|
The Company
Resonext Communication is a recognized fabless IC design company whose mission is to develop silicon ICs for Wireless Local Area Network (WLAN) applications. Their chipset consists of a radio chip for front-end communication and a base-band chip that implements algorithms for OFDM (Orthogonal Frequency Division Multiplexing) modulation. Both chips are designed to be manufactured using 0.18-micron CMOS process technology.
The Challenge
The management team at Resonext realized that device modeling was vital to the success of the company. They needed accurate RF models to improve their circuit design simulation and accuracy, which ultimately provides a competitive advantage.
The company decided to build a device characterization lab to test and evaluate device models from foundries and vendors as well as to build in-house models.
The device modeling group needed to:
- Build and extract in-house models for many CMOS devices, RF spiral inductors and MOS varactors.
- Customize extraction algorithms to fit specific process and device technology.
- Create RF extensions to standard models to improve design accuracy.
The Solution
Resonext Communication used IC-CAP software to control the Agilent 85122A Precision Modeling System while it acquired DC and RF measurements directly from silicon wafers. The measured data was then used to extract model parameters within IC-CAP.
The Result
For CMOS and BJT transistor modeling, model parameters from an external RF simulator were imported into IC-CAP and compared against the actual measured results. This allowed modeling engineers to provide instant feedback to the design team about the model's validity.
Next, Resonext engineers used IC-CAP to extract model parameters from the measured data to eradicate any discrepancies between the imported and the actual measured data.
Since many devices required special extraction algorithms, IC-CAP Parameter Extraction Language (PEL), similar to HP Basic, was used to develop their own extraction routines for important characteristics such as transconductance, gm, threshold voltage, Vt, effective channel length, Leff and so on.
"Sometimes the algorithms we used were not standard. The PEL programming environment allows us to develop our own algorithms easily and apply them to our devices," said Ali Rezvani, director of the Modeling and Characterization Division.
The resulting extracted model parameters include physical effects that can not be predicted by simulation software, thus significantly improving their model and design accuracy.
The modeling division also successfully developed accurate internal RF passive component models. The team again relied extensively on IC-CAP Parameter Extraction Language to develop their own extraction routines for many RF spiral inductors, MOS varactors, and resistors. The resulting models showed excellent agreement between the simulated and measured data.
 |
| S-parameter simulation (red) shows good agreement with measurements (yellow) |
"On-chip inductor modeling is extremely important to us," said Rezvani. "IC-CAP plays an important role in achieving accurate on-wafer RF inductor models because it supplies a powerful and flexible software environment to work with. We can develop different measurement and extraction algorithms for various scenarios. We are able to quickly take measurements and efficiently perform parameter extractions. This ability improves productivity and help our engineers meeting their project deadlines. IC-CAP helps us save engineering time and costs."
 |
| Modeling of Q-factor vs. frequency shows excellent fit between simulated (red) and measured (yellow) results |
For more information about Agilent Technologies Advanced Design System, visit our Website: http://www.agilent.com/eesof-eda/ For more information about other Agilent test and measurement products, go to: http://www.agilent.com/find/tmdir/
|
Contact Us
