In

by

IC Design Engineer, Novelda

Jørgen Andreas Michaelsen's profile picture

Novelda's takeaways from the 2015 International Solid State Circuits Conference

As I'm writing this, I'm returning home from San Francisco, having attended the 2015 International Solid State Circuits Conference (ISSCC).

As was pointed out in a conference keynote; analog design is all about learning. And in my opinion, the ISSCC is the perfect venue for it.

Before the conference started, I had the pleasure of attending a forum on high speed interleaved ADCs. In some respects, general purpose high speed ADCs are the competition. Our front end radar receiver can be viewed as a time interleaved ADC, with the current generation, the X2, sampling at close to 40 GS/s. Time interleaved converters have seen a substantial effort from academia and industry in improving speed and energy efficiency. It is interesting to see that our system is still competitive in terms of energy use. Not because we are more clever than the community at large, but because we have the advantage of controlling the radar pulse generator and the frontend sampling as one holistic system. This allows us to send, receive, and digitize the radar pulses bit by bit so to speak, rather than having to resolve the full dynamic range in one go. This touches on another interesting point from the forum, namely ADCs tailored to the application at hand, rather than treating the ADC as an ideal isolated device. This allows for saving power by relaxing the specifications that do not impact the quality of the final output of the system. This requires a detailed understanding of the system, signal processing, and application. Something we have been working hard on for our next generation of products, and something that will be a continued effort to allow for further optimizations. Another major topic in the forum was digital correction to undo some of the non-ideal behavior of the core circuits. A good analog design is still a prerequisite, but as one panelist pointed out, analog design is no longer enough. A deep understanding of digital signal processing is needed to mitigate the imperfections inevitably incurred by the analog frontend. Knowing what to optimize for in the frontend, and what is better left to digital post processing is a tradeoff that is not easy to make, but is an increasingly important factor for success. Intrinsic analog performance requires power and area, while digital processing is becoming ever cheaper as the process technology scales.

Outside the core technical content, a panel session was set up to discuss the balance between innovation and short term deadlines to deliver products to the market. For any company, shipping products is a key indicator of success. And innovation is important to be able to deliver value in the long run. However, innovation cannot be scheduled, but usually happens through focused work on important problems. An open atmosphere with an understanding that not all ideas will work out were pointed out by the panelists as important, with the key being recognizing as early as possible which ideas warrant further investigation. Also, assessing new ideas in terms of risk and potential value is important, where more risk is allowed when the potential benefits are greater.

Working towards tapeout deadlines also has its place for innovation as it helps to focus the innovation towards important problems, keeping the innovation relevant.

At the core of the conference are the technical sessions. Most of which are pushing technology limits through novel circuits and systems. This is a good opportunity to study the current trends in the circuit design community and to get an update on state-of-the-art. But most importantly, it is inspiring to study how circuit design challenges are solved across different applications. This has given us new ideas for advancing the performance of our radar system, and hopefully open up for new applications.

After almost a week of learning about the latest developments in IC design, I'm looking forward to getting back to designing circuits for our next radar project. But first, I have a new pair of skis, and a hope that there is fresh snow in the tracks around Oslo. I'm sure it will be a nice contrast to the vibrant city life of San Francisco.