INTRODUCTION High-density advanced packaging (HDAP) continues to be the promising “More” in the “More than Moore” approach for improved form factor, functionality, and integration of multiple dies built using different technology nodes. HDAP offerings from outsourced assembly and test (OSAT) companies and foundries are continuously increasing. However, the full commercial productization of such offerings will require the assurance of both an acceptable yield and correct (as intended) functionality. This assurance, like that for integrated circuits (ICs), will come from the availability of proven and qualified electronic design automation (EDA) tools and flows that can be used by the design houses to build HDAPs with the confidence that they are compliant with the foundry/OSAT requirements and recommendations. The need for and general concept of assembly design kits (ADKs) that provide proven, qualified flows for HDAPs has been previously discussed in multiple white papers. In addition, there have been analyses of the need for assembly-level layout vs. schematic (LVS) verification for HDAPs. Best practices for an assembly-level LVS process have been proposed, including the required inputs (data, formats, etc.), and likely hurdles and potential errors have been highlighted. There has even been discussion of how parasitic extraction could be achieved for packages. However, as HDAP technologies and flows mature, system-level designers want to know if package design rule checking (DRC), assembly-level LVS, and layout vs. layout (LVL) verification (die-to-package alignment, scaling, orientation, etc.) are sufficient to guarantee correct functionality and successful manufacturing of the HDAP. While this question may depend on how complicated the HDAP is, in general, the answer (for now) is no. As HDAP technologies become more and more similar to IC technologies, it is clear that, although the physical verification steps for HDAP may be considered good progress, they are only part of a much more comprehensive flow, one that must account for a more in-depth, system-level electrical analysis. Of course, at the same time, expanded EDA tool support is required to ensure fast, accurate, automated flows that ensure package designers can meet their market schedules and expectations. HDAP POST-LAYOUT ELECTRICAL ANALYSIS In the case of an HDAP design, the foundry/OSAT expects that each component is designed and validated to meet the required HDAP constraints and specifications. For an analog-based flow, the designer must simulate the HDAP system circuitry, including parasitics, to ensure it meets the intended performance specifications. For a digital-based flow, the designer must run static timing analysis (STA) on the complete HDAP system, including parasitics, to ensure it meets the overall system timing budget. From an EDA perspective, building an automated flow to support these checks/analyses provides assurance that these processes can occur in a consistent, repeatable manner while ensuring accuracy and minimizing runtime. In general, EDA approaches take one of two paths. SINGLE COCKPIT In the cockpit approach, an EDA supplier builds a single simulator infrastructure to support HDAP circuit simulation, parasitic extraction (PEX), and static timing analysis (STA). Although a single interface seems convenient, it forces the designer to use the same design tool for all components at all levels (die and package). This approach may be too restrictive, given that HDAP design and verification typically require the involvement of multiple groups with varying backgrounds and tool preferences. Although this approach would be useful when building “fully live” heterogeneous HDAPs (i.e., both die and package are under development simultaneously, and can both be edited for performance), this is rarely the case. More commonly, known good dies (which have already been taped out) are used to build an HDAP. TOOL-AGNOSTIC In the tool-agnostic approach, an EDA supplier enables the user to construct the needed system-level connectivity of the HDAP (including parasitics), regardless of which design tools are used to build any one die or the package. Once the system-level connectivity is available, it can be exported in the required format to any circuit simulation/STA tool to simulate or analyze the entire HDAP system. This approach introduces minimum disruption to existing tools/methodologies used for die and package design. This paper discusses the implementation of a system-level parasitic netlist process for the HDAP using the tool-agnostic approach.
Read full abstract
Oct 1, 2019
Jinn-Jong Sheu + 1
Open Access
