Getting the Check Signed When Buying Test Equipment

As engineers, we are accustomed to solving problems. After all, that is our job. If a piece of equipment or infrastructure provides part of the solution, why should anyone thwart a decision to buy it?

Unfortunately, the reality of corporate life is considerably different from that ideal vision. However much you want to spend the company’s money, management usually asks for a formal economic justification before issuing purchase orders or signing checks. You need to answer some difficult questions:

• Is there a less expensive alternative?
• How long before benefits exceed costs?

• Does the purchase follow established budget guidelines?

Go With the Flow

Economic models are inexact and imperfect at best. Their unpopularity with most engineers has prompted many vendors to provide assistance, often including some kind of software tool. Your numbers go in one end; a justification comes out the other.

Software tools have several drawbacks. Most vendor-provided tools contain a built-in bias in favor of that vendor’s products. This bias may be unintentional, but the uncertainties inherent in any economic model coupled with the vendors’ preference for their own products tend to push the numbers in that direction.

The biggest disadvantage is more fundamental. Some engineering decisions still require human analysis and judgment. As yet, computers cannot weigh factors as effectively as good human problem-solvers can. Software offers little understanding of the process that lies between input and output.

Principles of Payback

A justification looks at the relative merit of two or more solutions to a particular problem. In board test, the problem relates to manufacturing yields, fault diagnostics, and board repair. The engineer weighs the current scheme against one or more alternatives. Of several purchase scenarios, which provides the optimum mix of costs and benefits? The accuracy of this analysis depends heavily on how well you can predict the future.

Costs fall into three basic categories: one-time, recurring, and ongoing. One-time costs occur only once or rarely enough that you can delineate your analysis around one such expenditure. Capital costs for equipment and related infrastructure fall into this category. You may buy test equipment from time to time. But by analyzing each purchase as a separate project, within the confines of that project, it occurs only once.

Recurring costs happen frequently but not necessarily regularly. Test-program generation and fixture construction fit this definition.

Ongoing costs are analog in nature. That is, they can be infinitely subdivided without losing their meaning. You can express rent or labor costs per month, week, hour, or even femtosecond.

Calculating all of the costs that accompany a test activity may seem a daunting task. However, the purpose of a cost justification is to compare alternatives, not to define a budget. Any cost that remains the same regardless of the solution chosen is irrelevant, and you can ignore it.

For example, evaluating test alternatives costs money in salaries, production-floor time, and travel expenses. Because you must perform the evaluation whatever your final decision might be, it makes no difference whether you spend $5 or $500,000 in that effort.

In addition, facility-wide costs do not change if you add or remove a piece of equipment or an entire manufacturing line unless you expand, add, or get rid of an entire building. Adding a product or product line to an existing factory alters the allocation of costs among the ongoing projects, but the overall cost to the organization remains the same. Consequently, you can ignore facility-wide burdens such as rent or mortgage, utilities, and security. Table 1 (see below) lists the costs that remain after this elimination.

Table 1. Significant Costs for Justification

Some costs are relatively easy to obtain. For example, the company controller establishes some of them, such as burdens on raw hourly labor rates to account for benefits including health insurance, vacations, and sick leave that apply to individual workers.

Benefits of a strategy change can include reduced costs for fault identification and repair as well as intangibles such as increased sales from an improved reputation for quality. Here we focus only on cost reductions.

Justification

Several available techniques compare costs and benefits. Company management dictates which of them to use as well as the boundary conditions that you have to meet—a minimum threshold—to justify the expenditure.

Most common is the concept of payback: How long before the benefits of a strategy change offset its costs? Payback is relatively easy to understand and straightforward to calculate; however, it does not consider the time value of money. An expenditure made a year from today costs less than spending the same sum today because during that year you can put the money into a bank account or other income-bearing instrument. Any model that does not take this difference into account will overestimate future costs and future benefits. References 1 and 2 explore these techniques in detail.

On the other hand, most companies demand a very short payback period for test equipment and related infrastructure, sometimes less than 18 months or even less than a year. The value of money usually erodes very little over such a short period.

Using payback, you could argue for dropping the cost-of-money entry in Table 1. As a rule, I leave it there to compute the cost of holding extra inventory to compensate for false failures and board scrap. In most calculations, however, this cost is quite small, small enough to ignore in this example.

False failures during manufacturing test enter the repair-and-retest loop along with real failures. After several unsuccessful rework cycles, most companies pass the board to the next step, hoping that the failure was fictitious or subsequent test steps will find it. This procedure incurs both labor and material costs for the unnecessary repairs.

If we ignore inventory-holding costs, the cost of board scrap is the board’s manufacturing cost. To compensate for scrap, the operation also must make more boards than it plans to ship. Here we assume sufficient capacity to handle the extra production.

Payback-period justifications should always include the effect of taxes and depreciation. Any savings generated by a strategy change represents an increase in the company’s income (profit) and is subject to tax. Depreciation is a government-mandated formula for deducting part of a capital investment from the company’s income each year, which reduces the tax and increases the expenditure’s benefit.

In the example that follows, yield is defined as the aggregate percentage of good boards that passes to the next step. That is, a 92% yield on 100,000 boards out of in-circuit test means that, after all repair and rework, 92,000 good boards and 8,000 bad boards that the tester thinks are good pass to functional test. This sometimes is referred to as actual yield.

Setting Up the Example

Consider a small manufacturer of PC motherboards, shipping 100,000 boards per year at a cost of $400 each. The current test strategy consists of in-circuit test and functional test. Surface-mount and other space-saving technologies make finding manufacturing problems using a bed of nails increasingly difficult. The test engineer is considering adding an inspection station to prescreen the in-circuit test.

Assume that the inspection system will find some manufacturing defects that otherwise would fall to the in-circuit tester. The engineer will avoid looking for those problems twice, reducing the complexity of the in-circuit test fixture and program by enough to compensate for the cost to program the inspection system. Since both strategies will exhibit the same fixture and programming costs, we ignore them.

Figure 1 shows the factors used to calculate the cost of the current strategy. The in-circuit equipment already is in place and does not require new capital purchases. The in-circuit labor cost will not change because those workers still will perform the same tasks, nor will repair, maintenance, and spares costs for the existing equipment. We also assume that the alternative strategy will find more failures prior to functional test than the current one, but the added step will not affect the yield out of functional test.

The strategy change does not affect device scrap for defective boards because the same defects come through manufacturing. Only the labor cost for repairs will change. The inspection system will increase the number of false failures. However, when working only with differences, false failures are calculated once, in this case, under the alternative strategy.

For the purpose of comparison, the current-strategy costs include board scrap and functional repair. The 2,041 extra boards per year that come through manufacturing to permit shipping 100,000 cost $816,400. Since under the current strategy the actual yield is 92% out of in-circuit test and 99% out of functional test, the cost for functional repairs is $20 per repair for 7% of 102,041 or 7,143 boards, totaling $142,860. As a result, the cost base for comparison of the current strategy is $959,260 per year, as Figure 1 shows.

The Alternative Approach

The inspection station under consideration costs $300,000 including necessary infrastructure modifications. It will require one operator per shift for two shifts.

Because inspection like in-circuit test uncovers manufacturing faults and explicitly identifies failures, labor rates for inspection operators will be the same, as will the associated repair costs. Depreciation will be five-year straight-line, and the tax rate is 35%.

The chief benefits of the change in strategy are an additional 4% fault coverage and a sharp reduction in board scrap. The number of false failures rises from 2% to 5% of all boards through manufacturing, increasing those costs. Figure 2 illustrates the changes. 
Since the purpose of this example is to determine how long it will take for the savings generated by the inspection step to pay its cost, the initial calculated cost will not include the capital expenditure. The rationale for this decision will become clearer shortly. 
Although Table 1 lists training as a one-time cost, here we allocate $20,000 per year for that purpose. We have assumed that fixture and programming costs and the cost of engineering change orders (ECOs) and enhancements do not change between the two strategies.

Maintenance and spares cost 10% per year (a common service-contract charge) or $30,000. Assume a 2% allowance per year for downtime and consumed spares or $6,000. The two additional operators at $15 per hour, including benefits for 2,080 hours per year, total $62,400 per year. The alternative scheme produces less scrap, requiring only 100,503 board starts to ship 100,000 boards.

The current strategy, excluding false failures, repaired 6% of 102,041 boards or 6,122 boards at in-circuit. In the alternative case, 10% or 10,050 boards are repaired after inspection or in-circuit test, a difference of 3,928 boards. Functional test now repairs 3% or 3,015 boards.

The original strategy incorrectly flagged 2% or 2,041 good boards as defective and sent them through the repair-and-retest loop. This time, 5% or 5,025 boards suffer that fate, a difference of 2,984. If the boards go through three cycles before passing to functional test, the cost per board is three repairs at $3.33 and three scrapped devices at $5.00, totaling $74,570.

As a result, operating the alternative strategy costs $467,550 per year, and the before-tax savings totals $491,710 per year as Figure 2 shows. Since that sum represents additional income, it is subject to tax. With a 35% tax rate, the after-tax net is 65% or $319,612.

Because depreciation represents a tax deduction, it lowers tax by 35% of that deduction. In other words, the 35% tax that you do not pay is extra cash in the company’s pocket. With five-year straight-line depreciation, the equipment depreciates one-fifth of $300,000 or $60,000 per year. The benefit, then, is 35% of $60,000 or $21,000 per year.

Table 2 (see below) shows the resulting payback as 10.6 months. Such short paybacks are quite common with test equipment. By any controller’s criterion, the new inspection system represents a good investment.

Table 2. Payback Calculation

Again, this example will not decide your need for a strategy change. It merely illustrates the technique for comparing alternatives without relying on software. Only you can determine its application for your particular situation.

References

1. Davis, B., The Economics of Automatic Testing, Second Edition, McGraw-Hill, 1994.
2. Scheiber, S. F., Building a Successful Board-Test Strategy, Butterworth-Heinemann, 1995.
3. Scheiber, S. F., A Six-Step Economic-Justification Process for Tester Selection, Quale Press, 1997.

4. Scheiber, S. F., Economically Justifying Functional Test, Quale Press, 1999.

About the Author

Stephen F. Scheiber, principal of ConsuLogic Consulting Services, has spent more than 20 years in the industry, serving 13 years as contributing technical editor and senior technical editor for Test & Measurement World and as the first editor of Test & Measurement Europe. He is an author and lecturer and offers seminars, publications, and assistance on economics and other test-related issues. ConsuLogic Consulting Services, 276 Longhouse Lane, Slingerlands, NY 12159, 518-452-9228, e-mail: [email protected].

Published by EE-Evaluation Engineering
All contents © 2000 Nelson Publishing Inc.
No reprint, distribution, or reuse in any medium is permitted
without the express written consent of the publisher.

December 2000