jump to navigation

Best Practices for Fabless Semiconductor Firms – Part I October 30, 2007

Posted by Jeff in Business.
Tags: , , , ,
trackback

Introduction

Best Practices are those methods, processes, activities, or techniques that have been found to be most effective at generating a desired outcome.  By implementing best practices, an organization may operate more effectively.  With best practices, over time and across an organization, the results can be substantial, with both improvements in execution and reduction in training.

Best practices are typically determined by review and analysis, and often with aspects of trial and error experimentation.  The steps can be broken down as follows:

  • Assessment of current practices
  • Applying quantitative benchmarks to the measurement of practices
  • Analysis of costs and benefits of practices
  • Selection of the best practice and revision of the current practice

We should note, however, that best practices are under constant review and revision.  Changes in technologies or markets can have a major impact and allow new practices to be developed that are more effective.  Specific areas of change within the semiconductor industry include:

  • Wafer size
  • Automation level
  • Outsourcing
  • Technology

In this document, much of which is based on Chapter 11: Best Practices for Fabless Companies, from the handbook “Understanding Fabless IC Technology“, we focus on practices for a fabless semiconductor firm that address operations execution, such as engineering, quality, customer support, production control, and finance.

In several cases, the articles which are referred to or summarized here are located in the members-only area of the Global Semiconductor Alliance (GSA) (formerly FSA).  In this case, the link will take you to the GSA site.  If you are a GSA member, use your member’s account to proceed.

While this document is focused on “fabless” firms, the same principles apply to “fab-lite” firms, also known as “asset-lite” firms, which are semiconductor firms that have shifted away from the approach of building their own fabs, mostly due to cost factors.  It applies to firms using outsourced manufacturing, including those activities that are outsourced through joint ventures.

An important area of best practices that is outside the scope of this document is legal practices, including IP Protection.  This topic may be addressed in a future posting.

Value Chain of Fabless Processes

Fabless Production Value Chain

 We have focused on the design, production, and fulfillment processes.  Not shown in the chart, but clearly important, are the processes associated with marketing, lead generation, sales, design wins, etc.

Product Development Best Practices

There are numerous sites with information on best practices for the design and development of new products, but these are typically focused on product engineers and the design process itself.  Here we will focus on those practices that have a direct impact on the migration of technology from the product engineering to the manufacturing operations phase.

Based on a presentation given by Gina Gloski in 2006, “Best Practices for Operations” (filed under GSA/Resources/Startups), three key issues appear:

  • Define requirements to support your end product
    • Are you shipping a reference design or software with each chip?
    • Are your targeted customers going to require early ISO 9000/2000 certification?
    • Software control -you need a document control system at RTL time
    • Need quality/manufacturing engineer to support component, PCB and supplier selection at time of RTL
    • Are you using specialized fab processes such as RF, GaAs, or SiGe?
  • Plan for the resources needed to support your portfolio
    • How many designs or new tapeouts are anticipated each year?
    • 5+ design tapeouts per year may justify design tool expenditures and in house operations team
    • 3-5 tapeouts-portions of operations may be outsourced to reduce cost and underutilization of resources
    • For 1-3 design starts per year, consider outsourcing both physical design and operations
  • Cost management
    • Costs will be composed of
      • Fixed cost: staffing
      • Variable costs: materials
    • Factors driving Variable costs
      • NRE, Gross Die, Wafer, Defects, Assembly, Transportation

Members of the Serus staff and our Customer Advisory board suggested the following additional factors as being important:

  • Keep in mind that choices in volume and wafer size, plus the technology to be used, can greatly restrict the number of possible suppliers, leading to difficulties in cost reduction
  • Market demand can greatly change the nature of the customer base
  • There are several choices for design and simulation tools, as well as several options for design from scratch vs. use of libraries

Operations Efforts and Resources

There are four areas that operations typically has ownership of, with a target of procuring and delivering product to customer on schedule.  The four areas are:

  • Manufacturing Engineering
  • Quality and Reliability
  • Business Processes
  • Financial Processes

The typical organizational structure around Operations looks like:

Organization Chart

Operations tasks and practices vary as an emerging company grows.  A model has been developed for calculating Order Of Magnitude (OOM) estimates of operations efforts and cost as a function of company revenue.  This is based on a white paper by Rakesh Kumar in 2005 “Operations Practices for Emerging Fabless Companies” (filed under GSA/Resources/Best Practices).  The model estimates the operations expenses, both manpower and other infrastructure, as a fraction of company revenue.

Revenue/Year

$1M

$10M

$100M

Units shipped/Year

0.1M

1M

10M

Wafer starts/Month

14

140

1400

Wafer lot outs/Month

0.7

7

70

Operations Resources Required

1-3

3-5

10-17

Ops Cost to Rev Ratio

15% – 35%

5% – 7%

2% – 3%

Assumptions used in this model are a chip size of 6 millimeters by 6 millimeters, 130 nanometer 1P6M CMOS logic technology, 200mm silicon wafers, a $10 ASP and $100K per person per year.  These are typical guideline numbers that can vary from one company to another.

Factors to keep in mind when applying this model include:

  • Supplier costs/fees
  • Infrastructure costs
    • Office
    • Benefits
    • Travel
  • IP Management costs
  • Reticules
  • Product qualification cycle
  • Logistics fees
  • Yield per wafer and die

Manufacturing/Test/Production Engineering

This group has a broad set of technical responsibilities such as maintaining yield and quality.  Formalizing and documenting these processes is the first step towards establishing a best practice in this area.

  • Definition and Documentation of product
  • Supplier interface to address Product, Design, Process, and Test-related issues
  • Yield monitoring and tracking
  • Yield enhancement
  • Resolution of yield issues
  • Resolution of Quality and Reliability issues
  • Quality and Reliability maintenance coordination and execution
  • Engineering support for customer returns
  • Failure analysis and debug
  • Cost reduction, which can involve test reduction, technology revision, and supplier relationship changes

An article in FSA Forum, “Supply-Chain Aggregation Boosts Predictability for Fabless Semiconductor Vendors” (Sherwani and Eltoukhy, September 2005) described some best practices of a product engineering team:

  • To maximize chip yield, a product-engineering team monitors yields on a regular basis and works closely with both foundry engineers and the chip-assembly provider to address any process drift or changes that may cause yield issues.

Another recent article, reprinted from FSA Forum, “Best Practices for Cost-Effective Test and Yield Optimization of Embedded Memories” (Pateras, December 2006), describes approaches to using Built-in Self-Test Engines (BIST engines) in chip designs.

Quality and Reliability

This function plays a key role in setting up, monitoring, and maintaining the quality processes within the organization.  Best practices can be set up around each of the following:

  • Equipment qualification (often as part of supplier audits)
  • Documentation management
  • Qualification documentation
  • Continuous improvement
  • Change control

An example of a specification for practices in the Semiconductor group at TI includes the following:

  • Defining the Q&R requirements in conjunction with the Quality organization for new product designs and redesigns that assure Q and R performance leadership.
  • Ensuring that:
    • customer requirements have been integrated into designs, and evaluated during design review(s)
    • New designs and redesigns meet all TI and customer requirements.
  • Ensuring that, prior to production release
    • All products meet design rules, meet all qualification requirements, are tested to and are in compliance with customer requirements, and
    • Data sheets have been published, and
    • Products meet all published data sheets criteria.
  • Establish and maintain an effective quality improvement process.
  • Establish and maintain a comprehensive internal audit program to ensure that both TI and customer requirements are met.
  • Ensure compliance to requirements for major changes

Another survey of quality issues and standards within semiconductor production described the role of international standards such as ISO 9001 as follows:

For the most part, the semiconductor industry has completed its ISO/TS 16949 and ISO 9001 implementations. Many large semiconductor organizations, contract fabs and assembly subcontractors that were registered to QS-9000 have successfully transitioned to ISO/TS 16949 despite the scope restrictions of this technical specification.

Because the scope of ISO/TS 16949 meant it only applied to automotive products, many semiconductor organizations couldn’t make the transition from QS-9000 to ISO/TS 16949. These organizations, including many assembly houses, chemical suppliers or mask-set suppliers, were forced to adopt ISO 9001 instead.

Business Processes

These deal with the process for recording orders, scheduling and fulfillment:

  • Production planning and control
  • Demand forecasting
  • Order management and fulfillment
  • Customer relationship management
  • Supply chain management and logistics
  • Supplier relationship management

Sherwani and Eltoukhy (2005) described the following best practice:

  • One of the biggest advantages a third-party supply-chain aggregator offers customers is the expertise in helping the customer make supply-chain decisions that optimize the manufacturing flow with respect to cost and schedule predictability, and reliability of first-time working silicon. The optimization constraints are tailored for each individual customer, based on their particular unit cost, schedule, performance, non-recurring engineering (NRE) and other requirements. Customers can make informed choices that lower cost and reduce risk at each step of the ASIC manufacturing process. An up-front systematic cost analysis optimizes expenditures at each step in the supply chain. Customers can then choose from the firm’s portfolio of qualified manufacturing, test and packaging providers. The firm’s broader experience base with each of these providers, greater than can be expected with an individual customer, gives each customer several options at each manufacturing operation for optimizing their particular chip.

Once again, use of a library approach can be beneficial.

Likewise, Kumar (2005) describes the role of operations in the planning and forecasting process:

Operations has a key role in pulling the demand forecast together. This is clearly a multidisciplinary function and can be coordinated by different pieces of any organization – Sales, Marketing, Finance or Program Management. Since Operations is the group that has to communicate the Build Forecast to the Suppliers and to manage the Order Fulfillment Process, they are an excellent candidate for this very critical function. Operations roles are summarized here:

  • Coordinate Cross Functional Process
  • Receive Sales/Marketing Input
  • Receive “Do-ability” from Engineering and Suppliers – Quantity, Schedule, Quality
  • Coordinate “Judged” Forecast Generation
  • Keeper of Updated Demand Forecast
  • Communicate Forecasts to Suppliers
  • Capacity Allocation Management
  • Supplier Management

Financial Processes

In executing these activities it is assumed that the operations group works with the company’s finance group but is held accountable for achieving the operations goals:

  • Purchase order receipt
  • Invoicing
  • Budgeting for capital expenses
  • WIP ownership and liability
  • Inventory ownership and liability
  • Management reporting

In turn, the finance group handles the following:

  • A/R, A/P, and G/L
  • Cash management
  • Currency hedging

Most of these functions are based on data in the ERP and order management systems, with the exception of reporting, which may be carried out in a separate OLAP system that is fed from the ERP system, or reporting may be carried out using spreadsheets (a very common practice, though not one that we have found to be a best one, see our document Defining the Software Product Category: Operations Management).

Another function of the finance organization is to determine cost and valuation approaches and establish standards for such.  The organization will maintain cost history information.  There are two main approaches to allocating overhead costs:

  • Percent of direct costs
  • Activity-Based Costing

The latter, Activity-Based Costing (ABC), is a method of allocating costs based on the activities and services that have been applied to the components, such as wafers and die.  This is in contrast with the approach of treating all components as equivalent, and simply dividing up the costs.  Organizations typically use ABC when carrying out strategic decisions such as pricing, outsourcing, and measuring business process improvement efforts.

Customer Support Processes

Efficient and responsive customer support is another key area for a fabless IC organization.  Customer support practices are widely discussed for manufacturing firms in general; see for instance, the Customer Service Zone web site.

A specific example of the organization of processes for fabless customer support is described in Kumar (2005) and included the following:

Customers interact with the firm through the Customer Service Organization, provides a range of services to the customers from a unified perspective.  Internally, however, the Customer Service Organization may actually be relying on other organizations, including:

  • (CS or Finance) P.O. Receipt and Acknowledgement
  • (CS) Delivery Commitment and Performance
  • (Finance) Invoicing and Payments
  • (CS) Freight and Logistics
  • (CS) Returns and RMA’s
  • (Engineering) Failure Analysis Reporting
  • (Engineering) CARs (Corrective Action Reports)

In turn, the Customer Service Organization is part of the Sales and Program Management organization, who assures that the results of Customer Service activity is communicated to:

  • QA:  to track outgoing quality, to update measurements of Early Failure Rates and Average Failure Rates
  • Product Planning: to support Engineering Change Orders (ECO’s) which address quality issues, and to provide inputs for demand forecasting.

Sherwani and Eltoukhy (2005) described the following best practice:

  • As part of the debug process, the organization should perform failure analysis on any return material request (RMA) and issue reports explaining the root cause of the problem and the corrective actions that need to be taken to avoid the problem in the future. Additional roles include performing product characterization at different process splits and across operating conditions to guarantee the quality of the product, and to assist the customer during the chip’s validation and debugging phase by providing bench measurement and focused ion beam (FIB) work if required. Finally, the organization should provide a full suite of die and package qualification tests, including burn-in, electro-static discharge (ESD), latch-up and highly accelerated stress test (HAST), along with second-level reliability testing if requested by the customer.

Typically the resources shared by the Engineering and QA groups include the following:

  • Test lab
  • Acid baths
  • X-Ray machines
  • Tester equipment

Many of the best practices that we have found focused on tracking returns and failures as the best way to learn about problems and apply corrections.  In other words, returns and failures should not be ignored, but should be used as for the basis for learning.  A recent HBS article (“The Hard Work of Failure Analysis“, Edmondson and Cannon, 2005) describes this same approach applied in the medical and health care field, leading to the same conclusion.

In Part II, we will look at the role of the Foundry Manager in a fabless firm, and at the flows of information in such a firm.  We will look at how IT infrastructure and software applications can automate and manage this flow.

Advertisements

Comments»

1. Srikanth - October 28, 2009

Excellent article


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: