Innovation Challenges the Status Quo
We had the pleasure of speaking with Ms. Sherry Hess, vice-president of Marketing at AWR®. Ms. Hess recently joined AWR, bringing with her nearly 20 years of senior management, sales, marketing and international business development experience in the semiconductor and EDA industries. Prior to joining AWR, Sherry was with CebaTech, Inc., a VC-backed startup that specializes in EDA/electronic system-level (ESL) and telecommunication intellectual property (IP).
As vice president of business development, she was responsible for all business activities as the company transitioned from stealth R&D to successful product introduction. Before CebaTech, Sherry spent more than 15 years at EDA developer Ansoft Corporation, where she most recently held the position of vice president of marketing, with responsibility for internal and external marketing worldwide. While at Ansoft, she founded their European Operations as well as sold and supported their EM products for many years into key corporate accounts. Sherry holds both a bachelor of science in electrical engineering and an MBA from Carnegie Mellon University.
Hello Sherry, thank you for taking the time to talk with the Microwave Journal. First, I should disclose to Microwave Journal readers that you and I did work together when we were both at Ansoft. You moved out of the microwave field very briefly and are now back, working for a former competitor that seems to be paying considerable attention to an area that is a particular strength of your previous employer. Clearly from your background, you have a sense of the importance of EM simulation and its impact on first-pass success. From what I know about software development cycles, these projects must have been in the works before you came on-board. A key factor in your decision to join AWR must have been your discussions with senior management and what they saw in the marketplace that led to their decision to focus development efforts on enhancing the power of AWR’s EM simulation technology.
MWJ: You’ve just released a new EM product called AXIEM™. How is this simulator different from the current EM solver that is (or was) a part of the MWO suite?
SH: EMSight™ is and will continue to be AWR's EM simulation technology included as a standard component within our Microwave Office® design suite. EMSight, like the AXIEM technology, is a method-of-moments (MoM) solver, but that is where the similarities end. EMSight is a closed-boundary, gridded solver that assumes the structure is contained within a conducting rectangular box. For the classes of design problem that adhere to this assumption, EMSight is a very effective and efficient tool.
The AXIEM product, on the other hand, is an open boundary, non-gridded MoM formulation, which assumes the structure is unshielded or in a free space environment. While both EM technologies solve the same basic functions—Maxwell's equations—they were designed for different operating conditions/design flow use models; i.e., AXIEM is capable of solving problems an order of magnitude larger than EMSight.
MWJ: Does AXIEM replace the older AWR EM technology or is it an add-on option? Will the previous EM technology be discontinued? If so are customers being asked to upgrade to the new product?
SH: In an ideal world, one might think that there should be one EM technology that does all things for all people, but in reality there is no single EM simulator that will be right for all customers or all classes of problems. Speed, capacity, and accuracy are just some of the reasons why one tool may be chosen over another, whereas intuitiveness-to-use, technical support, and integration/openness to existing design flow tools are yet more.
Actually, Dr. James Rautio has a very informative video on the Sonnet website () that conveys why a designer would want to be able to use four different types of EM simulators for four different classes of problems. Together, AWR's EMSight and AXIEM tools address two of these four requirements/scenarios. So, EMSight is still an essential component within Microwave Office, and it will be enhanced and upgraded in the future accordingly. AXIEM is AWR's newest EM technology, which can solve problems that EMSight was never meant to address.
MWJ: I noticed that AXIEM can be purchased as a stand-alone product or integrated into the AWR design flow. Does that mean AXIEM works with other AWR products such as VSS and Analog Office in addition to Microwave Office? I assume this will be useful for applications such as system design verification and high-speed signal integrity analysis. Are these some of the intended applications?
SH: Yes, yes and yes. A founding principle at AWR is one of product openness and interoperability. For more than a decade now, we’ve been leveraging our proprietary unified data model (UDM) to readily extend the capabilities within all of our product lines. The AXIEM tool furthers the company’s vision of interoperability, and so yes, it can be added to complement all of the AWR product lines; the Microwave Office, Analog Office®, Visual System Simulator™, and Signal Integrity design suites. AXIEM’s advanced hybrid meshing technology increases its capacity reach and consequently makes it an ideal EM tool for signal integrity analysis of large RF PCB designs, for instance. Likewise, AXIEM’s unique thick metal capabilities also make it an excellent tool for 90nm and 65 nm RF CMOS problems, where line thicknesses are often in parity (1:1 ratio) with line widths.
MWJ: In terms of the underlying technology and the resulting capability, how does AXIEM stack-up to other commercially available EM products (planar and 3D)?
SH: The AXIEM technology was developed for 3D planar applications, the heart of electronic designs these days, and so we expect to stack up extremely well against other planar simulators for applications like RF PCBs and modules, LTCC, MMIC and RFIC designs. Generally speaking, AXIEM leverages the most recent advancements in EM solver and meshing technologies, as well as computer architecture, so its newness is, in fact, an advantage in that AXIEM’s speed benefits become more pronounced as the number of unknowns increases, and/or the accuracy requirements grow to be quite demanding.
AXIEM is not a general-purpose 3D solver and so it should not be viewed as a replacement for either 3D FEM or 3D FTDT codes that are quite good at analyzing highly non-planar 3D structures such as waveguides and antennas.
MWJ: Does AXIEM have any specific applications where it has a clear speed and accuracy advantage?
SH: Based upon our own algorithmic expertise and our beta customer designs, we expect it to perform quite well on thick metal simulations, i.e., silicon spiral inductors and very large/complex designs such as the analysis of a complete GaAs MMIC design. We also anticipate that as the number of AXIEM users increase so too should our list of known specific applications for which we offer exceptional speed and accuracy benefits.
MWJ: Microwave Office has been positioned for a number of years as an easy-to-use platform that integrates AWR and 3rd party simulation technology such as HSPICE® and Sonnet®, Zeland and CST. Does AXIEM represent a departure from a strategy of being a technology integrator to being a technology (simulation) innovator? If so, does this signal a shift in the direction of your R&D?
SH: As I stated earlier, AWR was founded on a principle of openness but that doesn’t mean we are not ourselves advocates of innovation. In fact, for more than a decade now, we’ve been innovating and releasing new technologies (UDM, iNets, ACE, RFA, and now AXIEM) that help designers do their job better. So, while we know our tools are among the best in the industry, we are also a company founded and grounded by design engineers who realize that it is impossible for one EDA company to have the absolute best of everything, for all applications, all the time.
Therefore, we ultimately recognize that it should be the customer who decides which EM, HB, SPICE, or DRC engine to use at any given time and for any given design. With that perspective in mind, our goal is to provide the best environment, inclusive of tools—whether they are our own or through seamless access to third parties—for customers to get their designs out the door quickly.
MWJ: Applied Wave Research has made a name for itself with numerous innovations in user interface (i.e. design entry, post-processing, etc.) and products that many engineers find very user friendly. How does AXIEM leverage the strengths of the AWR user interface technology?
SH: The AXIEM technology is integrated into the AWR design environment using the same EM Socket™ interface provided to all of our EM partners. This means that customers can access AXIEM by simply selecting it from a list of the presently-supported EM solvers: Sonnet’s em, Zeland’s IE3D, MEM Research’s EM3DS, Flomerics Microstripes, CST Microwave Studio, and Vector Fields’ Concerto. There is nothing new or additional to learn. There is no need to recreate preexisting structures. That’s it. It’s that simple.
To change simulators, a single mouse click is all that’s required and so the customer can readily and instantaneously compare speed, accuracy, capacity, etc. on the problem at hand. As stated earlier, since we recognize that there is no single EM simulator that is perfect all the time and for all applications, AWR’s unique software architecture empowers the customer to choose and access the best EM technologies the EDA community has to offer but with the added benefit that there is no need to learn multiple distinct GUIs. In this way, the majority of the user’s time is spent designing and optimizing performance of the design rather than on the learning curve for a new tool’s user interface.
MWJ: EM simulation as a design tool implies parametric analysis and circuit co-simulation. Is this the case for AXIEM? Do geometric structure changes due to parametric analysis hinder the intelligent meshing process that is being touted as one of the factors leading to AXIEM’s speed?
SH: The AXIEM meshing process is fully and intelligently automated so that there is no downside to using any of AWR’s parametric analysis capabilities. In fact, AWR’s novel EM extraction capabilities enable parametric structures to be readily mapped to AXIEM for optimization.
MWJ: It sounds like the hybrid mesh which uses a combination of rectangular and triangular meshing to accurately resolve structure shapes is well suited to addressing parallel and perpendicular traces along with arbitrary curved surfaces. Does the rectangular mesh have any speed advantage over a triangular mesh in terms of speed? If so, will structures with large metalized areas solve faster with this hybrid approach over a pure triangular mesh approach?
SH: The hybrid meshing approach that the AXIEM tool employs has definitive benefits over purely triangular meshing. Primarily the benefit is that a hybrid mesh offers greater efficiency and a significant reduction in the number of unknowns required to solve the same structure to the same degree of accuracy. Again, for designs from our beta customers, we’ve quantified this to be in the range of 5-10x that hybrid elements have over purely triangular ones. Similarly, the bigger benefit that a hybrid mesh has over a pure triangular one is for the accurate modeling of the current distribution on transmission lines. There are a number of publications that exist on this topic so I invite the reader to reference those is they’d like more information.
MWJ: Does the 3D mesh in AXIEM apply to any arbitrary plane or is it restricted to a plane perpendicular to the XY plane that defines the substrate and traces?
SH: This is a great question. In fact, I’d like to clarify what is meant by the words “planar solver,” since over the years, 2D, 2.5D, and 3D planar have all been used interchangeably and often to the confusion of the designer. AXIEM is a 3D planar solver; 3D in the sense that currents can flow in any direction (X, Y and Z) and planar in the sense that the substrate and traces are stackable and layered on top of one another. There are no limits on the number of metal layers, the thickness of metal layers, or, likewise, the number, size, or length of vias. Dielectrics are essentially stacked layers that are assumed to be infinite in the XY plane.
Okay, so now to answer the question. Yes, the initial release of AXIEM is restricted to support only facets perpendicular to the XY plane. This isn’t an inherent limitation of the solver technique and will be addressed in a future version. Zeland’s IE3D, CST’s Microwave Studio, Vector Fields’ Concerto and/or Flomerics’ Microstripes are all potentially good choices for arbitrary 3D geometries and can be accessed through the EMSocket.
MWJ: Does the 3D meshing capture coupling between 3D vias?
SH: This is another very good question, as EM tools on the market today differ in how they address via coupling. AXIEM can indeed handle any 3D planar structure and correctly account for all coupling in the X, Y, and Z directions. Consequently, there are no restrictions on the length of vias or thicknesses of conductors.
MWJ: You’ve recently introduced the circuit extraction tool known as ACE that allows engineers to incorporate layout-driven design directly into their circuit simulation. Does AXIEM simply provide a more powerful EM simulation technology for this layout-driven design flow?
SH: ACE™ is a unique technology in the marketplace. It is a fast tool that can be used for up-front rapid prototyping and optimization of a general design. It is not a generalized EM simulation tool but a technology that is built upon well-known underlying distributed/high-frequency circuit models. ACE has the advantage of being thousands of times faster than an EM simulator and just as accurate for structures that map well to circuit models.
For up-front design investigation, the AXIEM technology can compliment ACE in order to characterize ACE fractured interconnect for instances when no pre-existing circuit models are relevant. In this case, ACE and AXIEM were envisioned by AWR’s R&D team to work well together in order to streamline the product development cycle. Additionally, AXIEM is an excellent choice for back-end post-layout verification of the final design.
MWJ: Does AXIEM accept data from the major RFIC, advanced package or PCB CAD tools such as Cadence Virtuoso, Allegro or Mentor Expedition? Is there automation in place to help designers then incorporate extracted models or S-parameters into their circuit simulation?
SH: The AXIEM product leverages the same core technology as EMSight, and so it can import RFIC and PCB data through optional interfaces to Cadence products (Virtuoso and Allegro PCB), as well as Mentor Graphics tools (BoardStation, PADS, and Expedition). Additionally, AXIEM can also import GDS2, AutoCAD DXB files, and Gerber Mask files.
There is also an advanced automation capability within the AWR design environment known as EXTRACTion that drives EM analysis and the resulting S-parameters directly from schematic, thereby linking AXIEM and schematic-driven layout to simulation.
MWJ: Does the AXIEM product have 3D structure views and field visualization?
SH: AXIEM has very powerful visualization capabilities that provide for 3D viewing of the structure, its mesh, and currents. For the curious reader, it would be useful to refer to the AXIEM article in the September 2007 issue of Microwave Journal to see several 3D viewing/post-processing examples.
MWJ: Accuracy is often the reason designers resort to EM simulation. Given that AXIEM takes several new and innovative approaches toward meshing, Green’s function generation and solving the resulting matrix, how has AXIEM performed with regards to its robustness and accuracy?
SH: The electromagnetic tools marketplace exists, as you point out David, because of accuracy, or rather the failure of past approaches and techniques to deliver accurate-enough results, for the design of today’s high-frequency wired and wireless devices.
Needless to say, AXIEM’s accuracy of results is a must and a given in order for AWR to announce and release this product to market. Throughout the development process, we have worked with, and continue to work with, select key customers as well as academia to develop a portfolio of challenging real-world designs from which to test the innovations, algorithms, and mathematical techniques embodied in our new AXIEM product.
Based upon the existing class of problems and AXIEM results compared to measured data and/or other EM approaches, we are achieving the goals set forth for AXIEM in terms of accuracy and robustness, and along the way our confidence has been strengthened as we see AXIEM solving problems that other tools cannot.
MWJ: Any comments on what types of excitation ports are available in AXIEM? Are there any port limitations in this first release?
SH: The AXIEM technology has an auto-calibrated internal port that is quite novel in that it maintains its own current return path with relatively low error, can be further de-embedded to introduce even less error, and does not require any explicit path to a sidewall, such as edge ports.
This type of port implementation/excitation is ideal for complex structures such as PCB designs with many surface-mount pads. A prerequisite goal of AXIEM was to provide port excitations in such as way that it introduces as little error as possible into the EM solution, while avoiding the user requirement of de-embedding. AXIEM's internal port implementation has tested very well (low error) in all of the benchmarks run to date.
MWJ: The briefly mentions the importance of an accurate DC solution when the circuit contains active components. Are inaccurate DC solutions from EM simulation typically the case? If so, how does AXIEM improve its results at DC?
SH: It is not uncommon for today’s high-frequency EM simulation tools to break down as they approach DC. Some products actually provide for a hard stop at a few MHz while others incorporate a second technique, separate from the higher frequency technique, and then through mathematics stitch the two solutions together to provide results from DC to daylight. But often times, this stitching of two independent solutions introduces non-casual results and so both of these approaches lead to inaccuracies in the range of DC to a few MHz.
So you might say, what’s the big deal? The big deal is that accuracy of results down to DC per se is essential for getting the bias conditions right and setting the DC operating points for active structures. AXIEM has the ability to simulate to DC so there is no design hiccup associated with using its results in a harmonic balance and/or transient simulation within the AWR design environment.
MWJ: Does AXIEM address problems where the substrate is finite?
SH: Up until now I’ve been able to say that our new AXIEM product is terrific at everything, but here you finally got me. The initial release of AXIEM will not support finite substrates, but we recognize the importance of this feature, and I can assure you that it is in the plans for a future release.
MWJ: Well, it wasn’t my intention to “get you”. Clearly a lot has gone into this first release. What is the approximate list price for AXIEM?
SH: The US list price for AXIEM is approximately $30K for a node-locked license. We will also be offering special upgrade options and promotions to AWR customers who hold current maintenance contracts.
MWJ: Will current customers of AWR products receive AXIEM as part of their maintenance upgrade or is this considered a new product or add-on option?
SH: While AXIEM is considered a new product and priced accordingly, we will be providing evaluation copies to all of our customers with valid maintenance contracts at the same time we issue their AWR 2008 upgrades, which are slated for early in the new calendar year. For all of our active AWR customers, we’ll provide for an extended evaluation period for AXIEM, during which time we’ll offer a range of incentives/promotions for upgrading their Microwave Office, Analog Office, Visual System Simulator or Signal Integrity design suites to include AXIEM.
David, thank you for your insightful questions and for providing us with the opportunity to elaborate further on our newest product, AXIEM.
MWJ: Thank you Sherry and good luck with your newest product.