By Stephen Rolston
There was a time when electronics manufacturers focused most of their efforts on developing ways to increase production capacity to meet consumer demand. However, when consumer demand languishes and the communications industry hits hard times, electronics manufacturers must look for ways to improve product quality and increase yields. This is especially true in the optoelectronics industry, an industry that has run from hot to cold and could heat up at a moment's notice.
One of the major costs facing optoelectronics manufacturing is from products never making it to market because of defects that happen during the production process. By reducing the time it takes to identify and fix production problems, companies can increase yields, improve quality and bring more products to market faster.
Automation Impact
Recent advances in automation technology have helped optoelectronic manufacturers establish repeatable manufacturing processes that enable them to bring products to market before their competition, adjust production to meet customer demand and reduce manufacturing time. However, implementing automated assembly and manufacturing applications is no panacea.
If manufacturing processes are not operating correctly, large amounts of scrap can be produced in a short period of time, causing both financial and opportunity costs. This is a problem for high-volume operations, as well as small-batch production, where each step may be a time-intensive, expensive process.
Automated Inspection
Quality management plays an important role in helping manufacturers cut costs, and automatic inspection is one of the best ways to achieve this. Notifying factory personnel in real time about production problems gives them the data to take corrective and preventive action (CAPA). User-defined business rules identify problems, and the manufacturing execution system (MES) application instantly notifies personnel by e-mail, phone, pager or handheld device so that they can take immediate action and prevent more serious quality problems. The MES system has pre-defined action plans that determine who should be contacted about specific problems. However, tools like this do not replace good manufacturing procedures and reporting — they facilitate them.
MES Systems
To truly improve yield is to feed continuous, real-time inspection data to an e-manufacturing or MES system that facilitates proper and accurate manufacturing processes (Figure 1). Generally, MES data can come from parametric process data from the manufacturing line or from engineering tests. This information can be input into the system by the operator; sent directly by the equipment via object linking and embedding (OLE) for Process Control (OPC), Microsoft's automated control standard; Generic Model for Communications and Control for SEMI Equipment (SECS/GEM), semiconductor equipment control standard, or gathered from an interface to an external information system.
 Figure 1. Feeding continuous, real-time inspection data into a detailed product history report facilitates proper and accurate manufacturing processes. |
null
Data collection is attached to a transaction that occurs during processing of that step. Data can also be gathered upon demand with an adhoc data collection. If the factory has a truly integrated MES system, production data should only have to be entered into the system once, and this information can then be sent to other applications to enable personnel to make smart decisions and/or take the appropriate action.
MES software should serve as the factory's nervous system, reporting what is happening during the manufacturing processes and alerting the appropriate people if actions need to be taken. In an ideal situation, this data is sent to the company's enterprise resource planning (ERP) software. With this “live” manufacturing data, the ERP application can make smart decisions that can save the company time and money.
SPC Method
Statistical process control (SPC) is a software method that ensures any deviations from processing norms are detected and corrected quickly, which ultimately improves consistency and production throughput. It is accomplished by gathering data, from an operator entry, automated interface or integration with an external system. This data is evaluated by the SPC control system. Different formulae can generate the charts for each spec.
null
When an out-of-control process needs to initiate an action, the SPC can place lots on hold, send messages to operators or engineers and can even place the equipment into a down state. In addition, the lot may be moved to another workflow, typically for rework.
Gate Rules
Gate rules inspect a sample from a certain number of manufactured products to determine if the lot is error-free. If a few or none of them fail inspection, the assumption is made that the entire lot is acceptable, and the lot moves to the next manufacturing step. If some of the units fail, the lot is either placed on hold or moved to a 100 percent inspection step.
 Figure 3. Online traveler reports allow management to view manufacturing performance from any PC. |
null
The gate rule states the sample size (either in units or in percentage of total lot size), number allowed to fail (in units or percentage) and action that needs to take place (i.e., place the lot on hold, set a failure flag or both). The failure flag redirects the lot to the 100 percent inspection step. This gate rule should be defined and attached to either a spec or to a bill of process. When the lot reaches a step with an attached gate rule, partial processing is allowed; the operator can display the gate rule before using it to determine the appropriate sample size.
Retesting
When a factory must test completed parts, a standard procedure is to retest failed parts to ensure that the tester made a proper connection. Although the part may pass subsequent tests, the ideal is for the part to pass the test the first time. Measuring first pass yield is an important statistic, as well as recording the actual number of tests performed because this directly affects the cost of the part.
When a track out is performed, the system checks the equipment to see if retests are allowed (Figure 2). If they are allowed and the maximum retest limit has not been reached, the retest quantity field allows retests to be entered. Any quantity entered in the retest field counts as losses when calculating first pass yield and adds to the cycle count on the tester, but will not actually be lost. Any retest quantity remains as quantity that still needs to be processed to complete the current operation.
Evaluating MES
When evaluating MES software, it is important to find a system that meets the factory's specific needs. Ideally, the system should be standards-based and leverage the factory's existing technology infrastructure. This not only shortens the time it takes to integrate the system into the factory's existing IT environment, it also saves programming time and causes less factory downtime.
The MES system also should offer a Web-based interface. This not only makes it easier to use, but also allows it to manage manufacturing operations at multiple sites — whether they are located across the street or around the globe. In addition, management can view manufacturing performance from any PC (Figure 3).
As optoelectronics manufacturers continue to strive to meet changing customer demand and technology advancements, more often they will rely on e-manufacturing solutions like automated inspection and MES. It is only with these tools that they can truly maximize their investment in manufacturing automation technology.
STEPHEN ROLSTON, senior consultant, may be contacted at Camstar Systems Inc., 900 E. Hamilton Ave., #400, Campbell, CA 95008; phone: (408) 559-5700; e-mail: [email protected].