NF: A lot of microwave engineers still prefer lab-based measurement techniques to simulation tools. What do you have to say to those folks?
LW: Lab-based measurements have their place; simulations have their place. It's not an either/or decision. Engineers will always perform both. Of course, I believe strongly in the value of simulation because of its flexibility and efficiency. Years ago, at Hughes, we built the phased-array antenna for the B2 bomber. The feed on that antenna had cross-guide couplers with crossed dog-bone-shaped slots coupling from one to the other waveguide. In the lab, we had a custom-machined test fixture with roughly 100 unique slot inserts of varying slot length so that we could measure coupling and return loss as a function of slot length. Days and days of measurements were performed to build up a design curve for that coupler. Today, I can run a parametric simulation in HFSS to fully characterize that coupler in just a few minutes! What's more, I can connect that coupler with models for the rest of the feed in a circuit simulation and optimize power distribution, phase, and bandwidth while including all 3D EM effects. Many of our customers trust simulations so strongly that they now avoid prototyping such a system.
Measurements are essential for characterizing materials and/or validating results of larger systems that may exceed the capacity or scope of simulation. For example, measurements are essential for characterizing nonlinear devices for which simulators do not exist or are unreliable.
NF: In your opinion, what major obstacles keep microwave companies from adopting design-automation tools?
LW: Most microwave companies use design-automation tools. The level to which they adopt, however, depends upon the design challenges they face. For instance, a company that produces microwave components like connectors, small antennas, couplers, or discretes will need field solvers with parametric capability. Likewise, a company that produces radio-frequency integrated circuits (RF ICs) or monolithic microwave ICs (MMICs) will require, at a minimum, microwave circuit simulation with layout and possibly some field solvers.
A company with larger design challenges will adopt simulation to a much greater degree. Consider a fabless semiconductor company that delivers a multimedia wireless-handset system-on-a-chip (SoC) that includes microwave radio circuits and baseband processing on a single chip. That device must operate in conjunction with the IC package, handset printed-circuit board (PCB), antennas, and other components. Complicating issues is the proximity of other potentially noisy circuits like switching power regulators, the system liquidcrystal- display (LCD) driver circuits, and other digital circuits. The company with these design challenges will adopt design automation software to a high degree. Radio designers use circuit simulation, layout, and field simulation. The system engineers use field solvers for IC package modeling and extraction of PCB parasitics. Full simulations for system integrity, radio receiver desensitization, and EMI are performed by combining RF circuit simulation, 3D, and PCB field solvers and digital circuits.
NF: What are the biggest failings in today's software tools in terms of the needs of microwave designers?
LW: Today's "biggest failings" are the same as yesterday's "biggest failings." Engineers will always need more speed, capacity, and accuracy! Over the years, software vendors have continued to deliver simulators that can handle larger circuits, greater nonlinearity, electrically larger geometries, and greater accuracy. In HFSS, for example, we released a new version with a completely automated high-performance-computing (HPC) option that allows engineers to divide a problem and solve it across a network of computers.
NF: How do you think software tools have most impacted the microwave and RF design process?
LW: The greatest impact is that simulation leverages the creativity of the microwave engineer. Engineers can try any number of design ideas and quickly determine their feasibility. Concepts that would be too time consuming or costly to attempt in hardware can be examined on the computer. Another result is that engineers can gain understanding of the operation of components, circuits, or systems by performing simulations. Finally, the cost and time to deliver microwave and wireless systems would have been prohibitive without simulation.
NF: How do you think they will revolutionize microwave design in the future?
LW: In the future, even greater leverage of the engineer's creativity will be achieved as companies adopt advanced computer hardware and HPCnot only to run larger problems, but many more iterations across the design space. For years, microwave designers have used Monte Carlo analysis to examine the design corners of microwave circuits. Now, those circuit simulations can include parametric electromagnetic simulations and entire systems can be solved for thousands of permutations across a network of computers.
The next phase will also include more of the relevant 3D physics. RF power amplifiers, for instance, dissipate significant heat. An engineer can run comprehensive harmonic balance simulation coupled to full-wave 3D models for the packaging plus circuit board parasitics and yet still not quite predict the performance of the amplifier due to the temperature distribution. Including 3D thermal analysis in this example would be a very valuable addition.
NF: Your firm's HFSS tool has been used in many research projects. What are some of its most unusual applications?
LW: The original design challenge that inspired HFSS was the quest for uniform cooking in a microwave oven. Ansoft's founder, Zoltan Cendes, began his quest to create a full-wave 3D finite-element field solver when he worked at General Electric. It was there that he discovered that a direct decomposition of Maxwell's Equations at the nodes of a finite-element mesh fails to predict fields reliably. Later, at Carnegie Mellon University, he invented tangential vector elements that solved this problem.
We have customers who use HFSS in the oil and gas business. Modern oil exploration uses magnetic-resonanceimaging (MRI) techniques when drilling deep underground to sense materials and detect fluid reserves. HFSS is used to design the RF coils and to simulate MRI detection effectiveness.
In the medical industry, we have worked closely with researchers at Duke University in their application of hyperthermia for cancer treatment. It turns out that the effectiveness of traditional chemotherapy and ionizing radiation treatment of cancerous tumors is greatly enhanced if the tumors are heated. HFSS has been used to design a phased-array antenna that concentrates microwave energy in the tumor to raise its temperature while leaving surrounding tissue unaffected.
NF: The acquisition of Ansoft by ANSYS created quite a simulation powerhouse in the Pittsburgh area. How has your group leveraged the ANSYS technology?
LW: ANSYS has a strong financial foundation and vision and we concentrate on engineering simulation. We have already leveraged the technology by combining multiple physics into microwave simulation. For example, the hyperthermia application mentioned earlier has been solved by computing electromagnetic losses in HFSS and mapping those to the ANSYS thermal solver to compute the temperature gradients. We have also linked ANSYS DesignXplorer to HFSS and Ansoft Designer. DesignXplorer drives HFSS in Design of Experiments (DOE) studies, optimization, and sixsigma design for manufacturing. We worked with one of our customers to use the ANSYS mechanical simulator to compute the deformation of a spaceborne reflector antenna due to solar radiation and then computed the farfield antenna pattern of the deformed dish in HFSS.
NF: Can you tell me anything about your plans to leverage other ANSYS expertise going forward?
LW: ANSYS has some fantastic bidirectional links to mechanical computeraided- design (CAD) frameworks like AutoCAD, ProEngineer, Dassault, and SolidWorks. We will leverage these links for our 3D solvers so that users can optimize geometries in HFSS and then immediately push the dimensions back into the native CAD framework.
We also will leverage the deep relationships that ANSYS has with customers in the mechanical-engineering world to develop even more advanced solutions that include electrical effects. Our R&D scientists and developers are working closely with one another to build on the synergies and techniques used in the various areas of physics and numerical methods.
NF: Engineering companies often balk at the cost of simulation.
LW: Simulation is only expensive if you don't have significant engineering challenges. Engineering firms spend tens of thousands of dollars on a simulation only if they have problems worth the expense.