Executive Interview with Jim Rautio
James C. Rautio received his BSEE degree from Cornell University in 1978, his MS Systems Engineering degree from the University of Pennsylvania in 1982, and his PhD degree in Electrical Engineering from Syracuse University in 1986. From 1978 to 1986, he worked for General Electric, first at the Valley Forge Space Division, then at the Syracuse Electronics Laboratory. At this time he developed microwave design and measurement software, and designed microwave circuits on Alumina and on GaAs. From 1986 to 1988, he was a visiting professor at Syracuse University and at Cornell. In 1988, he went full time with Sonnet Software, a company he had founded in 1983. In 1995, Sonnet was listed on the Inc. 500 list of the fastest growing privately held US companies, the first microwave software company ever to be so listed. Today, Sonnet is the leading vendor of 3-D planar high frequency electromagnetic analysis software. Rautio was elected a fellow of the IEEE in 2000 and received the IEEE MTT Microwave Application Award in 2001.
MWJ: How and when did you start Sonnet Software?
JR: I started Sonnet in June 1983 to sell software I had developed for the amateur radio market. Then I used the company as a conduit for funding for my PhD studies under Roger Harrington at Syracuse University. Upon completing my PhD, I saw the opportunity to make the big jump and take Sonnet full time to commercialize the results of my research.
MWJ: Who was Sonnet’s first customer? What types of problems was the product originally used for?
JR: Sonnet is used for very high accuracy analysis of planar microwave and RF circuits. We had substantial early financial support from GE, HP, and RCA Sarnoff. Our first official customer was either Hughes or TRW (now Northrop Grumman) in 1989. We got a telephone call confirming a purchase order from one, and literally minutes later, we got a fax of a purchase order from the other!
MWJ: EM analysis was a new concept for many people back when you started. Was it difficult to educate potential customers back then?
JR: One of my most vivid early Sonnet memories was when, as a shy self-conscious graduate student, I gave my first presentation describing my technique at a small workshop run by Harrington and Sarkar (both Syracuse professors). There were quite a few very well known microwave engineers there. One came up and emphatically told me, “All this EM stuff is useless ivory tower academic junk.” At the time, he was pretty much right, we could do only very small problems on the recently introduced 4.77 MHz IBM-PC with 640 kB of RAM (my computer at the time). Several years later, he sought me out and told me that because of my work he had changed his mind.
MWJ: How well-prepared are the engineers entering the work force today when it comes to understanding microwave circuit theory and the use of EM tools such as Sonnet?
JR: Some engineers are very well prepared. There is a problem when, say, an analog designer with no microwave background is suddenly told that he is now an RF designer. This is difficult for everyone involved. And this is where one of our greatest strengths comes into play. Our support is by far the best in the industry. While we do not do their design for them, we will sit down with the less experienced designers and patiently provide a huge amount of guidance. We very much want all of our customers to be successful, and that takes both world-class software tools, and world-class support. As for preparing future microwave designers, we offer exceptionally motivating educational packages, especially for universities on extremely limited budgets.
MWJ: Sonnet was one of the first (if not the first) companies to offer a stripped down version of its product for free. Was this a radical idea at the time? How did the introduction of “freeware” impact Sonnet overall business?
JR: Yes, it was a very radical idea at the time within our niche. And it was a serious risk. What if it short-circuited too many sales? Then, when I first announced it at a special MAFET (Microwave Analog Front-End Technology program) session at IMS (International Microwave Symposium), everyone in the room got up and gave me a standing ovation. This is something people really wanted. Our first SonnetLite customer (when we opened the web site the next day, 15 June 1999) was from TRW. Several other EM software vendors grumbled pretty loudly, but we had copy-cats within a year. Today there are over 75,000 registered copies of SonnetLite in the field. And many full Sonnet customers are SonnetLite alumni, so I count SonnetLite as our most spectacular success.
MWJ: How has customer support for RF engineers using Sonnet products changed over the years?
JR: Other than in the number of support engineers, our support has really not changed. One thing that bothered me a lot in the eight years I spent doing microwave designs, prior to Sonnet, was long response times for support. “Your call is important to us,” was a seriously bad omen. So, first we put a huge emphasis on software quality (my business partner spent 10 years at Bell Labs on software quality control) so that we would not waste support time on poorly tested software. Second, we did extensive testing and human factors design on the interaction between the user, the software, and the documentation. We wanted to make that really smooth, so support would not be swamped with nuts-and-bolts questions. In fact it takes only 45 minutes for the average new user following the tutorial (Help->Getting Started) to start doing real work. Third, I put skilled microwave engineers on the support staff and I concentrated on retention. In fact, all support engineers we have ever hired are still with the company. Roughly 90% of our support calls are connected with an engineer when they call. That is even though we tell all customers that if they wonder about something for more than five minutes, they should call. Best support in the world for EM software, that’s important.
MWJ: Many advanced software products are so feature-rich that groups of “power users” often evolve. Does Sonnet have a similar community? Has your company been able utilize these power users through special support forums or live user group events to promote the sharing of information?
JR: We have an active Sonnet User’s Forum (Sonnet’s web site), where skilled Sonnet users from around the world share their insight. We keep a close eye on it so if there is any incorrect guidance, we can correct and clarify it immediately. SonnetLite users are especially encouraged to obtain help here. We generally recommend staying away from the plethora of un-monitored forums scattered over the internet as quality of advice can cover a wide range. We also hold regular SonnetLite training classes; check our web site for details.
MWJ: Where do the ideas for new features come from?
JR: New ideas come from anyone and everyone. We religiously log all suggestions from everyone in a database designed specifically for that purpose. When we enter the definition phase for the next release we go through the entire data base and organize suggestions by importance to our customers and by degree of difficulty to implement. The winners define our next release.
MWJ: So which EM software is best?
JR: I have seen forum postings asking this question, and answers are posted without ever knowing what the application is! Of course, it depends on the application. Our own software is best for RF and microwave analysis of planar circuits where high accuracy is needed. If you get into 3-D arbitrary geometries, then I have a couple other answers, again depending on the specific problem. CST Microwave Studio, whom we represent in North America, is a major answer in that area. If all you want is a quick, “good enough” answer, then it might be Sonnet, it might be something else. There is a video on our web site where I discuss this in specific detail. One specific I strongly recommend is that designers use framework software from vendors who encourage interoperability. To do otherwise locks a designer down to tools that might be sub-optimal for the task at hand. From the Sonnet point of view, interoperable framework vendors include Agilent, AWR, and Cadence. These companies have actively assisted our interoperability even though we might have some degree of competitive overlap. They support interoperability because it is in the best interest of their customers, and that is important.
MWJ: Could you discuss multi-core machines and how they will impact future designs?
JR: CPU clock speeds have stopped increasing, so now CPUs are going multi-core. The main limitation with multi-core for most applications is memory bus bandwidth. The numerical bottle-neck for Sonnet is matrix solve. We have to use a full matrix solve. If we use approximate iterative solves, we become an approximate EM tool and we don’t really go after that market. A full matrix solve has memory accesses scattered all over the memory map. It is a rat’s nest. We figured out how to straighten that out so each core has its own little section of memory and it pretty much stays there. Result: Dual core matrix solve is a full two times faster. Eight core is average seven times faster. Combined with new meshing, our latest release easily does 30 times faster and more on large problems. When a job takes two minutes instead of an hour, it totally changes how you do designs. Our highest end customers actually sport clusters of multi-core computers. Throw in another 10 times faster…there is no speed limit!
MWJ: What has been among the biggest advances to Sonnet software in the past couple of years?
JR: Last year it was the huge speed up I just discussed. The year before it was perfectly calibrated internal, “Co-calibrated™” ports. By “perfectly calibrated” we mean perfect to within numerical precision provided the port connecting lines are not over-moded. Perfect calibration can only be done in a shielded (i.e., in a box) environment where all ports have exactly the same perfect ground reference and where the Green’s function (i.e., the coupling between subsections) is calculated to full numerical precision. And Sonnet is the only EM tool that does this. Among the things enabled by Co-calibrated ports is the port-tuning methodology. We put ports anywhere, even in the middle of a high-Q narrow band filter. Then we use access to these ports to tune the filter or circuit after the EM analysis is complete. Bottom line: We optimize a circuit in real time but with nearly full EM accuracy. For example Dielectric Laboratories has taken their filter design cycle from two weeks down to one day. Seriously powerful.
MWJ: Accuracy for EM software is a big selling point. How does one know their software gives the correct answer?
JR: My answer might sound bizarre at first. I will personally 100% guarantee everyone that our software will always give the wrong answer! After you think about it, it’s obvious. Only an un-enlightened salesman would try to convince intelligent microwave designers that EM software gives the right answer. Engineers must assume as a matter of course that all software always gives the wrong answer. Our job as engineers is to figure out how much wrong the answer is, and then make sure that the error will not kill our design. I have personally performed and published a lot of results on EM analysis error. We know all the error sources and we know how to quantify error, minimize error, and otherwise control and eliminate error. It is like opposites attracting. Approximate EM tools, which have larger error, have to spend time talking about accuracy. Sonnet is the most accurate EM tool in the world for planar circuits, so we talk about error.
MWJ: Where are the areas of greatest strength for Sonnet in comparison to other tools on the market?
JR: The number one area we have emphasized for over 25 years is the highest possible accuracy. We have the most accurate tool for planar circuits in the world. As I described above, that is why we speak freely about accuracy’s opposite: analysis error. Next is the best support in the world for microwave software, bar none. That comes from my early frustrations with vendor support as a microwave designer. Hand in hand with support comes probably the world’s easiest to use general purpose planar EM user interface. And this is regardless of whether one uses Sonnet standalone, or together with Agilent ADS, AWR MWO, or Cadence Virtuoso. Another strength: Most of our technology is fully published. If you don’t believe my claim about perfect port calibration, for example, you can read all the details on how we do it in fully peer reviewed archival papers. Many of these papers are available on IEEE Xplore and on our web site. I have a hundred or so papers published over the years with all the details on nearly every aspect of our software. Sonnet is pretty much unique in this matter with all other EM vendors having few if any technologically detailed publications authored from within their organization, especially in the last 10 years. And then there are literally thousands of papers published by Sonnet customers. It is personally incredibly gratifying to me to see that kind of response steadily growing over the last 25 years.