(Continued from page 1)
The x-axis of a parametric x-y plot does not have to be a parametric value, but
can also be a series of discrete values such as a date range or product
serial numbers.
By using the parametric x-y plot in this way, managers identify the trend of a parametric
value over a series of prototype or production units. For example, a wireless networking
company's design group needed to look for trends in the crystal operation frequency
for prototype units as shown in Figure 7. Furthermore, this same company wanted
to analyze the average power output of each product by date. For this second report,
the TDM system averages the values for output power for all units tested on a particular
date
and then plots the average versus date.
|
Figure 7: A wireless networking company's design group needed to look for trends in the crystal operation frequency for prototype products.
Enlarge Image
|
A TDM system further analyzes this type of data with statistical process control
(SPC) reporting including plotting the range and standard deviation of the parametric
values within each grouping.
Other common test performance analysis reports are the test volume report, cycle
time report, and station utilization reports. Test volume reports show the number
of times a test executes by date and thereby helps managers identify changes in
test capacity. The cycle time report shows the amount of time the test station requires
to complete each test, and is used by managers to improve the efficiency of their
tests. Finally, station utilization reports show the operational capacity of each
test station so that managers can maximize throughput and their investment in automated test systems.
Test Performance Analysis Reports
Test performance analysis reports change pass/fail data and test system utilization
data into information for management. The most common types of test performance
analysis reports available in an enterprise TDM system are yield reports. Yield
reports provide information on the number and percentage of units passed and the
number and percentage that failed by presenting the information in a histogram or
Pareto chart. These reports also present the number of times a product was required
to go through a test station to pass. A large aerospace company, for example, wanted
to measure first-pass yield of sub-assemblies through a test station (Figure 8a).
For further evaluation, the company can examine sub-assemblies that have a low first
pass yield by running an N-pass yield that shows the number of times a product had
to go through the test system in order to pass (Figure 8b).
|
Figure 8a: A large aerospace company wanted to first evaluate the yield of sub-assemblies that could passed on their first time through the test station.
Enlarge Image
|
|
Figure 8b: For further evaluation of sub-assemblies that have a poor first-pass yield, the company can run an N-pass yield report to determine how many passes through the test station a particular product required in order to pass.
Enlarge Image
|
To hone in on the main causes of the yield problems, engineering managers filter
data according to time and location related information to generate yield reports
by operator, date, lot, plant, production line, and station. Other filters commonly applied to yield reports are related to the product family and include yield reports
by product line, by product, by sub-assembly, and by contract manufacturer or component
supplier. A third common group of filters are related to the test itself. For these
reports, managers filter data by test, by test step, and by test version.
An enterprise TDM system can combine multiple filter criteria to create very specific
reports to find the exact cause of a poor product yield. A manager can change his
view of the test data within seconds from a total yield report by test for a specific
line of the plant, during a specified date range, for a particular product family
to identify
which test or even test step is responsible for low yields. For example,
a wireless broadband company uses Arendar with their contract manufacturers (CMs)
to quickly identify the cause of failures by product, by test, and by test step
to maximize their yields.
Managers are also changing their test data into information by applying a line chart
to yields to see trends and easily determine if yields are improving. The wireless
broadband company previously mentioned monitors first pass yields by date to see
if yields at their CMs are improving (Figure 9).
Another power supply company analyzes
the standard deviation of the yield to show how well their manufacturing process
is staying in control over time (Figure 10).
|
Figure 9: By monitoring the first pass yield by date, a wireless broadband company can identify the trend of their yields at their contract manufacturer.
Enlarge Image
|
|
Figure 10: A power supply manufacturer wanted to calculate the standard deviation of the yield to show how well the process is staying in control over time.
Enlarge Image
|
Any Time, Any Place
Historically, generating test data analysis and test performance analysis reports
was a tedious, manual process that took days to create. Now, enterprise
TDM
software creates these reports in seconds and makes them available throughout the
company with a standard Web browser. Problems are identified quickly and engineers
and managers take immediate corrective action resulting in large savings. With this
instant access to a company's worldwide design and manufacturing test information
at any time and at any place, managers obtain a competitive advantage by making intelligent decisions that improve their products and increase their productivity.
Ask the Expert
Email questions concerning this article to:
expert.testpronews@vi-tech.com
|
|
|
|
Read about the latest information for integrating automated test systems throughout the enterprise. |
|
|
|
 |
Download PDF |
|
|
