IBIS Survey Results 1-3-97 Research Question: Basic question we wanted to answer was "Would a >$100k hardware IBIS extraction solution be a viable product?" To look at this we wanted to understand "How are people creating IBIS models today?" Methodology: An open ended set of 8 questions were posted on an IBIS reflector site on the WWW. We had 11 responses, 9 were filled out in a way as to add valuable information, one was a request for information on IBIS applicability, and one was a request to show products at Design Supercon. Conclusions: Component suppliers (active digital I/O) would be the best target customer, because of the existing modeling groups and some budget. Syed Huq from National Semiconductor represents the best contact to investigate further. When asked for new approaches he listed "Automation in creating a model from empirical data and easier tools for model validation." We'd have to see what Syed actually has in mind. While component suppliers (active digital I/O) represents the best opportunity, I'd be concerned that we didn't get responses from the other big digital I/O companies (except SGS-Thomson), and what this says about the size of the market. CAE vendors who develop IBIS models could also match the product, but I don't see more than 5-10 potential sales from this area. PCB simulation users who develop IBIS models don't have the budget or manpower allocated today to support a >$100k piece of hardware. A simpler curve tracer with software to develop IBIS models would be a better start. A commercial spice to IBIS converter, supported, documented and simplified would be the best first step, but also has an upper price bound imposed by the availability of the free version. Detailed Summary: I. 3 catagories of IBIS model developers A. PCB Simulation users (3) B. Component suppliers (4) C. CAE vendors (2) II. How do you go about developing IBIS models? A. 7 of 7 derive IBIS models from SPICE B. 4 of 7 also mention deriving IBIS models from measurements 1. Two of these mentioned SPICE first then measurements 2. Two of these mentioned measurements first C. Measurement based models appear to be low frequency/without slew rates 1. Curve tracer - IV curves 2. IC tester D. Three or four possible candidates for investigating IBIS model development III. PCB simulation users A. No "modeling" budget 1. Designers, not modelers a) Not a modeling group b) Modeling is just a means to an end 2. Use existing equipment 3. Lack of manpower allocated to task B. Benefit of IBIS modeling to simulation user 1. Faster simulation of behavioral models enables being able to simulate full PCBs with layout effects. Selecting drive strengths of chip buffers is design variable designers can control. C. Curve tracer sets the expectation for cost and complexity of IBIS development 1. Not looking for n-outputs on a chip. 2. Package data comes from other sources D. HSPICE seems to be where existing SPICE models come from IV. Component suppliers A. Modeling people do exist and are budgeted 1. Size of budgets and strength of groups is unknown. B. Benefit of IBIS modeling to component suppliers 1. All are developing IBIS as a response to meet customer requests C. Two sub-groups 1. Active a) Looking for automation in creating models from empirical data and validation of models 2. Passive a) Looking for multi-coupled line behavioral models V. CAE vendors A. Modeling people do exist 1. Some have IC testers for other purposes B. Benefit of IBIS modeling to cae vendors 1. Builds applicability and sales potential for simulators