Electronics research succeeds after chaotic start

A joint venture involving Analog Devices and University College Dublin shows how companies can benefit when they join forces with university researchers.

The partnership started with a careers talk, expanded into chaos theory and led eventually to significant cost savings at an electronics manufacturing plant in Limerick. Of such unlikely ingredients are collaborative ventures made.

Researchers at UCD's Department of Electronic and Electrical Engineering developed innovative software now used by Analog Devices BV to test its digital to analogue converter (DACs) products. The co-operation has worked well for both partners, explains Mr Tom O'Dwyer, Analog's DAC product development manager.

"We have been quite happy with the collaboration and will stick with it," he says. His company's testing costs have been reduced and UCD also gained by being able to apply students to a decidedly real-world electronic engineering application.

One completed his MEng on the strength of the project and another spent a year developing the complex test software, explains Dr Peter Kennedy of UCD.

Mr O'Dwyer recounted how he had been invited by Dr Kennedy to speak to students about careers in electronic engineering. The two began to chat afterwards. "I was describing some of the technical challenges we had and Peter said he would be interested in working on them."

Dr Kennedy's first impression was that the difficulties could be solved using chaos theory. "My background is in non-linear circuits and I am interested in chaos," he says. Though chaos was not the ultimate solution, the result of the effort, LEMMA (Linear Error Mechanism Modelling Algorithm), has proved a winner for Analog.

The company manufactures a range of semiconductor components for data converters, mobile phones, computers and disc drives. It employs 1,200 people at its plant in the Raheen Industrial Estate, from which it has operated since 1976.

DACs are "mixed signal devices", Mr O'Dwyer explains. They change digital signals, the ones and zeros handled by computers, into analogue sounds and signals which can be sent, for example, down a telephone line. The company applies strict quality control so each DAC is tested using a $1 million (£730,000) fully automated electronic device. Duds are picked out by a robot arm and pitched into the reject bin.

There are 256 checks needed to validate an 8-bit DAC, 4,096 for a 12-bit and 16,384 for a 14-bit DAC. Engineers at Analog developed software for the test equipment that reduced the number of checks per DAC while enabling the company to maintain quality standards. The challenge for Dr Kennedy was to run fewer checks without loss of confidence in the tests.

Analog agreed to fund research carried out by a Masters student, Mr Adrian Wrixon, supervised by Dr Kennedy during the academic year 1995-96. Mr Wrixon, now studying in California, developed a mathematical algorithm, a formula, which showed that just 100 checks on a 12-bit DAC were enough to guarantee quality.

The following academic year, a final-year student, Mr Paul Grogan, made the trip to Analog, this time to write the computer software that made use of Mr Wrixon's algorithm. This is now running successfully on the test rig in Limerick.

The new software gives a twofold improvement in the number of checks compared with Analog's own software, according to Mr O'Dwyer. "It is of value to Analog. It saves costs in the manufacture of DACs by reducing the amount of testing on each component." It has also been of value to Paul Grogan, who is now employed at the Limerick plant.

Mr O'Dwyer praises the partnership with UCD. "Peter has been strong in getting projects that are relevant to industry," he says. It is a success story that could be repeated a hundred times over if academics and business people realised the potential there was in collaborative research.