The future for analogue is brighter than ever: Interview with Bob Dobkin

There are many people who have, for various reasons, discounted analogue technology, believing digital is the way to go. But the world remains analogue and that means there will always be the need for analogue technology to sit between the world and the digital domain. So what differences does Dobkin see between 'now' and 'then'? "There are many big differences," he said. "Circuits are more complex but, because we have the ability to simulate with good accuracy, problems can be identified before we go to silicon. But complex circuits with significant digital content can now be combined with analogue circuitry on the one chip." What have those advances enabled? "Higher performance analogue circuitry," he responded, "as well as complex power switching circuitry combined with feedback and telemetry. Data converters with high speed and high accuracy are in production at multiple manufacturers and high speed amplifiers with high accuracy are commonplace." Advances in analogue technology are also enabling new kinds of product. Amongst the areas which companies like Linear can now address are higher resolution data converters. "Data converter speed and resolution continue to increase," Dobkin observed. "A 20bit data converter is now available with a resolution of less than 1ppm, as well as a similar linearity. These benefits have trickled down so that 16bit data converters have now become the 'workhorse' of the industry, with 18bit devices positioned as the upgrade path." How have companies such as Linear achieved these levels of performance and has this entailed any trade offs? "We have been able to make these parts as a result of better process characterisation," he said, "and following many years of development. But the biggest trade off is development time; the market doesn't accept parts that bring big trade offs." One of the issues that separate the analogue and digital worlds is Moore's Law. While digital technology is, by and large, manufactured on leading edge processes, analogue often requires the benefits provided by the trailing edge. But are there analogue components which can benefit from more aggressive process technologies? "High speed data converters and RF ICs," Dobkin suggested. "Linear circuits are typically several generations behind the cutting edge of digital technology. Few linear circuits are being made of technologies smaller than 65nm," he noted. "In fact, many analogue circuits needs bigger feature sizes. "Meanwhile, high voltage and high current do not benefit from scaling transistors; relatively large sizes – such as 0.35µm for MOS and 0.7µm for bipolar – are the norm." Linear's products reflect this. Dobkin said the company's devices are generally made on processes ranging from 0.18 to 0.6µm. "But the norm for bipolar transistors is from 3 to 10µm." Does Dobkin see any benefit in using digital technology to enhance the performance of analogue circuitry? "I don't see digital techniques helping analogue to work on smaller processes," he asserted. "As a designer, you have a function and specifications to achieve. You then pick the process and features that will do the best job. You have to approach the problem with the idea of doing the best job possible with appropriate tools." But what of the future for analogue design? There are some who typify analogue designers as 'silver backs' who've been doing the job for 40 years; when they retire, the industry loses the expertise. Dobkin countered: "We have a company full of analogue designers all younger than me. We are hiring analogue designers from college and I'm continuing their training while they're here," he continued. "But we also hire experienced designers and have an easier time hiring them because Linear is known for analogue design. Other companies may find that task more difficult and therefore see analogue designers as a precious commodity." Do those designers have different skills to their digital counterparts? "The skills are the same," Dobkin contended, "understanding how circuits work. But the knowledge base is different; a discrete application designer must not only understand how the IC works, but also the characteristics of the discrete components. And, if they're designing ICs, they need a good knowledge of silicon. "We clearly know more about analogue circuitry and processes today than we did when Linear was founded. Analogue IC development is more scientific today and less of a 'black art', although there is still a good portion of 'art' involved." Is there more 'hand holding' involved in helping customers to apply technology? "We've been hand holding for more than 30 years," Dobkin retorted. "It's important to help customers apply the part properly. We look at support as part of the job – and the greater the degree of integration in the product, the greater the need for application support." And is he enthused about the future for analogue design? "I've been in analogue design for many years and it's still interesting. We are discovering new things in silicon and designing new circuits to make new products. Analogue is still needed, with ever higher performance. The future was bright when we started Linear Technology," he concluded, "and it's much brighter now." Bob Dobkin Bob Dobkin founded Linear Technology in 1981 with Bob Swanson, initially serving as the company's vice president of engineering. Since 1999, he has been Linear's chief technical officer and VP of engineering. Immediately prior to founding Linear Technology, he was director of advanced circuit development for National Semiconductor.