Use best-in-class tools
The final critical element in the success of Agilent’s oscilloscope project was the selection and use of the best possible design tools. In this case, that decision meant use of Agilent’s own Advanced Design System (ADS) electronic design automation software for RF, microwave and high-speed digital applications.
According to Dave Dascher, lead hardware design engineer with Agilent’s Oscilloscope Product Division and 2012 recipient of Agilent’s internal Bill Hewlett Award, this decision “wasn’t because the engineers on the team were told to use their own internal tools. Rather, it was simply the tool-of-choice of the design team. Additionally, it offered us the integrated environment we needed to allow all of the different engineers on the design team to work together seamlessly, plus it gave us access to the InP fab.”
The project team used ADS to design and simulate the performance of all six of the new InP ICs. Good agreement between ADS simulation and measured results provided the team with confidence that its results were correct. Although, as Dave points out, “While it’s ideal to have good agreement between simulated and measured results, this is not the ultimate objective. You can spend a lot of time modeling everything to the nth degree to obtain perfect agreement between simulations and measurements. The ultimate objective, however, is to complete the design by meeting the required measured specifications as quickly as possible. By using ADS as our main simulation tool, we were able to meet this ultimate objective.”
ADS’s Momentum 3D planar EM simulator proved especially useful, enabling the design and simulation of the 32-GHz input matching network, and the 16- and 32-GHz anti-aliasing filters of the preamp IC (Figure 4). ADS’s optimizer was also employed on the matching network and anti-aliasing filters. By themselves, the amplifiers and filters didn’t have quite the desired frequency response. However, when optimized together using the ADS optimizers, the design team was able to achieve the desired combined response.
Figure 4. Highlighted here are the spiral inductors of the anti-alias filters on the preamp IC.
Momentum also played a key role in the design team’s physical modeling of key interconnects. As Dave explains, “When the first chips we worked on came back from fabrication, they were oscillating. This occurred because we had not adequately modeled the interconnect. The answer was simple: to increase our modeling accuracy. Our use of Momentum to model the interconnect was absolutely critical to achieving this goal.”
Using Momentum, the team was able to identify which nets were likely causes of the oscillations. Moreover, the tool told the team which parts of the design to modify to eliminate the oscillations.
Additionally, the project team used ADS in more generic ways, such as for tasks like probing issues with the ICs and DSP correction. In addition, ADS’s AC, Transient, S-Parameter, and Harmonic balance circuit simulators were used with models for the heterojunction bipolar transistors provided by the IC foundry for use with ADS.
Another use of ADS was demonstrated by Mike McTigue, lead engineer in Probing with Agilent’s Oscilloscope Product Division. Using ADS, he built an S-parameter model of the launch into a probe that had minor discontinuities getting from the RF coax line into the 50-ohm section in the probe. A model of the discontinuities was built out of multiple short transmission line segments. ADS optimizer was then used to tune the z0 of each transmission line segment with the goal of making the time-domain reflectometry (TDR) of the measured and modeled circuit the same (Figures 5).
Figure 5. This plot compares the TDR of the measured probe launch to the TDR of the modeled probe launch.
As Mike explains, “When the optimizer reached its goal, I ended up with a simulation model of the launch that I could use in system measurements or for de-embedding. As a result, using a best-in-class tool like ADS, just made my life a whole lot easier.”
The success of any design project is based on many variables and factors, both external and internal. While no two projects will ever be the same, there are certain best practices that can and should be adopted to stack the odds in a design team’s favor. As Agilent’s oscilloscope design team found out, those practices include: finding a synergy that works, being open to taking educated risks and, whenever possible, using best-in-class tools to accomplish your design tasks. Utilizing these practices helped Agilent successfully develop the world’s fastest and first 33-GHz real-time oscilloscope and the industry’s first and only 30-GHz oscilloscope probing system.
These practices are also used within Agilent’s research labs, where work is continually ongoing with Agilent’s various instrument divisions to develop and introduce breakthrough technology. It was within this state-of-the-art engineering environment that the ADS solution was first developed. Today, feedback from engineers engaging in this sort of leading-edge work continues to provide the impetus for ADS improvements. Such successes make it easy to see why adoption of a few key best practices can go a long way in helping any design team achieve project success.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for todayís commercial processor giants such as Intel, ARM and Imagination Technologies.