-
- News
- Books
Featured Books
- smt007 Magazine
Latest Issues
Current IssueBox Build
One trend is to add box build and final assembly to your product offering. In this issue, we explore the opportunities and risks of adding system assembly to your service portfolio.
IPC APEX EXPO 2024 Pre-show
This month’s issue devotes its pages to a comprehensive preview of the IPC APEX EXPO 2024 event. Whether your role is technical or business, if you're new-to-the-industry or seasoned veteran, you'll find value throughout this program.
Boost Your Sales
Every part of your business can be evaluated as a process, including your sales funnel. Optimizing your selling process requires a coordinated effort between marketing and sales. In this issue, industry experts in marketing and sales offer their best advice on how to boost your sales efforts.
- Articles
- Columns
Search Console
- Links
- Events
||| MENU - smt007 Magazine
Creating the Intelligent Factory
December 31, 1969 |Estimated reading time: 9 minutes
In a typical high-mix environment, lean thinking cannot be applied only to the material flow, but also has to include the information flow throughout the factory.
Mattias Jonsson and Simon Sandgren, MYDATA Automation
Even with a high degree of general automation, production engineers are spending the majority of their time on information handling and system maintenance, activities that are non-value-adding from a strict lean perspective. In an intelligent factory, lean methods are applied to every aspect of production, both physical and digital. The intelligent factory is characterized by effective use of production data and digital information. Production data or information is registered only once; never reinvented; and made immediately available to all potential users in planning, kitting, testing, or purchasing. As a consequence, production engineers may have to register data of which they are not the primary user, nor fully understand. This requires better and more intuitive user interfaces with a higher ease-of-use than today’s. Training will focus on the production flow as a whole rather than as isolated tasks, enabling streamlined information handling. Understanding how incorrect data affects efficiency and creates bottlenecks at later stages is an essential motivation. In the same way as lean methodology stipulates that production teams cooperate in finding the most efficient way to transfer a partly-built product from one cell to another, teams must find the most efficient way to create and transfer information between cells.
Consider registration of component carrier types. The primary user of carrier type information is the planner. To optimize and predict pick-and-place assembly time, the planner must know, for each component, if it is available on reel, stick, or tray. Traditionally, this has been a manual, time-consuming task, using information sources of varying reliability. We should ask ourselves where this information is truly originating from, and how it can be reused once captured.
The earliest safe source of carrier information is at incoming inspection, when the material is unpacked and registered. Another safe information source is kitting, when components are loaded onto feeders, but this stage is too late, since planning includes the process of deciding which feeders to use. The fact that a component is physically loaded onto a feeder can merely verify information entered at an earlier stage.
From a lean information-handling perspective, the most efficient way of registering this data is at the incoming inspection, even though stockroom personnel receive no direct benefit in entering this data. Avoiding a tempting shortcut in one cell, the stockroom, means that substantial time savings can be achieved in later stages, in this example in planning and kitting. A fundamental requirement is that the system as a whole has the connectivity required to make this data available to the planner. This connectivity is achieved in two ways. The obvious way would be to use the same system in both the stockroom and in planning, but an equally good solution is having separate systems with a shared dataset. The details of the implementation are not critical, as long as connectivity is considered a key element when deploying the system.
This example highlights the importance of making reliable information available to the right user early. But the information itself is not what is important; instead, it is the decisions made based on the information. These will yield higher throughput and improved quality.
Any decision based on the factory-wide information set must be taken locally, as late as possible, since the production environment and its associated data are ever-changing. If you make a decision too soon it limits options and inhibits quick responses to changes. A prerequisite for just-in-time decision making is that process intelligence and process rules to a large extent are built into the assembly equipment itself, rather than being applied at an earlier step in the preparation phase. While a process engineer may have the training and experience to take into account nozzle size restrictions or vision limitations, it is not realistic to expect the machine operators to take over this responsibility when implementing just-in-time decision making. The only realistic solution is that assembly equipment has built-in intelligence to handle complexity in a fast-changing environment.
To implement the intelligent factory, machines need to send or share information within the facility. Electronics assembly machines must be looked at as part of the entire production process, not isolated islands. In the intelligent factory, data is transmitted between external systems and the assembly machine seamlessly. A modern machine should therefore be connected to the factory network and preferably use open and standardized interfaces to enable easy information exchange.
The intelligent machine should not be soley characterized by its ability to communicate. A great deal of information is continuously generated and used. To secure overall process quality and minimize errors, the information fed to machines must always be correct. Avoid generating redundant data. Each given set of data should have a unique location, like a central memory. The assembly machines in a line or several lines use a database where they share all relevant information. Despite the fact that new information is continuously added and updated, this database solution confirms that all machines, at all times, access the same information. If the database engine is a modern relational database, it will also enable easy exchange of data with external systems.
Traditionally, industrial machines are controlled by programming techniques where the outcome is deterministic, like an NC program. However, this does not allow a system to control a machine with just-in-time decisions based on factory events. With modern computers and software technology, it is possible to develop a more advanced and better-performing system by allowing the machine program to make intelligent decisions based on its memory bank of pre-programmed knowledge.
Simple mount sequences are not optimal assembly machine work orders. These should be formatted as a product description, with PCB dimensions, positions/rotations of components, part numbers, etc. Not specified is what feeder to use, at what location, for a certain part number. This information has already been retrieved from a shared database when the feeder was inserted, if the feeders have a unique identifier linked to a part number. The machine also collects information on package type with process-relevant information to understand the product description.
Upon start-up, the machine initiates a background optimization process, calculating the quickest mount sequence for the given feeder loading. Critical to such an optimization process is accounting for different restrictions on the mount order. A typical restriction is avoiding nozzles colliding with previously mounted components. This occurs often with 0402s and 0201s on densely populated PCBs. These considerations have been the responsibility of the process engineer, using in the best case software tools for modeling the assembly steps, or in most cases relying on experience. The intelligent machine must be able to model this, allowing just-in-time decisions.
Assembly optimization running continuously as a background process can recalculate strategies if prerequisites change, as when a feeder runs out of components and is replaced by a different unit in a new location, or for hot-swapping of feeders and next-job preparation while the placement system is running. The processing overhead required to achieve machine intelligence, compared to a more deterministic machine behavior, is well compensated for by non-stop production and instant coping with unpredictable events.
Monitoring and continuous process improvements have become a vital part in modern manufacturing technology. In the intelligent factory the machines must, in real time, provide detailed information on performance and status. The intelligent machine should have a built-in ability to store all relevant assembly events in its memory banks and make these accessible so that higher-level systems can process local information into a complete picture to identify bottlenecks.
The end result with the above design philosophy is a quicker and more accurate response to disruptive changes in the production. Not all situations can be solved on the local level alone, but relevant data is shared with neighboring equipment and to top-level systems if an issue has to be escalated to other decision makers.
Conclusion
Distributed intelligence combined with seamless data sharing will enable individuals to take a larger responsibility for carrying out tasks, and bottlenecks can be eliminated at the production floor instantly instead of escalating. The result is more products manufactured, at lower cost, and with minimal information overhead. SMT
Mattias Jonsson, product manager, software solutions and Simon Sandgren, product manager, assembly technology, MYDATA, may be contacted at mattias.jonsson@mydata.se; www.mydata.com.
Adding Intelligence at the Component Feeder
Pick-and-place systems have been with us since the beginning of surface mount technology. Early systems were adapted from basic robotics using a Cartesian multiple-axis machine with pneumatically actuated feeders. As time went on, these systems evolved to include more intelligence, enhancing the ability of these basic robots to handle smaller and finer-pitched components.
Machine vision added the capability to center components on their leads. Image resolution made it easier to identify, center, and place multi-lead fine-pitch devices. Similar improvements to inherent placement system capabilities allowed increased placement rates and product quality.
It still took decades before anyone started to improve the capabilities of the lowly component feeder. Over time, advancements in feeder technology took us from pure pneumatic-mechanical designs to electro-mechanical products. Until recently, the changes stopped at this design.
Recently, some suppliers looked at the task at hand and developed completely new concepts for tape feeders. These take the best of the electro-mechanical fundamentals and marry them with high-level intelligence designed to implement materials management into the lowly feeder.
Using intelligent feeders brings extended benefits to production equipment operators. They allow operators to load component information – vendor, lot code, component type, and quantity – directly to the feeder. Since the data follow the loaded tape feeder, the information can now be shared with the placement system once the feeder is installed onto the machine. Once loaded with both components and data, the feeder is loaded onto the docking feeder cart.
Intelligent feeders enable users to track component lots and consumed and remaining inventory. This eliminates guessing about how many components are left on a reel. And low-supply indicators illuminate to give staff time to either splice another reel onto the feeder or replace the unit with a new supply of parts.
Complete tracking information is available once production is completed, providing detailed reports on parts usage. Reports can follow the production lot or be included in serialized board records.
Using a system designed-in by the equipment manufacturer allows for a more proactive approach and ensures maximum efficiency and yields. Taking advantage of the feeders’ built-in intelligence, docking cart verification system, and integrated lot-tracking system maximizes profitability by reducing and eliminating waste. Adding intelligence at the feeder level allows you to reap the benefits of a leaner, meaner, and more efficient operation from purchasing to final assembly. SMT
Mike Foster, GM, Dynatech Technology, mike.foster@dynatechsmt.com.