Man of letters

The STE software is getting rave reviews from ST engineers in the UK and France, as well as from contractors in the US who are doing work for ST's UK wireless LAN business unit. "I wanted our engineers to be able to focus on their test work, rather than compiling code," said Boz. Although STE remains an internal tool, Boz would like to patent it for wider use. The software can be easily reconfigured to handle a number of different applications—functional tests, debugging, soak testing, calibration, and production test.

Boz's journey to his present position has spanned geography as well as technology. Now the principal test systems engineer in ST's Wireless LAN Business Unit, he was born in Turkey and became interested in technology while attending professional high school—roughly equivalent to a cross between high school and junior college in the US. He scored well in a university qualifying exam, enabling him to enter a four-year electrical engineering program at Ankara University.

Boz's initial interests were in electronic cardiographic sensors for medical-imaging applications, but on completing his EE degree, he took a job in the factory-automation field, where he worked on a project involving process temperature control. That project, he said, provided a vantage point from which he could develop both hardware and software and appreciate the links between them.

A requirement for military service ended his factory-automation career, and on completing his military service, Boz joined the Turkish firm Mikes, which, in cooperation with Lockheed Martin in the US, manufactured electronic warfare systems for Turkey's F-16 fighter planes. Mikes, Boz said, needed someone with both hardware and software skills that could be applied to test-system development, and Mikes hired him as a test system engineer. "I quite liked the concept of test system engineering. I liked being involved from the initial system-level design stage and developing test-requirements documents. I liked working with systems engineers, digital engineers, and RF engineers."

Furthermore, he liked working with expensive instrumentation, which wasn't widely available to him in college. His university program, he said, was very strong on math and physics. At Mikes, he said, he augmented his background in theory with practical hands-on experience, and he discovered an affinity for working with instrumentation: "I felt quite comfortable getting any equipment to do what is required," after a brief review of the relevant manual.

During his tenure at Mikes, Boz met his England-born wife-to-be, who was teaching in Turkey and with whom he now has two young daughters. A Christmas holiday in England visiting his wife's family prompted Boz to consider seeking a job there. "I said to my wife, 'I'd like to practice interviews in a foreign land.'"

Test Engineer of the Year Zafer Boz escapes work to spend some time with his daughters.

According to engineering manager Terry O'Sullivan, Boz's boss at ST, Boz signed on with Synad in April 2000, soon after the company's founding. O'Sullivan himself joined the company a few months later. The original goal of the company, O'Sullivan said, was to develop intellectual property that would speed the development of RF integrated circuits, but the founders believed they needed to focus on an applications area to which Synad itself could apply the IP it had developed. The company decided on wireless LAN chips.

O'Sullivan said, "Over the years, the synthesis tool has continued to evolve. When we are redesigning or designing new chipsets for WLAN applications, that tool is the first port of call. It does reduce the cycle time significantly, which ultimately saves money, so it is quite a benefit to the company at large. In fact, it enabled us to be one of the first to implement a dual-band chipset, at the time when ST acquired us in 2003. Some other companies were trying and failing [to implement dual-band chips capable of 2.5- and 5-GHz operation], and we believe the synthesis tool was a reason for our success. We felt we were on the cusp of some major wins."

Engineering manager Terry O’Sullivan encouraged Boz to develop a test strategy that would allow rapid and accurate characterization and verification of reference designs.

He continued, "Zafer's role in that regard was to effectively implement the environments—and that's plural—to allow us to do evaluation/characterization work and then ultimately to allow customers to very easily run through a set of production tests to ensure that their solutions were actually performing as required. So, it's quite an involved program technically. Wireless LAN was new to everybody back in those days—it was only five years ago, but it seems like a lot more. So, we had to pick up a lot of new techniques in terms of how to check out modulation schemes, how to check out transmitters' spectrum masks and receivers' sensitivity, how to deal with regulatory issues, and ultimately how to ensure those cards not only worked in a very structured environment from a test point of view but would go on into operation in a system-wide sense."

That's when Boz and other members of O'Sulllivan's team got to put into practice their conception of the real world vs. the idealized world of a laboratory equipped with high-performance test equipment. "In the laboratory environment," said O'Sullivan, "everything is controlled. With the test equipment, everything is '50 V in, 50 V out'—everything is very controlled, and it all works reasonably well. But you take your device out of a test system, and it doesn't work so well, because you've got multipath transmissions, you've got noise, and you've got interference."

Boz's job, he said, was to initially examine the very deep technical requirements for evaluation and characterization as well as for production test in a structured test-equipment environment. Ultimately, he said, Boz had to develop "a structure whereby we could get a view of how a card would perform in a real system. It was not just 'set up some test equipment and play with it manually in a lab environment'—it was very structured, and it required a lot of innovation to come up with the techniques we ultimately used and depended on to get these cards through the process."

O'Sullivan continued, "WLAN at that point was very heavily focused on the laptop PC market, and we were developing chips you would load into a laptop PC and also into access points. Data rates involved there were anywhere from 1 Mbps to 54 Mbps over the air, and that was really stretching the technology."

Designing to IEEE 802.11b was the first challenge, according to O'Sullivan. "B operates to 11 Mbps in the 2.5-GHz ISM band, so we faced all the problems that plague unlicensed bands. One of the biggest issues we had actually was microwave ovens—we used to have times where we had to tell everybody at lunchtime, 'You can't use the microwave oven because we are testing.'"

He further explained that moving to the 802.11a standard brought its own challenges: "The emerging standard brought with it different issues and problems because it pushed the data rate up to 54 Mbps and operated in 5-GHz band, so you had the attendant issues of RF problems. A layman might think, 'Well, so what, you are only doubling the frequency.' But when you actually get down into the details of circuit design and analysis it's actually quite tricky. There are a lot of things to consider about how you route signals, how you avoid interference, how you screen yourself from your environment, how you calculate losses in your test setup—these are all actually quite important issues. If you don't take them into account in your test setup, all hell would break loose, and whatever you hand over would not be fit for the purpose."

In addition to technical challenges, the ST team was also facing marketing issues, which began to surface before the ST acquisition. O'Sullivan said, "It became apparent that in the PC market, the prices of WLAN chipsets were being pushed down so hard and so fast that it was becoming quite a bloody marketplace. Despite our best efforts working with ODMs [original design manufacturers] in Taiwan to control costs, we questioned whether this market made sense. So, we decided to focus elsewhere, and we chose the mobile market."

The ST acquisition made this choice particularly fortuitous, according to O'Sullivan. "One of ST's main strengths is to take a collection of die and other bits and pieces, like maybe some passive components, and to implement [them] as a multi," or multiple-die package. Those packages provide the small footprint and low power consumption necessary for use in PDAs or high-end enterprise phones, and as costs and power consumption fall further, they could increasingly find use in midrange and even low-end phones.

Of course, the technology wasn't standing still, and further complicating the picture was the emergence of 802.11g, which moves A's 54-Mbps data rates into B's 2.5-GHz ISM band. Emerging chipsets need to embody G capability while continuing to minimize cost, footprint, and power consumption.

"I can't go to a customer and say, 'I'll give you a much smaller and less power-hungry device, but it only meets half the 802.11 specifications,'" said O'Sullivan. "We refer to the standards as the table stakes. You can't come to the table until you meet them. Then, you have to demonstrate your superiority with respect to cost, footprint, and power consumption."

To leverage ST's multichip-package capabilities while meeting the technical issues, the division needed a test strategy that would allow rapid and accurate characterization and verification of reference designs. "Zafer has helped me push that," O'Sullivan said.

"At Synad it was amazing," Boz commented. "I was employed as a test development engineer, and I started working with evaluation cards and got involved in software designs as well as hardware designs, with the help of RF engineers. I was doing initial chip characterization, so I managed to do multiple disciplinary things. At Synad, I've worked quite hard because I always get good motivation from the people around me, who are working hard as well. I was very happy to get lots of experience with chip characterization, which I didn't have experience with in Turkey, where we were working with huge boxes for the military."

But test software development was an area where Boz proved to be particularly skilled. "I was always developing software applications for different purposes, like controlling RF synthesizers. One day, Terry mentioned that we needed software that we could reconfigure to change inputs and outputs and other parameters without the need to recompile code." Thus was born STE, which manages test racks containing instruments such as Rohde & Schwarz SMIQ vector signal generators and FSQ signal analyzers, controlling signal-generator parameters, acquiring signal-analyzer measurements, and supporting the export of measurement data to spreadsheets for further analysis and documentation.

Figure 1.  Within STE, a tree structure provides access to databases and external libraries. A watch window provides for changing parameter values on the fly.

Figure 2.  Device characterization requires consistent, repetitive testing under control of STE to determine, in this instance, transmission power relative to temperature and scaling code.

Figure 3.  Spectral mask testing is one of many functions that STE and test-scripts developed with STE can control.

Boz does caution that STE is for engineers: "You have to know what you are doing. It is very powerful but very dangerous, too, if you don't know what to do."

Not content with resting on his laurels, Boz is completing a master's degree program at Surrey University, where he is impressed with the practical hands-on experience students get. "I was amazed to see students actually involved in launching data-communications satellites." Boz himself is working on wireless subsystem design at Surrey, specializing, on the recommendation of O'Sullivan, in power-amplifier linearization.

Boz noted that Surrey is well funded when it comes to lab facilities. And in addition to having access to Surrey's RF measurement hardware, he also arranged with Agilent Technologies to obtain a license to Agilent's ADS electronic-design-automation software for RF and microwave applications. ADS, he said, will help him design the power amplifiers that embody the linearization techniques he is developing.

Boz's experiences at Surrey have motivated him to want to help develop more funding for Ankara University. As Test Engineer of the Year, Boz is entitled to assign a $20,000 grant to an engineering school, and he has designated Ankara University to receive the donation.

Boz and Professor Ahmet Kondoz discuss prospects for cooperation between Surrey and Ankara universities.

With the grant, said Kondoz, "We hope to start a relationship between Surrey and Ankara University, which I hope to continue with other funding from EU framework programs."

He added, "We hope to have key people from Surrey go to Turkey and give seminars, and we are exploring the possibility of some of their people coming to Surrey. We also want to set up students' prizes in Ankara University and pursue other ideas." Details, he said, are still under discussion.