Improving the Flash value stream with new technologies: inline systems provide just-in-time programming of Flash memory devices, improving production line throughput and increasing yields.

In today's world of fluctuating production demands and quickly evolving products, factory managers have been forced to find new ways to limit costs and add flexibility. The emergence of the Internet has encouraged managers to look beyond their factory walls and consider the value streams that connect them to their customers and suppliers.

The Flash value stream is the pathway that a Flash memory device takes from its creation in a semiconductor fabrication facility to the programming stage and then to its installation into a product (Figure 1). In the same way, the intellectual property (IP) resident in the device flows from the original equipment manufacturer (OEM) to the end user. The intellectual property is the unique, proprietary programming that brings electronics products to life. For example, the program instructs a Flash device to act like a cell phone in one product and like a personal digital assistant (PDA) in another product. IP defines the unique set of features for that product, and it is often updated based on customer demand or feature pressures from competitive products. One advantage that electronics products have over other products is that value can often be added to the product without requiring changes in its physical design.

[FIGURE 1 OMITTED]

In a traditional Flash value stream, the OEM and the semiconductor manufacturer collaborate on the features of the Flash memory device. The OEM tells the semiconductor manufacturer about the programs that it wishes to store in a device, and the semiconductor manufacturer advises the OEM on the selection and programming of the device. The OEM sends the software containing the desired feature set to the programming center. The programming center burns the firmware into the devices and then ships a quantity of the devices to the electronics manufacturing services (EMS) provider. The EMS provider assembles and tests the products using the programmed devices as one component of the manufacturing process.

Figure 1 also shows the programming equipment manufacturer, which works with the semiconductor manufacturer to acquire the appropriate programming algorithms and understand the technical requirements of programming the device. The product's end users create constant change in the value stream by showing which new features and new products they would like to see in the product.

In the relationships between the companies in the value stream, each company has its own set of business drivers. The OEM is concerned about market share and time-to-market. The semiconductor manufacturer is interested in technical leadership that enables higher margins. The programming center and the EMS provider are driven by minimizing costs and working in volume.

These key business drivers define the character of the Flash value stream. In turn, constant changes in products, product features and programming algorithms drive frequent changes in manufacturing orders. New products are rushed to market as quickly as possible, and manufacturing must be performed without delays that would idle lines and defer time-to-money.

Because of the characteristics of the Flash value stream, inventory tends to build up in many places (Figure 2). The semiconductor manufacturer naturally has a large inventory of unprogrammed devices. Once the devices are programmed with intellectual property, they become different items with different part numbers. The programming center carries these devices as separate inventories until the devices can be delivered to the next partner in the supply chain, the EMS provider. The EMS provider carries the devices in inventory until they are installed into the product.

[FIGURE 2 OMITTED]

The problem with such inventories is that they are based on forecasts that are subject to change. Inevitably, excess Flash device inventory appears at the OEM every time a change occurs in the device firmware. This excess inventory must be reworked by the programming center, thereby introducing delays and cost. In addition to inventory build-up, companies also engage in other tasks that do not add value, such as sending purchase orders (Figure 2).

The EMS provider has similar problems. Although it tries to match production to actual demand, in the real world, this task is impossible because, if the OEM's customers want a new feature set, the firmware will be changed. This demand results in excess inventory, which must be sent back to the programming center to be updated.

Every time the firmware changes, a wave of disruption moves across the supply chain. The programming center has to reprogram its inventory of the suddenly outdated devices, and the EMS provider has to reset its production lines.

Because of the business forces driving the electronics industry, changes to firmware happen with great frequency. By repeated disruptions, the flow of value in the chain slows down, and inventory and rework time increase manufacturing costs. Perhaps even more significant, disruptions slow product time-to-market and lengthen the payback period for production equipment investments.

Technology to the Rescue

Fortunately for factory managers, new technology is emerging that can change the Flash value stream so that excess inventories and device rework are virtually eliminated. The new technology has two characteristics.

First, it uses new systems to program Flash memory devices at the point of their placement on the board. These systems attach to the pick-and-place machines and match the pace of programming exactly to the pace of manufacturing; the devices are programmed only as they are used. This just-in-time programming eliminates inventory of relatively expensive pre-programmed Flash devices. More importantly, the strategy eliminates the disruption in the flow of devices and intellectual capital that otherwise occurs every time the firmware changes.

Second, manual programming systems located at engineering design centers can be connected to high-speed automated programming systems at geographically dispersed manufacturing facilities. Communication is achieved via a network, such as an Internet virtual private network (VPN) or corporate Intranet. By sharing the same electronic programming architecture and using the same programming protocols, engineers can quickly communicate firmware changes and accelerate changes to production runs. Additionally, this connected programming strategy allows for centralized process control and monitoring of software revisions across the entire design and manufacturing supply chain (Connected Programming Strategy Manages Supply Chain Via Internet).

One alternative to programming at the pick-and-place stage is to push the programming to the end of the manufacturing process: the testing stage. However, the test equipment that is available today is not designed for high-speed programming. Especially with large Flash memory files, programming at the test station would considerably reduce throughput. Moreover, a defective Flash device would not be identified until the device was already installed, causing an entire board to be scrapped or reworked.

In contrast, the inline solution now available tests each device immediately after programming, to ensure that the device is not defective and that the programming was successful. Failed devices are moved to a discard container.

To keep pace with the pick-and-place operation, an inline programmer must be able to program and test a device very quickly. Recent advances in programming architecture have resulted in programming systems capable of the speed required for inline programming. For example, the manufacturing process may require a 64-Mbit device to be programmed in 0.375 seconds per Mbit.

In addition to throughput speed, an inline programming system must provide sufficient yield; for example, by programming four devices at the same time. Like any piece of production line equipment, an inline programming system must be evaluated on its output rather than on throughput alone. Job setup times and downtime are part of the equation. Fortunately, programming systems have advanced to the point that inline programming can now be achieved.

Another hurdle to inline programming was the need for the programming system to interact with surface-mount machines. Instead of being a separate station in the manufacturing line, the programming system attaches directly to the pick-and-place machine. Therefore, the system had to be very compact, and relationships between the programming systems manufacturer and the pick-and-place machine manufacturer had to be developed to assure compatibility.

A Case Study

Theory is one thing; real-world practice is another. For example, a major supplier of wireless handsets applied lean manufacturing principles and the new just-in-time programming technology to its manufacturing operation and realized a significant positive effect on factory economics. This OEM's experience makes an excellent case history for any manager considering a similar move.

The OEM had been using a programming center to program its Flash memory devices; one 16-Mbit Flash device to be installed on each board. Using four production lines, the programming center was able to program 163 units per hour. The firmware changed about six times per year, and each code change required about four hours of reset time. In addition, other delays occurred, such as two interruptions in device supply per year.

The OEM decided to streamline its value chain by moving the programming function in-house. It installed an inline programming system with Windows-based control software. Because the software was connected to the OEM's internal network with a gateway to the Internet, firmware changes could be easily communicated to the programming system, and the programming system could be reset remotely. Therefore, the time cost of accommodating a code change was only five minutes versus the hours of time previously required by the programming center.

Right away, the OEM benefited by avoiding certain losses: time and output lost during code changes, downtime when the programming center did not supply on schedule, and reprogramming errors. The monetary values of these avoidance costs are shown in Figure 3. Regarding operating costs, the new system showed only small costs for handling code changes, programming errors and inventory holding costs, as well as a significant savings in maintenance costs (Figure 4).

[FIGURE 4 OMITTED]

Before investing in new programming equipment, any factory manager should consider what would happen as demand increased. For example, at what point would an OEM need to buy a new inline programming system to handle increased demand, and what would that do to the economic equation? In this case study, the inline programming system had enough capacity--up to 467 units per hour--to handle significant increases in demand, even when device density was factored in. Therefore, the OEM also benefited from demand and density protection.

In this example of just-in-time programming and networked programming systems, the OEM realized a cash benefit of more than $615,000 from a one-time inventory burndown, a $210,000 yearly benefit from freeing resources for productive activities, and a $983,000 reduction in yearly operating costs. The cost of the inline system was a significant investment, but the payback period was only four months and the return on investment in one year was nearly 300 percent (Figure 3).

Conclusion

New approaches in programming can have a significant positive impact on factory economics. These gains can be expanded when these approaches are applied to the entire Flash value stream. By communicating firmware changes electronically, rather than shipping pre-programmed chips between supply chain partners, IP value moves swiftly through the supply chain with far less disruption. As a result, factory managers get innovations to market faster and at lower cost, while EMS providers shorten their time-to-market and control unpredictable costs.

Figure 3: Costs avoided by using networked, inline programming.

                            Cost Type
                                          Inventory
         Code     Programming   Damaged    Holding    Programming
        Changes     errors       Leads      Costs         Fee

value   11,200      95,918         0       123,157      981,260

                      Cost Type
                                   Total-
        Maintenance   Equipment   Operating
           costs       support      Costs

value    -228,000         0        983,535

RELATED ARTICLE: Connected programming strategy manages supply chain via Internet.

A new vision of integrated circuit (IC) device programming has been proposed that could make manufacturing more responsive and efficient. This networked or connected programming strategy is for managing the firmware supply chain, where programming systems are networked across the stages of manufacturing and the supply chain. The common element is HTML- or XTML-based programming control software provided by the programming system vendor.

This concept has recently emerged for several reasons. First, manual programming systems are now available that enable software designers to use the same control software as the programming systems on the manufacturing floor. These systems are the first to include an Ethernet connection to the LAN and, by extension, to the Internet. Now, a software change can be seamlessly communicated from the designer to the production line.

Programming requires complex data: algorithms, data files and programming profiles. The data are frequently changing as new software features are added, as parts change and as algorithms are updated. To prevent costly mistakes, manufacturers need a way to organize and direct programming across the supply chain. With this new approach, network capability provides a mechanism for central control of critical programming files. In addition, the control software can automatically download algorithm updates, by connecting to the website of programming systems vendors.

The real value is that company managers can be assured that programming is traceable and failsafe. Moreover, statistical information flows from the programming system to the user's control station, enabling the operation to become visible to operators, supervisors and managers.

The benefits spread to each member of the supply chain, because programming houses and EMS providers can easily reset programming job profiles, by receiving firmware changes via the Internet into their own control software package.

Obviously, this approach requires a high level of compatibility between software packages. For now, this interconnectivity can be accomplished by standardizing on a single programming systems vendor. In the future, standard data formats will have to be developed for communication between partners using programming software from different vendors or even from third-party vendors of supervisory control and data (SCADA) software.

Another driving force behind the connected programming vision is that the network control supports just-in-time programming. By enabling a remote operator to quickly send code changes to multiple manufacturing lines and to remotely reset programming equipment, delays in the manufacturing process are greatly reduced.

Larry Godt is president of Emerge Business Strategies LLC, Portland, OR; e-mail: [e-mail]. D. Judd Boone is product marketing manager for Data I/O Corp., Redmond, WA; e-mail: [e-mail].