Test/Diagnosis/Rework Process Step Calculator

This web page presents a stand-alone version of a test/diagnosis/rework analysis model for use in technical cost modeling of electronic assemblies. The approach includes a model of test operations characterized by fault coverage, false positives, and defects introduced in test, in addition to rework and diagnosis operations that have variable success rates and their own defect introduction mechanisms.  The model can accommodate an arbitrary number of rework attempts on any given assembly and can be used to optimize the fault coverage and rework investment during system tradeoff analyses.  The model is documented in:

The example implementation below allows all inputs to the model (except number of rework attempts) can be represented as probability distributions thereby accommodating inevitable uncertainties in input data present during tradeoff activities and uses Monte Carlo methods to determine model outputs.  This model enables quick access to decision-level data allowing the location of test and rework operations in a process to be optimized along with the level of fault coverage invested in, thus resulting in higher quality systems at lower cost.

For information on how to use the model click here.

User's Guide   [Input Fields] | [Probability Distribution Inputs] | [Output Fields] | [Buttons] | [Known Bugs]

In the following we use the word "part" to refer to any item (die, chip, board, module, etc.) that is being addressed by the test/diagnosis/rework process.

Input Fields:

Incoming cost = total cost per part incurred by processes that take place prior to this test/diagnosis/rework activity.

Incoming yield (%) = yield of parts after processes that take place prior to this test/diagnosis/rework activity.  Must be entered as a percentage, i.e., 100% yield must be entered as 100.

Yield before test (%) = yield of processes associated with handling and loading of the tester.  Must be entered as a percentage, i.e., 100% yield must be entered as 100.

Test cost = average cost per part of performing the test.

Fault coverage (%) = percentage of faults detected by the test.  Must be entered as a percentage, i.e., 100% fault coverage must be entered as 100.

False positives (%) = percentage of parts tested where a positive test results in parts that do not possess the attribute for which the test is conducted.  Must be entered as a percentage, i.e., 10% false positives, meaning that 10% of the parts will have a positive test (test defective) even though the parts do not have the fault the test is trying to detect, must be entered as 10.

False positives = the basis for the false positives fraction above.  This can be either all parts or only the good parts.

Yield after test (%) = yield of processes associated with handling and unloading of the tester.  Must be entered as a percentage, i.e., 100% yield must be entered as 100.

Rework cost = average cost of performing rework per reworked part.

Rework yield (%) = yield of the rework process.  Must be entered as a percentage, i.e., 100% yield must be entered as 100.

Rework fraction (%) = percentage of parts reaching the rework process that can be successfully reworked (others are scrapped).  Must be entered as a percentage, i.e., 100% reworkable parts must be entered as 100.

Rework attempts = maximum number of times a part is reworked before it is scrapped.

Diagnosis cost = average cost of diagnosing per part.

Diagnosis fraction (%) = percentage of parts reaching diagnosis that can be successfully diagnosed (others are scrapped).  Must be entered as a percentage, i.e., 100% diagnosable parts must be entered as 100.

Monte Carlo = "Yes" a Monte Carlo analysis will be performed, "No" the analysis will be performed using most likely values only.

Number of runs = number of Monte Carlo samples used (only enabled if Monte Carlo is set to "Yes").

Probability Distribution Inputs:

When the Monte Carlo input is set to "Yes" then the input parameters can be optionally represented as probability distributions.  To enter and access the distribution data, click in the desired field or on the desired slide bar and press the "Enter" key.  A dialog box will appear with the following inputs:

Distribution type = the type of distribution desired - Fixed Value (no distribution), Triangular, Normal, Uniform, or Lognormal.

Most likely value = value of the distribution with the highest probability of occurrence.  This value is take from the main interface, this value must be changed on the main interface.  This value is irrelevant for Uniform distributions.

Low value = low value of the distribution (applicable only to Triangular and Uniform distributions).

High value = high value of the distribution (applicable only to Triangular and Uniform distributions).

Standard deviation = defines distribution width (applicable only to Normal and Lognormal distributions).

Output Fields:

Both mean and standard deviation outputs are available for the following quantities.  The standard deviations will only be calculated when the Monte Carlo analysis is used.

Outgoing cost = effective cost per passed part after the test/diagnosis/rework process.

Outgoing yield (%) = yield of passed parts after the test/diagnosis/rework process.

Pass fraction (%) = percentage of parts that start the test/diagnosis/rework process, that pass the test/diagnosis/rework process.

Scrap fraction (%) =  percentage of parts that start the test/diagnosis/rework process, that are scrapped by the test/diagnosis/rework process (1 - pass fraction).

Buttons:

Default = populates all the fields (including the probability distribution inputs) with consistent default values.

Compute = causes analysis to take place.

Bar Chart = plots Monte Carlo results (Monte Carlo must be set to "Yes" to enable this button).

Known Bugs:

The following bugs are known to exist for this tool:

1) The black arrow graphics may not appear when the web page is initiated.  Scrolling the browser window or resizing the browser window should cause the graphics to refresh.

2) The browser window may hang after using the tool and attempting to close the browser window.  The hang may require the deletion of the process using CTRL-ALT-Delete.  The problem is confined to the browser window you have been using the tool in and does not affect other processes on the computer.