Part of the  

Solid State Technology

  Network

About  |  Contact

Archive for September, 2014

Automotive Opportunities Present New Challenges for IC Verification

Thursday, September 25th, 2014

By Matthew Hogan, Mentor Graphics

The automotive electronic landscape is changing. Some of the changes are being driven by an increase in the number of safety and critical systems controlled by electronic control units (ECUs) [1], while others are from the explosion in in-vehicle infotainment systems. Back in 2009, it was estimated that even low-end cars had 30-50 ECUs [2], driving not only the complexity of the underlying hardware platform, but also that of the software used to control complex user interactions.

Some plans aimed at improving safety require complex systems to interact not only within a single vehicle, but also in concert with the surrounding traffic. One such system is the upcoming proposal (requirement?) of the US Department of Transport (DOT) to mandate vehicle-to-vehicle (V2V) communications [3] [4]. Many positioning pieces have been written on this topic, both for and against [5] [6] [7], expressing not only privacy concerns, but also technical challenges on product feasibility and implementation within such a short time. A number of established automotive partners have already begun research in this area [8] [9], working in tandem with the US government.

A common theme for these safety systems involves increasingly complex integrated circuits (ICs) and the need for exceptional reliability. While in-vehicle infotainment systems need high-reliability IC designs as well, out-of-specification performance for these ICs results in customer inconvenience (sometimes significant), but not the degree of concern felt if an air-bag controller, brake sensor, or other critical operational IC should fail. The harsh environment present in automotive electronics [10], combined with the high reliability expectations required for verification of these ICs [11], provides additional design and verification challenges that may not be of concern when designing and developing ICs used in gentler or less demanding scenarios. One common example where reliability in electrical overstress (EOS) environments plays an important role is protection and verification against time-dependent dielectric breakdown (TDDB) in interconnects (often called voltage-aware DRC [12]), or providing additional protection in the form of guard rings around devices to prevent possible latch-up scenarios. Both types of issues require larger design areas to avoid failure, but both are critical to mitigate in high reliability IC designs.

Understanding the reliability needs of these ICs can be challenging, particularly for companies just entering this market. Presently, the automotive market seems very attractive, with high growth and IC requirements for multiple new application areas. Questions on which electrostatic discharge (ESD) or EOS compliance standards need to be met are often answered in standards documents. What’s not exposed are the challenges, design trade-offs, and best practices used to achieve these standards. Speaking for myself, I have found that interactions with individuals who are experienced in these fields provide the greatest insight on many of these topics. IEEE conferences are one place where these experts congregate; another, somewhat less-tapped resource is that of “workshops.” These are often smaller events with a narrowly focused topic field. Two such upcoming workshops include the International Integrated Reliability Workshop (IIRW) [13] and the International Electrostatic Discharge Workshop (IEW) [14] [15]. The latter deals with more than just ESD (as its name might imply), including EOS and other issues ranging from IC design, test, and implementation all the way from the device to the entire system. Full disclosure, I am the general chair of the upcoming 2015 IEW event.

This is a time of incredible change in the systems and number of ICs used in our automotive vehicles. New players entering the marketplace, together with new expectations of how cars are built and driven, present an exciting time for those involved in the design, development, and verification of the ICs used in these systems. Everything from driver safety aids to “driver-less” cars is being implemented and trialed. How fast will new capabilities emerge, and how quickly do you think we will progress? How do you see the adoption of these new systems going mainstream, given the complexity and reliability verification challenges you see ahead? It will certainly be an exhilarating and interesting time!

References

[1] http://www.newelectronics.co.uk/electronics-technology/growing-number-of-ecus-forces-new-approach-to-car-electrical-architecture/45039/

[2] http://spectrum.ieee.org/transportation/systems/this-car-runs-on-code

[3] http://www.nhtsa.gov/About+NHTSA/Press+Releases/NHTSA-issues-advanced-notice-of-proposed-rulemaking-on-V2V-communications

[4] http://www.safercar.gov/v2v/index.html

[5] http://rt.com/usa/183208-dot-nhtsa-rulemaking-v2v/

[6] http://www.eetimes.com/document.asp?doc_id=1323617

[7] http://www.eetimes.com/document.asp?doc_id=1323968

[8] http://www.autonews.com/article/20140818/OEM11/140819888/u.s.-details-plans-for-car-to-car-safety-communications

[9] http://www.its.dot.gov/research/v2v.htm

[10] http://electroiq.com/blog/2004/05/the-changing-automotive-environment-high-temperature-electronics/

[11] http://www.eetimes.com/author.asp?section_id=36&doc_id=1322554

[12] Using Static Voltage Analysis and Voltage-Aware DRC to Identify EOS and Oxide Breakdown Reliability Issues, 2013 EOS/ESD Symposium, Matthew Hogan, Sridhar Srinivasan, Dina Medhat, Ziyang Lu, Mark Hofmann,http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6635948&tag=1

[13] http://www.iirw.org/home.html

[14] http://www.esda.org/IEW.htm

[15] http://blogs.mentor.com/foundry/blog/2014/09/07/friendly-but-shy-bears-and-other-eosesd-issues/

How to Survive the Perfect Storm of Changing Fill Requirements

Thursday, September 4th, 2014

By Jeff Wilson, Mentor Graphics

Even if you think you’re prepared for anything, the perfect storm can erupt unexpectedly, creating havoc and chaos. In integrated circuit (IC) design, we’re currently seeing the makings of such a storm when it comes to the growing complexity of fill. The driving factors contributing to the growth of this storm are the shrinking feature sizes and spacing requirements between fill shapes, new manufacturing processes that use fill to meet uniformity requirements, and larger design sizes that require more fill.

Before 90nm, fill was used primarily to improve planarity during the chemical mechanical processing (CMP) stage of manufacturing. Designers added only as much fill as needed to achieve the desired density targets. Because the file size was manageable, the runtime for adding fill did not add excessive time to the schedule, so designers typically waited until near the end of the design flow to run fill at the top level of the design.

In a typical 20/16/14nm design, fill can now easily exceed a billion shapes. At leading-edge processes, fill is used to mitigate the effects of rapid thermal annealing (RTA) and stress effects, and to improve the results of electrochemical deposition (ECD), etch, and lithography. These new uses of fill means fill strategy has shifted from minimizing the amount of fill in a design to maximizing the amount of fill added, which obviously impacts both fill runtimes and database size.

Two new fill techniques aimed at reducing the impact of these changes are the ECO fill flow and the hierarchical fill flow. I’ve discussed the ECO fill approach previously, so now let’s explore the hierarchical fill method.

Using hierarchy in the fill process can provide greater control for the designers and improve consistency throughout the design. Equally important, hierarchical fill can also reduce both fill runtime and file size, in a couple of ways.

The first technique is to use hierarchy within the fill shapes themselves. The key to this method is to adopt a cell-based approach to fill (Figure 1), which benefits both flat and hierarchical runs because the fill output now contains cells, rather than billions of individual fill shapes.

Figure 1. Cell-based fill allows designers to define a fill pattern (fill cell), using correct by construction techniques. These cells can be placed within the designated regions, reducing both runtime and file size.

The second method of using hierarchy to reduce the impact of increasing fill is to make use of the design hierarchy. If you have multiple instances of a block, it is logical that filling it once reduces both the time required to add the fill shapes and the file size. The cell-based fill approach not only simplifies coding for deck developers and reduces file size, but by placing cells with a correct-by-construction fill solution that considers all the spacing rules across all layers, insertion rate increases while concurrently reducing runtime and file size.

In addition to reductions in runtime and file size, there are a number of design consistency issues that benefit from using a hierarchical fill approach. For example, the requirement for a uniform fill around interconnect layers reduces the available spacing and increases the electrical impact of the fill. A hierarchical approach keeps the capacitance from the fill shapes consistent across different instantiations of the same block, ensuring that the electrical impact and timing of these blocks are the same throughout the design. This hierarchical fill consistency is also valuable in reducing the post-fill runtimes of both design rule checking (DRC) and layout vs. schematic (LVS) comparisons because the checks only have to take place once. Post-tapeout tools and processes can also take advantage of this improved consistency, such as manufacturing operations performed in the foundry, like multi-patterning decomposition, mask data preparation, and optical process correction (OPC).

Obviously, then, there are a number of positive aspects that this type of flow delivers, but of course, it is not without challenges and trade-offs. This approach does require additional effort over the traditional “push the button at the end” fill technique, because the designer must decide the appropriate level at which to add fill. As with any hierarchical approach, operating at too high or too low of a level in the design hierarchy significantly reduces the benefit. Too high, and the results start to resemble flat fill. Too low, and designers may reduce the benefit of hierarchical fill by making the interactions between blocks too complex.

All of these constraints complicate the placement of hierarchical fill cells, but as technology continues to shrink, and runtime and file size continue to grow, the question is not if your design flow will need to support a hierarchical fill flow, but when. Design methodologies for hierarchical fill range from a user-driven flow with a hands-on approach to a completely automated flow using new EDA tool capabilities.  Companies must determine the proper fill methodology for their design style, but regardless of the methodology they choose, there are a number of questions that must be answered. Designers must determine the optimum level of hierarchy at which to add fill, and define the correct halo size required to meet the gradient density rules. If companies want to implement an automated fill solution, they must determine if they have a complete solution available that includes both the EDA tools with the required capabilities, and a rule deck that supports the hierarchical fill flow and methodology.

Given how easy running a flat fill at the end of the design process is to use, it is logical that companies will want to use that approach as long as possible. However, I strongly recommend that your design team look ahead, and begin learning and planning for the transition to hierarchical fill. Familiarizing yourself with the methodology, acquiring the necessary technology to support your design flow, and analyzing your designs in progress are essential to a successful and timely implementation. Approaching the end of a design cycle with time-to-market pressures mounting is not a good time to find out that your flat fill is going to take over 24 hours to run and add billions of shapes into your design, significantly increasing the post-fill runtimes for both DRC and LVS and delaying your product’s introduction to market. That’s a storm that can destroy everything you’ve worked for.

Jeff Wilson is a DFM Product Marketing Manager in the Calibre organization at Mentor Graphics in Wilsonville, OR. He has responsibility for the development of products that analyze and modify the layout to improve the robustness and quality of the design. Jeff previously worked at Motorola and SCS. He holds a BS in Design Engineering from Brigham Young University and an MBA from the University of Oregon. Jeff may be reached at jeff_wilson@mentor.com.