TECHNOLOGY AREA(S): Electronics
OBJECTIVE: To develop trusted computer simulation software to accurately and quickly analyze the end-to-end circuit behavior of completely general time-frequency waveforms in complex non-linear RF circuitry.
DESCRIPTION: Anecdotally, a dozen fabrication cycles, 1000 or more engineers, and billions of dollars were required to develop the prototype for a modern cellular modem. Current commercially available circuit simulation software is not capable of fast, accurate analysis of the response of an entire complex linear and non-linear circuit to modern time-frequency waveforms. RF circuits are currently designed based on intuition derived from the analysis of many summations of steady-state functions such as sine waves or the transient analysis of circuit response with relatively small dynamic range. RF circuits tend to be designed starting from a steady-state analytical solution, followed by extensive trial and error fabrications. Small parts of the circuit are simulated for short time intervals and the results are combined based on engineering intuition. However components of RF circuits can respond in unexpected modes when subjected to wave forms which have formulations in both time and frequency. A simple example is pulses of sinusoidal waves. The response of filters and in particular non-linear circuit elements to these time-frequency waveforms can be substantially different than would be expected from a steady state wave form analysis. Even relatively simple 5G waveforms such as OFDM and CDMA modulation can be hard to accurately analyze. More general time-frequency waveforms where the frequency content varies with time and RF transients can dominate the response may be of interest for jamming and EW or the construction of LPI waveforms. Circuit simulation tools commercially available do not have the dynamic range to address these waveforms, the number of state variables required can grow exponentially, and computation time can take weeks for a single circuit for even a limited circuit time period under analysis. These simulation tools can have dynamic ranges on the order of 80 dB, while a dynamic range above 140 dB may be required, as well as the capability to reduce the number of state variables. An appropriate simulation will also require the capability to handle true time delay and memory effects in a physically correct manner. Macro-models will be needed to address these performance and computational speed requirements. These can include accurate behavioral models and reduced-order models. Fractional calculus and complex basis functions such as wavelets may be useful in constructing these macro-models. The software should be capable of simulating the 5G waveform response in a generic smart phone front end, with center frequency in the 1 to 5 GHz range, four orders of magnitude faster than a Spice simulation. The approach should be state-variable based and capable of the accurate simulation of arbitrary state variables (including multi-physics variables), physically correct true time delay, circuit memory effects, stochastic circuit and component variation, and greater than 140 dB dynamic range. Consider leveraging various macro-model techniques, such as behavioral modeling, advanced basis functions, tensor trains, and fractional calculus. Leverage published or commercially available macro-models. The simulator should provide a “dial an accuracy” capability to allow user tradeoff between accuracy and speed.
PHASE I: Demonstrate the feasibility of a software approach to linear and non-linear circuit simulation capable of simulating the detailed circuit response, sampled at any point in the circuit, tocomplex waveforms such as a 5G OFDM (Orthogonal Frequency Division Multiplexing) or CDMA (Code Division Multiple Access), pulsed ultra-wideband, multi-carrier, frequency hopping, and non-periodic pulsed frequencies. The simulator must accurately and efficiently handle circuit non-linearity, repeated transient behavior, and full duplex operation (with simultaneous very high and very low power signals). The feasibility of the approach will be supported by a block diagram of the software routines and analysis based on performance estimation from parameters reported in the literature for the various algorithms required. Results of research in the last two decades provide optimism that such a very high performance non-linear circuit simulator can be formulated. As an example, an advanced circuit solver approach (fREEDA), from North Carolina State University, is described in ref. 1, which has a dynamic range exceeding 160 dB in transient simulation. It is publically available for download (ref. 2). It has a physically correct true time delay capability (ref. 3), can accommodate arbitrary state variables and multi-physics variables (ref. 4), and can handle distributed networks (ref. 5). Reference 11 is a link to a Sandia Laboratory generated circuit simulator (Xyce) which addresses many of the same issues as fREEDA. It is also publically available. Both have GPL public licenses. One approach to this topic would be to combine features from Xyce and fREEDA, since both can be leveraged for commercial application under their public licenses. Establish and present a transition plan for the software package, with details of specific transition partners and consideration of software documentation, maintenance and customer service.
PHASE II: Develop a trusted software package capable of the simulation described in phase I. Determine the fundamental limits of macro-models proposed and the measures of uncertainty introduced by each. Formulate a plan to demonstrate the capabilities of the simulation. Develop a validation plan for the simulator based on specific experiments and other limited capability simulators or simulations with extensive run time (which would not be practical for most applications). Demonstrate the simulation and validate against the criteria in the validation plan. Document the validation in a journal article for peer review which will effectively advertise and promote the simulation capabilities to the professional electronics community. Deliver a beta version of the software, including source code, to a designated government lab for testing. Provide on-site support of the government testing. Develop and deliver a comprehensive transition plan to make the software available to the government and commercial market place, with detailed outline of the roles of transition partners, an updated business model, and updated market analysis. Develop and deliver a GUI (graphical user interface) with schematic capture integrated in the simulation. The software must be implemented in a common computer language such as C++ or Python. The simulator meeting these requirements will have significant capabilities not found in commercially available software and even (to our knowledge) in dedicated software internal to industrial programs. In particular the software will be capable of comprehensively simulating relatively long time steps and time delays (milliseconds) during (for example) an EW attack, of simulating with a scalable accuracy-vs-runtime tradeoff, of accurately modeling perfectly general waveforms in two circuits coupled at major distances electromagnetically, and easily encompassing the extensive device model libraries developed for other software products.
PHASE III: Phase III work will advance the beta software version to a robust circuit simulator for sale to commercial and military markets. The capability to accurately and quickly simulate the propagation of advanced waveforms end-to-end and to observe the effects of individual circuit elements will significantly reduce the cost of RF chip design and therefore the electronic system itself. It is expected to be of interest to chip designers throughout the RF electronics industry, to universities and government laboratories for analysis of innovative RF circuit and waveform concepts, to agencies regulating spectrum usage, and to the electronic warfare community. The expected transition path would be for the company to establish its own software maintenance, customer support, and sales capability; to partner with an existing larger commercial software company; or to sell the license to a major software company.
KEYWORDS: circuit simulator, non-linear circuits, complex waveforms, 5G
N. M. Kriplani, S. Luniya and M. B. Steer, “Integrated deterministic and stochastic simulation of electronic circuits: application to large signal noise analysis,” Int. J. Numerical Modeling: Electronic Networks, Devices and Fields 21, 303 (2008).; https://go.ncsu.edu/freeda-download; S. Priyadarshi, C. S. Saunders, N. M. Kriplani, H. Demircioglu, W. R. Davis, P. D. Franzon, and M. B. Steer, “Parallel transient simulation of multi-physics circuits using delay-based partitioning,” IEEE Trans. on Computer Aided Design of Integrated Circuits and Systems 31,1522 (2012).; S. Priyadarshi, T. R. Harris, S. Melamed, C. Otero, N. Kriplani, C. E. Christoffersen, R. Manohar, S. R. Dooley, W. R. Davis, P. D. Franzon, and M. B. Steer, “Dynamic electrothermal simulation of three dimensional integrated circuits using standard cell macromodels,” IET Circuits, Devices & Systems 6, 35 (2012).; C. S. Saunders and M. B. Steer, “Passivity enforcement for admittance models of distributed networks using an inverse eigenvalue method,” IEEE Transactions on Microwave Theory and Techniques 60, 8 (2012).; J.C. Pedro and S.A. Maas, “A Comparative Overview of Microwave and Wireless Power-Amplifier Behavioral Modeling Approaches,” IEEE Trans. Microw. Theory Tech. 53, 1150 (2005).; F.M. Barradas, L.C. Nunes, T.R. Cunha, P.M. Lavrador, P.M. Cabral, and J.C. Pedro, “Compensation of Long-Term Memory Effects on GaN HEMT-Based Power Amplifiers,” IEEE Trans. Microw. Theory Tech. 65, 3379 (2017).; S. Barmada, A. Musolino, R. Rizzo, and M. Tucci, “Multi-resolution based sensitivity analysis of complex non-linear circuits,” IET Circuits Devices Syst. 6, 176 (2012).; J.S. Ochoa and A.C. Cangellaris, “Macro-modeling of electromagnetic domains exhibiting geometric and material uncertainty,” Applied Computational Electromagnetics Society Journal 27, 80 (2012).; A.S. Yang, X. Chen, J. E. Schutt-Ainé and A. C. Cangellaris, “Adaptive Wavelet Stochastic Collocation for Resonant Transmission Line Circuits,” in Proc. 2017 IEEE 26th Conference on Electrical Performance of Electronic Packaging and Systems (EPEPS), San Jose, CA, USA, October 2017.; https://xyce.sandia.gov