Despite the dominance of digital, analogue components are holding their own – notably, within the medical sector. So could analogue technology be giving digital a taste of its own medicine?

Digital technology continues to push process technology to its limits, with products now appearing on 45nm design rules. But it’s a different story for analogue electronics. There has always been a gap between the leading edge requirements of memory manufacturers and the ‘sweet spot’ for many analogue components. The gap appears to be getting wider. Is this a good or a bad thing? It depends on the particular analogue components. There is no doubt that analogue components follow a curve similar to that of Moore’s Law. The difference is the speed at which progress is made. Robbie McAdam is group vice president, analogue products, for Analog Devices. “We’ve tracked data converter and linear products over a long time and you can see Moore’s Law being applied, but at different rates.” In McAdam’s opinion, data converters are, in general, shrinking to half their previous size over 24 months. “If you look at linear products – amps and so on – it’s more like 36 months because these devices are less dependent on lithography.” Like many others, McAdam believes the analogue ‘sweet spot’ remains at 0.25um, with power management devices suited to a 0.35um process. Mike Britchfield, product line director for Analog Devices, agreed with McAdam. “Analogue does follow digital, but at longer step sizes.” The reason? “One of the biggest issues with analogue parts,” he continued, “is that many of them are multi year, multi market products so you don’t get the benefits of a process shrink.” He contrasts this position starkly with that of digital companies like Intel. “One of the benefits which Intel gets from moving to new nodes is the ability to obsolete old parts. This just isn’t possible in the analogue world. While you can see new products transitioning, it’s not possible to move existing products to new nodes.” Nevertheless, there is work underway to support the manufacture of analogue parts on 90nm and 0.13µm processes. “But it’s more circuit innovation,” McAdam contended, “than pure lithography.” Britchfield gave an example. “In some areas where Analog Devices is focused, I would say process shrinks are attractive, particularly where there are families of high performance ARM based devices featuring high performance analogue. Going to the tightest geometry possible from the analogue point of view allows us to port reasonable high performance data converters.” But Britchfield pointed out that it’s ‘horses for courses’. “Many process technologies are simply not optimised for analogue components. Many processes are optimised for such factors as 1/f noise or for leakage. This means there is a trade off.”