controllers on massively parallel processing systems.
EDN: How long have you been at NI?
ES: I started at NI as an application engineer in 1997 directly out of college. I had the opportunity to immediately begin working on customer applications and help our customers do some pretty amazing things. I was hooked right away, and continued my career in various roles within our marketing team working on product definition, enabling customer applications, representing NI within the industry and positioning our product offerings in the marketplace. I’ve been fortunate to be able to work on our key platforms for automated test during this period – from GPIB to our NI TestStand test executive software to NI LabVIEW system design software and PXI.
EDN: What is your educational background, and how did you get involved with test?
ES: In school at the University of Virginia, I focused on communication system design in terms of my EE concentration, but then got involved in a research project doing digital design with computer architectures. I was basically an ASIC developer and learned VHDL and digital layout “on the job.” The work was very interesting, and unlike many university research projects, we were well funded and had some pretty cool toys. I remember getting to spend $65k on a new workstation in the mid 1990s– a quad-processor Sun with 1 gigabyte of RAM, which was outrageous at that time. I always think of that purchase when I buy a new tablet or smartphone, which at this point probably has more processing power than that workstation!
EDN: What has surprised you most about your career?
ES: I am constantly impressed by the incredible spectrum of challenges in society that engineers are addressing and the pace at which technology has changed in a fraction of a lifetime. At NI, our business touches virtually every discipline of engineering, and I’ve been fortunate enough to work on applications ranging from the latest life saving medical technologies, the coolest mobile devices, new alternative energy systems, and even new frontiers in basic science. The ingenuity that engineers are bringing to each of these areas is inspiring to me. And this is true of test in general – as a test engineer you get to touch so many different aspects of technology and engineering disciplines. There is never a dull day.
EDN: You are involved with several industry groups, including PXI Systems Alliance and the Semiconductor Test Consortium. In general, what frustrates you about the ‘standards’ process? How could we do it better?
ES: Standards are necessary in our industry and can ultimately be a tide that raises all the ships of both vendors and customers. They work best when the line is clear between necessary pre-competitive infrastructure we all need and the areas where vendors will differentiate and compete. This is critical because our customers want multivendor compatibility and they also don’t want us all to reinvent the wheel each time; of course, they also benefit from competition and the innovation that it inevitably drives.
EDN: What makes your job fun?
ES: Certainly working with customers is at the top of the list. I get to visit with our customers throughout the world, and I enjoy learning not only about different applications, but also about different cultures. I also enjoy the variety of challenges I get exposed to; I have a short attention span, so this suits me well!
EDN: What are the next biggest challenges for the T&M industry?
ES: I do believe we are facing some major disruptions. It starts with cost of test. The devices we are testing are getting more complex and yet the cost to produce them is decreasing – both consequences of Moore’s Law. Despite this, test hasn’t kept pace and therefore continues to be a larger portion of the cost to invent and produce new technology. This has to change, and I believe software-defined modular systems, which also benefit from Moore’s Law, are fundamental to bending this cost curve.
EDN: How has the role of the ‘test engineer’ changed?
ES: First, test engineers have had to become more proficient in software development as test systems have become more and more software centric. They have also had to learn new technologies at an increased pace. Take wireless test, for example. This used to be the domain of the specialist – we all know a guru microwave engineer. But, as wireless has permeated nearly every device, more and more test engineers have had to also learn to test wireless among numerous other functions on their device under test.
EDN: Would you consider yourself a ‘software’ or ‘hardware’ person? Or does that division even realistically exist anymore for test?
ES: I don’t think it does. Software is increasingly important, but the need for good hardware design will never go away. I think some of the most interesting innovation happens at the intersection of the two. Take an FPGA, for example - a piece of hardware that can be reprogrammed with software.
EDN: Anything you would do differently if given the chance?
ES: I don’t believe in regrets. However, I would do many things differently based on learning from mistakes I’ve made. That list is a fairly long one…
EDN: Any advice for new engineers?
ES: Get as many opportunities to try new things early in your career – you might be surprised at what you love and where some hidden talents may lie. Also, don’t be afraid to take risks. Our CEO, Dr. Truchard, likes to say “Even a turtle has to poke his head out every now and then.”
EDN: Any funny stories?
ES: Here’s one that stays on topic: During my research project in school, we had just gotten our first ASIC back. We spent around $100k for the fab and received around 20 chips to test. We put one down on a test board I designed and were all excited to see a year’s worth of work come to fruition. Unfortunately, it didn’t work – we got all garbage data out. We spent several days trying other chips from the lot, debugging at the board level, and using a probing station for diagnosing the chip. We hit a lot of dead-ends and were very frustrated. I was sitting around a table brainstorming with the team and staring at the PCB when I noticed a stub of a trace on the top level of the board that didn’t look right. Sure enough, it was an error in the layout tool and caused a short. It just happened to be on the top layer of a multi-layer board so we literally took an Exacto knife to it, cut out the stub, and voila! - everything worked. The lesson – creativity and persistence are a must, but nothing beats dumb luck!