Is there a future for advanced analogue processing? The question was debated yesterday at the IET/GSA Semiconductor Forum in London yesterday.

"It breaks below 65nm", Nicky Lu, founder and CEO of specialty DRAM manufacturer Etron, told the forum, "I don't believe analogue will be designed as low as 45nm because of the signal to noise ratio. Below 65nm, dual voltage CMOS is the best approach," and the channel length can be relaxed. Is this the best way to implement true mixed signal chips?""

Responses to Lu's question were mixed: "45nm is difficult and expensive, but I believe it will be used for analogue", said Jens Kosch, CTO of X-Fab.

"Next year, I believe there will be a lot of work on 45nm for analogue", said Carlos Azeredo Leme, CTO of MIPS, adding "I can't say we have had much experience at 45nm."

"We are supplying 45nm", said Andy Buick of IBM, but we have very few analogue engagements at 45nm, and we expect analogue 45nm to be a year behind digital 45nm."

"For analogue and mixed signal design the drive is to use CMOS because of the advantage in cost, power and integration even when we are offering high performance", said David Srodzinski, CEO of Elonics, "for mixed signal we inevitably go down the CMOS route, it's a no-brainer."

Kosch disagreed: "For digital-dominated mixed signal if the performance is driven by digital content then we use the most advanced digital process, but in analogue dominated ICs, where we require high performance analogue processes, we need modular processes which are very flexible so they can fit either high volume or low volume products."

The effect of advanced of advanced processes is that "everything can be integrated in theory but does the business case make sense?" asked Peter frith, CTO of Wolfson Microelectronics, "the driving forces for the argument are: cost, risk, time-to-market, product differentiation, silicon process availability and capability, and technical issues like: Can a range of voltages be supported on one piece of silicon? Can the current be supplied? Can the heat be dissipated? Can the performance be achieved in the presence of other, potentially noisy, circuits?"

See also: Mannerisms, the blog of David Manners. Updated twice daily, it's the distinctive, entertaining, authoritative and never dull commentary on the semiconductor industry, from someone who knows. Sign up for the Mannerisms eNewsletter.