Manufacturing

Additional Manufacturing Reporting Concepts

Throughput Yield

While overall yield is simply P/(P+F), we can consider the "throughput yield" of our process by multiplying the FPY of each stage:

FPY(overall) = FPY(test1) x ... x FPY(testN)

This overall throughput yield gives us the probability that any unit will make it through the entire process without failing.

Note that the more steps you have will result in a lower probability of a unit escaping the line. Every decrease in FPY along the way compounds further down the line.

Units-per-hour (UPH)

UPH is a measure of the total number of units produced in a given hour of production time.

Cycle Time

Cycle time is a metric for how long a process takes. Cycle time can be evaluated for each individual step or for the overall process.

Cycle time is usually reported for each test step. If UPH requirements increase, additional stations for high cycle time tests are required (or test methodology must be improved).

You can track the overall cycletime for a unit through the production line (from start of input to packout). Note that each unit's cycle time can vary depending on any number of reasons:

  • Repairs
  • line down to adjust a station
  • Lunch break
  • Shift change

As such, I recommend tracking an average of each unit's overall cycle time.

Work in Progress (WIP)

Units that have not cleared a test stage or cleared the test line can be considered WIP. Tracking WIP gives you a better idea of what is actually being included in your overall output and yield.

In Repair

I like to have my CMs add two stations to their process flow:

  • FA Checkin
  • FA Checkout

This allows the CM to generate reports indicating how many units are in repair at any given time. The CM can also report how many units have returned to the line but not yet started testing.

Test in Progress

This is a subcategory of failing units.

"Test in progress" tracks units that are currently marked as "fail" but have not fullfilled the retest policy. Since the retest policy is not fulfilled, they are not yet "true failures".

Fail on Line

This is a subcategory of failing units.

"Fail on line" tracks units that have failed beyond the retest policy but haven’t checked into repair.

Loopers

Loopers are units that enter into your repair / return cycle repeatedly. This can simply be 1-2 loops, or like the Hotel California: you can checkin, but you can never leave...

Crash Course on Manufacturing Yield

At its simplest, yield is an easy concept: your yield is P/(P+F), or put into plain English:

(total # of passing units) / (total # of units input)

This concept gets complicated much further to provide more insight into the health of our manufacturing line. We will review the following concepts in relation to our manufacturing yield:

  1. Retest
  2. Repair
  3. First pass yield = yield without including units that went to repair
  4. Overall yield = yield including units that were successfully repaired
  5. Retest rate: % of units that retested pass

Retest

When a unit fails a manufacturing test, it is not immediately removed from the line. CMs will retest the DUT multiple times before declaring it a failure and sending it to repair. CMs will usually retest/repair DUTS until they pass. Less upright CMs will find tricks or swap in other units in order to pass the failing DUT down the line.

This fact of retest means that units can be reclassified as a pass or fail over time. Your CM will usually tell you "we had 100% yield", but this does not give you actual insight into how many units entered repair and how many units needed to retest pass.

Repair

When a unit retests fail enough times, it will enter into repair. This unit counts as a failure until it is made to pass again.

After repair, the unit will return to either the station or the start of the test line. This is determined by the repair policy you set with your CM.

First Pass Yield

In order to get a better picture of the yield/rework breakdown we can categorize our passing units into different groups.

"First pass yield" (FPY) is the measure of units that tested pass on the first test attempt (or retest-pass) before being checked into repair. Once a unit fails enough times to enter repair, it counts as a "fail" under FPY.

FPY is typically calculated for each process step.

FPY gives you a metric for your yield without repairs being included. Consider FPY as a time-zero metric for passing units - units may fail at a later time. Later failures do not remove it from FPY calculations.

Overall Yield

Overall yield is a measure of your actual output yield including repaired units. This number is the typical yield number we consider: P/(P+F).

Retest Rate

When we count passing units, we are actually counting "no retest pass" + "retested pass".

Retest rate can have a large impact on your overall throughput, as DUTs must spend more time at a station before moving down the line. High retest rates can also be used to identify fixture, software, or SOP issues.

To calculate your retest rate, simply consider the number of units that retested pass.

Pass to Fail Transitions

Note well: it is possible for a unit to pass in your first pass yield but fail in your overall yield (or later a field failure).

Example Calculations

Consider the following example for calculating yield & retest rates:

  • Input 30 boards
  • 25 boards passed
    • 4 boards failed once, but retested pass on the second try
  • 5 boards went to repair
    • 2 boards could not be repaired
    • 3 boards were successfully repaired
      • Two boards retested pass on the second try after repair
  • First pass yield = 25/30 => 83%
  • Overall yield = 28/30 => 93%
  • Retest rate = 6/30 => 20% (This is high - look into it to improve UPH!)

MAC Allocation Tracker

If your company has a MAC address block, you need to keep track of address allocations so you don't use the same address for multiple devices.

We developed a simple Excel template to keep track of MAC address allocations. This tracker can handle both single address and bulk allocations. It also shows the current number of addresses allocated and number of addresses remaining.

Get this Free Template

You can download our MAC Address Tracker template for free from the Embedded Artistry store.

MACPreview.png

Change Log

This article was updated on 21 July, 2018 to link to the Embedded Artistry store.

Related Posts