User Stories 

“Circuitlevel simulations took three days. With VerilogA they took 20 minutes—still too long to enable sufficient exploration of design alternatives. Using MATLAB and Simulink, we reduced simulation time to just one minute.”
Jun Uehara, Epson Toyocom
As manufacturers strive to pack more functionality into ever smaller devices, the demand for compact yet highly accurate crystal devices has increased. Crystal devices are essential to many of the advanced features in today’s consumer electronics—for example, they are used to implement clock functions, mobile TV tuners, surface acoustic wave filters, and gyrosensors used in cell phones, digital cameras, and automobiles.
As components are miniaturized, it becomes much harder to deliver stability and precision at high frequencies using purely analog designs. To solve this problem, engineers are turning to mixedsignal integrated circuits (ICs) that combine analog and digital circuits on a single chip.
Epson Toyocom used MATLAB^{®} and Simulink^{®} to develop a 16bit deltasigma analogtodigital converter (ADC) for a mixedsignal IC. Modeling and simulation in Simulink enabled Epson Toyocom engineers to make informed design decisions, verify their design early in development, and reduce the number of hardware prototypes, helping the team complete the project within a twomonth deadline and saving Epson Toyocom millions of dollars.
When Epson Toyocom engineers began developing the deltasigma ADC, they had no existing design that could be adapted or reused. Without a reference design, the team would have to evaluate a wide range of parameters, including input capacitance, output voltage, operational amplifier (opamp) gain, and slew rate. “We needed to run numerous simulations to sufficiently explore all the alternatives,” explains Jun Uehara, QI project, Epson Toyocom.
The team had just two months to design and lay out the ADC, which meant the specification had to be completed in the first two weeks. There was not enough time to use circuitlevel or VerilogA–level simulations to explore design options. Furthermore, to meet the project deadline, the team would need to rapidly simulate and verify any design changes as soon as they received the first IC prototypes.
Epson Toyocom used MathWorks tools for ModelBased Design to model, simulate, and verify the 16bit deltasigma ADC.
The engineers used Simulink to model the system architecture and the initial system design. The input capacitance, opamp gain, and other key characteristics were implemented as modifiable design parameters.
After verifying the basic functionality of the Simulink model via simulations, the engineers developed MATLAB scripts that programmatically updated the system parameters in the model. They then performed numerous simulations in which parameter values were systematically varied across their ranges to identify the combination of values that minimized thermal noise and maximized the signaltonoise ratio.
They used MATLAB to visualize the simulation results, such as power spectral density as a function of frequency. They also evaluated design tradeoffs using MATLAB. For example, the team weighed using a higher input capacitance to improve noise reduction against the additional expense of the highquality amplifier that would also be needed.
Once they had identified parameter values that fulfilled all the system requirements while minimizing costs, Epson Toyocom produced engineering samples of the design. Lab testing of these IC prototypes revealed a higher level of harmonic distortion than expected, which was reducing the signaltonoise ratio.
The engineers used Simulink to diagnose the problem, identify its cause, and gain deeper insight into the behavior of their design.
Using MATLAB, they developed a compensation algorithm to address the distortion problem. They incorporated this algorithm as a functional block in the Simulink system model. After running simulations to verify that the distortion had been sufficiently reduced, they updated the IC design with this fix.
Develop a mixedsignal integrated circuit with a new 16bit analogtodigital converter in two months
Use MathWorks tools for ModelBased Design to model, simulate, verify, and improve the systemlevel design