(Last year's presentation slides, links to talks: www.cpmt.org/scv/meetings/cpmt1010w.html).
Thursday, October 27, 2011
Please register in advance for this IEEE Workshop, using our IEEE Council's DoubleKnot registration site.
|
You may register yourself, plus others from your company/institution, for |
Location:
Cisco Systems, Inc.
Speaker / Company | Title and Abstract of Talk |
---|---|
Austin Lesea, Xilinx |
Soft-Error Architectural Vulnerability Factor for (Xilinx) FPGA Devices
Speaker Handout (PDF) Replay of Speaker's Presentation (35 minutes, WebEx) Soft error effects of designs programmed on FPGAs need to be properly evaluated in order to know if the system will meet its reliability and availability goals. While SRAM-based FPGAs are designed to have the most robust base storage cells in the industry, they are nonetheless susceptible to soft errors as bit flips in configuration memory may cause errors in a design until the bit is re-written with its correct value. There exists a configuration- time- and design-dependent nature to these soft error effects, and not all bit flips cause functional failures. In a Xilinx FPGA device from 2% to 10% of the bits affect the functionality. The microprocessor industry has measured soft error susceptibility in terms of a derating factor or architectural vulnerability factor (AVF), which measures the robustness of a design to soft errors. Thus far, there has been limited research into the estimation of this AVF value specific to FPGAs. We will describe our approach to estimating AVF at RTL based on resource and toggle rate estimations. Modeling is performed with beam-tested designs, and results are verified using an on-chip emulation environment. Providing FPGA designers with this AVF value at this early stage in the design flow will aid in the deployment of better mitigation techniques and meeting the reliability and availability goals for a system. |
Guillaume Warot, Lboratoire Souterain de Modane |
Alpha Emission Measurement by Gamma Ray Spectrometry
Speaker Handout (PDF) Replay of Speaker's Presentation (22 minutes, WebEx) This presentation will provide a brief introduction of the Laboratoire Souterain de Modane, a French underground laboratory which hosts ultra low level gamma ray spectrometer. This equpment allows the measurement of radioactive contamination of any material and the contamination level is then modeled o calculate the alpha emission of the material. Comparison between direct alpha measurement and gamma will be provided. |
Helmut Puchner, Cypress |
Correlation of Life Testing to Accelerated Soft Error Testing
Speaker Handout (PDF) Replay of Speaker's Presentation (26 minutes, WebEx) We present experimental data correlating acclerated Alpha and Neutron data to Life testing data. Life testing has been conducted at an Neutron acceleration factor of 9.22 as well as a cave setting with a Neutron level of <1e-5 n/cm2.hr. Good correlation is achieved for 90nm SRAM devices. Especially the cave life test data confirmed alpha counting experiments for the mold compound emission rate. |
Michael Gordon and Ken Rodbell, IBM |
On the Calibration of Alpha Sciences Proportional Counters
Speaker Handout (PDF) Replay of Speaker's Presentation (30 minutes, WebEx) By now, Alpha Sciences proportional counters are ubiquitous and are in use to measure the alpha-particle emissivity of materials used in semiconductor plants. Although the background counting rates are a few counts/hr, with long enough counting times, alpha-particle emissivities of a few cts/khr-cm2 are routinely measured. One parameter that is often misunderstood, and not accurately adjusted, is the discriminator setting on the Alpha Sciences proportional counters. We have made a low-energy alpha-particle source and have used it to determine the alpha-particle energy associated with various discriminator settings. The count rate of the source in the proportional counter is compared to that of the source in a chamber in air and after passing through the same mylar window as used in the counter with the use of a silicon surface barrier detector. Results of measurements and simulations will be presented. |
Nelson Tam, Marvell |
Investigation of Accelerated Alpha Testing with Vacuum
Speaker Handout (PDF) Replay of Speaker's Presentation (26 minutes, WebEx) This presentation will discuss the different characteristics of alpha SER rate as a function of the distance between source and DUT under vacuum and atmospheric pressure. In this study, 40nm SRAM SER rates are measured using a 0.1uCi Am241 sealed source in the form of a metallic disc that is approximately 5cm in diamater. The distance between the source and the DUT was varied from right on top of the test socket to about 1cm above the top surface of the socket. Two pressure conditions were tested where one was at atmoshperic pressure and the one was at below 0.2 torr. The observed SER rates were found to be varying by more than a factor of 2 and the amount of multi-cell upset events were also varied from a few % to over 10% of the upset events. In vaccum measurements, most of the variations can be attributed to the geometric effect. In the cases of the atmospheric pressure measurements, the shift in energy due to absorption by air makes for very interesting comparison to the vacuum results. |
Bharat Bhuva, Vanderbilt University |
Alpha-Particle-Induced Soft Error Rates for Flip-Flops Designed in a 28-nm Bulk CMOS Process
Speaker Handout (PDF) Replay of Speaker's Presentation (19 minutes, WebEx) With each new fabrication technology introduced by a manufacturer, mechanisms affecting soft-error rates become more complex due to close proximity of transistors, decreased nodal capacitances, and lower transistor drives. As a result, it is imperative that designers test their flip-flop designs for each new technology generation to verify the predictive accuracy of simulations and to identify any new mechanisms affecting the soft error rates. This paper will present results of an experiment for measuring FIT rates for various flip-flop designs in a 28 nm bulk CMOS process. Results show that soft error vulnerability at the 28-nm technology node is getting worse compared with earlier technology generations. Soft error mitigation techniques employed in the flip-flop designs show that FIT rates may be reduced to manageable levels through circuit design and redundancy. |
Gerry Maloney, Xilinx |
Flip-Chip Packaging Materials Alpha Characterization and Challenges
Speaker Handout (PDF) Replay of Speaker's Presentation (22 minutes, WebEx) Xilinx’s flip chip material set and alpha count requirements will be reviewed. Xilinx’s alpha count measurements of key flip chip materials will be presented as well as challenges of alpha count sample preparation. In addition, raw material, substrate and assembly supplier’s alpha count monitor methodology will be reviewed. |
Robert Baumann, TI |
Determining the Impact of Alpha-Particle-Emitting Contamination from the Fukushima-Daiichi Disaster on Japanese Semiconductor Manufacturing Sites
Speaker Handout (PDF) Replay of Speaker's Presentation (29 minutes, WebEx) We consider how the major radioactive isotopes were released and dispersed into the air, water, and soil following the Fukushima-Daiichi accident. The risk of contamination from uranium, plutonium, and other isotopes is considered, and the first ultra-low background alpha-particle counting measurements on wafers from two Japanese semiconductor manufacturing facilities. |
Brendan Dwyer-McNally, XIA |
Recent Advancements in Next-Generation Alpha Counting Technology
Speaker Handout (PDF) Replay of Speaker's Presentation (33 minutes, WebEx) Currently available alpha counters have background signals caused by alpha particles emitted from internal counter components. This ‘traditional’ background signal is generally many times larger than ‘ultra low alpha’ (ULA) activity levels (1 alphas/khr-cm2), thus requiring long counting times in order to extract the small sample signal amidst a much larger background – provided that the background is accurately known. A next-generation counter being developed at XIA is designed to suppress this ‘traditional’ source of background in an effort to drive measureable sample activity rates an order of magnitude or more lower. At the frontier of ultra-low activity measurements, new ‘non-traditional’ sources of background signal have been encountered and identified. We will review how XIAs counter suppresses traditional sources of background, thereby allowing direct measurements of samples with sub-ULA activity rates. We will also explore the identification of new, ‘non-traditional’ sources of background, their implications and methods to mitigate. |
SCV Chapter
Home Page |
How to Join IEEE |
Contact our Chapter Chair |
CPMT Society
Home Page |
IEEE Home Page |
Email
to Webmaster |
Last updated on