Data processing: measuring – calibrating – or testing – Measurement system in a specific environment – Quality evaluation
Reexamination Certificate
2002-01-22
2003-09-16
Shah, Kamini (Department: 2863)
Data processing: measuring, calibrating, or testing
Measurement system in a specific environment
Quality evaluation
C702S181000, C702S182000, C702S184000
Reexamination Certificate
active
06622101
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The invention relates to a method for monitoring a quality of a plurality of particularly different technical products types which are produced in a manufacturing process.
2. Discussion of Background Information
In every production, the quality of the manufactured products has to be monitored. In this regard, statistical process control is used for controlling and improving quality and for monitoring manufacturing processes. In principle, there are two requirements to be met by a quality monitoring.
The operator of a machine in a production line needs a rapid decision basis according to which he can evaluate the batch which has just been manufactured and measured.
The person in charge of quality control wants a rapid and complete survey of the manufacturing processes and the quality of all products manufactured. This survey does not have to be on-line, but on the contrary should indicate long-term tendencies and constitute a decision basis for process modifications or error analysis.
Until now, each parameter to be controlled of a product has to be monitored with the help of an individual representation, e.g. a control card. Since a product passes through many manufacturing steps, many control cards are required. When a lot of different products (product types) pass through a production line, a particularly great number of control cards are required which all have to be analyzed. For this purpose different ideas realized with the help of computers are known, according to which the great number of cards are automatically monitored with respect to exceeded control limits and which try to display large data quantities as distinctly as possible.
FIG. 2
illustrates a manufacturing process consisting of a sequence of processes (process flow) which is passed through by j different product types (p=1 . . . j), an individual card (e.g. a Shewart-card according to
FIG. 6
) being set up. This procedure has some principle disadvantages:
A possibly very large number of individual cards cannot be visualized distinctly. There is no method of displaying the monitoring of the complete quality of all technical products of a production line.
When the different product types have for example a high positive correlation and when the process is monitored by several individual cards, it cannot be determined sufficiently safely whether the manufacturing process is still under statistical control: When within the control limits, random samples average values of the different product types differ much more than it is to be expected by said high correlation, the manufacturing process is most probably no longer under statistical control, although no trespassing of the limits by one of the products is signalized.
SUMMARY OF THE INVENTION
Therefore, the object of the invention is to provide for a method for keeping a plurality of different (several) product types, which are produced in a quasi-parallel manufacturing process, statistically monitored, for distinctly representing or displaying the data quantities produced and for improving the expressiveness of the measurements.
Claim
1
,
12
or
14
are proposed for solving said problem.
A multiple products control is effected by representing or displaying a course of a state variable by which the manufacturing processes and the quality of the products manufactured in productions comprising a large number of product types (e.g. ASIC production) may be controlled rapidly and distinctly. In each manufacturing step, the quality of all product types, which pass through said at least one manufacturing step, may be monitored and distinctly displayed together by said control method. Measured values from a manufacturing station are collected during a defined time interval, said measured values being displayed as an obligatory state value only after the expiry of said time interval.
Said multiple products control (for controlling product average values) may be used for simultaneously monitoring said large number of products. Said representation or display expressively illustrates a statistical state of a manufacturing process.
The invention provides a method for rapidly and reliably monitoring the quality of all products passing through a respective manufacturing step in productions comprising a large number of products (broad product range) and for distinctly visually representing or displaying the data pertinent to quality. Said multiple products control is also well-suited for a long-term analysis of quality and as a decision basis for possible slow modifications of the manufacturing processes.
Hotelling's T
2
statistical analysis as known from literature serves as a statistical basis for said multiple products control. According to the invention, said statistics is used for monitoring a technical parameter for a plurality of technical product types manufactured in a “quasi parallel” process (e.g. in semiconductor production) (claim
12
).
A quality monitoring by said T
2
multiple products control has considerable advantages compared with prior art:
A rapid and reliable information on the statistical state of a technical manufacturing process is available.
By also considering in said display correlations that can exist between differently designed products, a statistically definite information independent of the degree of said correlation is provided.
The quality of all products manufactured in a manufacturing process is distinctly displayed.
A time-compressed representation provides for a long-term quality management and quality control, particularly for a recognition of tendencies in a manufacturing process.
The method uses average values of a random sample of products modified in their manufacturing state. Said average value is positioned as a part of a multi-dimensional average value. In fact, said multi-dimensional average value is convertible into a state value at any time, but it is converted into such an obligatory state value only after a determined time interval, said state value being represented either on a screen or graphically. Said determined time interval results from the manufacturing process. It is either a maximum time interval (claim
2
), it may, however, also result from again subjecting a product type of a first type number to a physical modification in a monitored manufacturing station. In this case, the same product would provide two measured average values, and prior thereto, a multi-dimensional average value established until then is converted into an obligatory state variable. A further criterion may be determined by said defined time interval. Once all product types monitored having been physically modified in said monitored manufacturing station, an obligatory state variable may also be calculated, since no further measured average value is required for completing all product average values in said multi-dimensional average value. However, said aspect is a subordinate case of the criterion mentioned before; the subsequent requirement of a product type in said monitored manufacturing station would in any case satisfy the criterion mentioned before and release a calculation of said obligatory state variable.
Consequently, said state variable may also be calculated more frequently (claim
10
), whenever a new product average value is supplemented to said multi-dimensional average value. Said value remains hypothetical, it is not registered in the obligatory course of said state variable.
If the number of monitored product types changes, i. e. the “first type number”, a new initializing run has to be effected. The dimension of the used vectors changes as well as the supposed product average values (claim
11
).
The supposed product average values are those values which are determined from preliminary calculations for a respective product type (claim
9
, claim
3
). A product average value replaces to a limited extent a measured average value in a multi-dimensional average value (average value vector). If in a calculation or in a forthcoming calculation of the state
Hennig Peter
Oechsner Richard
Pfitzner Lothar
Schneider Claus
Strzyzewski Poitr
Fraunhofer-Gesellschaft zur Foerderung der Angewandten Forschung
Greenblum & Bernstein P.L.C.
Shah Kamini
LandOfFree
Quality surveillance of a production process does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Quality surveillance of a production process, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Quality surveillance of a production process will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3060784