The economics of x-rays

Use both quantitative and qualitative information to help make a solid case for purchasing x-ray inspection equipment.

Stephen F. Scheiber, ConsuLogic Consulting Services, Slingerlands, NY -- Test & Measurement World, 2/1/2001

Few company controllers sign purchase orders for new equipment based on technical necessity alone. So, if you think an x-ray inspection system will help your company, you must show how the benefits outweigh the costs. Developing financial models that show realistic savings will help you make a forceful case for a purchase.

Unfortunately, many of the factors that enter into an analysis aren’t well defined. When you choose ATE or a machine-vision system, the alternatives may differ, but the economic requirements and technical benefits remain much the same no matter what equipment you choose. With x-ray inspection, even the definition of an inspection system varies.

You must consider manual or automated operation, in-line or off-line inspection, operator-driven or unattended systems, systems where an operator makes the pass/fail decision or systems that include software to aid an operator, and so on. Capital costs can range from about $50,000 to more than $500,000. And you’ll have to add costs for labor, facilities, infrastructure, and training.

You should start your analysis by finding out how much money your company can allocate to capital equipment. This amount will suggest the cost—and thus the type—of x-ray-system you can afford. But keep an open mind as you look at alternatives; your final justification may provide a savings sufficient to justify a larger-than-usual purchase.

With a general spending figure in mind, you can start to evaluate how you would apply an x-ray system. Perhaps the first decision involves where you could use an x-ray system. A centrally located x-ray system that serves several production lines generally precludes in-line, real-time inspection. Such a system requires that operators withdraw samples from each PCB lot and send them to the x-ray system.

An x-ray system placed in-line on a single production line (Figure 1) can reduce the cost of scrap and rework as it helps operators find and classify defects quickly. The in-line approach adds a step to the manufacturing process, requires more floor space, and can increase the inventory of boards on the production line. An in-line x-ray system also can slow throughput and cause a production bottleneck. Because high-volume manufacturers cannot tolerate any production-time extension, they may install several x-ray systems that run in parallel on a single line. Of course, those additional x-ray systems increase the capital investment and operating expenses.

Low-volume, high-value, and high-margin PCBs also often justify an in-line system. PCBs such as those meant for medical or aerospace applications must be manufactured correctly the first time. Manufacturers of these high-value products can’t risk shipping faulty products, so many of them add a step to production to accommodate in-line x-ray inspection.

Look at alternatives

Your next step is to decide whether to use a manual or an automated x-ray system. Low-volume operations where cost is of primary concern may favor a simple and relatively inexpensive x-ray system operated by a technician who makes a pass/fail decision. Manual systems, however, suffer from human faults such as inconsistency and operator fatigue that raise test and repair costs further down a production line.

Image-analysis software provides better consistency, but it can have difficulty properly classifying characteristics at or near tolerance limits. The resulting false failures or false passes—usually called false calls—increase the number of PCBs in the repair-and-retest loop.

These false-call PCBs may end up back on the production line or in the scrap bin. There’s a tendency for companies to keep scrap in inventory rather than absorb its cost. In most cases, the promise that someone will eventually repair these PCBs and they’ll get sold proves false. Simply by reducing the number of scrapped PCBs, you may justify the cost of an automated x-ray system.

As you develop your justification, consider whether an x-ray system will supplement or replace an existing test step. Because x-ray inspection identifies solder-joint problems and other manufacturing defects, some companies try to replace an in-circuit tester with an in-line x-ray inspection system. The success of this approach depends on the types of faults a PCB may experience.

If solder anomalies, missing components, and off-pad components make up the majority of faults, an x-ray system may suffice. Some manufacturers of surface-mount PCBs report that solder-related problems can represent up to 80% of all process faults. On the other hand, if your PCBs don’t exhibit the types of failures just mentioned, it’s unlikely any yield improvement due to an x-ray system will justify the loss of in-circuit testing.

If you decide that replacement of an in-circuit tester is justified, you should determine whether your company can use the tester elsewhere, can resell it, or must scrap it. Resale or reuse may bolster your financial plan.

At the heart of any economic analysis lie predictions of yield improvement, scrap reduction, false-failure rates, and so on. Your company may have information about test strategies, product history, case studies, and engineer experience that you can use to help predict how a process would benefit from x-ray inspection.

Figure 2. After passing an x-ray examination, good PCBs travel on to other test stations. Rejected PCBs get sent to repair stations. Repaired boards get reinspected. Boards that can’t be repaired end up in a scrap pile.

Use quantitative models

Once you know how you would use an x-ray system, you’ll need to demonstrate that a purchase is economically justified. You should run some financial models to show how an x-ray system can save your company money. The following examples illustrate some of the considerations you should include in your models.

Suppose each year your business ships 100,000 PCBs worth $300 each. Real yield (good PCBs) from manufacturing amounts to 80%; that is, 80,000 good PCBs and 20,000 bad PCBs come off the line each year. Of the 20,000 bad PCBs, 80% (16,000) “fall out” at in-circuit test, 15% (3000) at functional test, and the remaining 5% (1000) at system test. Repairs per PCB cost $10 at in-circuit test, $25 at functional test, and $100 at system test. Capacity is sufficient to handle any necessary retests. For the moment, assume no false failures and no scrap.

You decide that increased throughput requires preventing as many problems as possible during PCB assembly and finding any remaining problems as early as possible during testing steps. By adding an x-ray system, you expect to catch 93% of the defects during manufacturing, 5% during functional test, and 2% at system-level test. For simplicity, assume the cost for repair after x-ray inspection equals the cost of repair after in-circuit test. (That’s a reasonable assumption because both methods pinpoint failing parts and bad solder joints explicitly.)

Table 1 shows the costs for using in-circuit testing vs. x-ray inspection. In this case, the one-year saving of $84,000 is generally insufficient to justify an x-ray system. (The T&MW Web site provides all the tables mentioned in this article. )

Now consider PCBs that are more difficult to test. A higher percentage of SMT components, higher required throughput, more inaccessible nodes, and higher logic complexity contribute to a reduced yield of 60%. And this type of PCB increases the load on the tester and on the repair-and-retest loop. In this case, the in-circuit tester finds 60% of the faults, functional test finds 35%, and system-level test finds 5%. Table 2a compares the costs of in-circuit test and x-ray inspection for these PCBs. The cost reduction of $288,000 per year shows more promise for justifying an x-ray system.

Improve your process

Now assume you’re using the information from the x-ray system to improve your manufacturing process. You’ve corrected problems with the solder-paste printer and the pick-and-place machine and now obtain a manufacturing yield of 85% (85,000 good PCBs).

Also, assume that all of the improvements come from correcting manufacturing defects that in-circuit test would find. As in Table 2b, the cost benefit increases to a more attractive $538,000 per year. (Note that taking all of the now-good products out of the in-circuit repair step is the most conservative assumption we can make. If some of the improvement comes from not finding as many problems downstream, repair costs drop even further, increasing the x-ray benefit.)

Electrical tests cannot pinpoint numerous types of faults that x-ray inspection can find, including solder opens, solder voids, misplaced balls on ball-grid arrays, and insufficient solder (or no solder at all). These problems can cause unrepairable bad PCBs that end up as scrap. Now suppose 1.5% of your manufactured PCBs fall into this category, as shown in the top half of Table 3 for the simple PCB. To ship 100,000 PCBs, you must start with 101,523 PCBs, thus increasing all repair costs slightly. Remember, you can’t declare a PCB as scrap until after you try to repair it. The value of those 1523 scrapped PCBs at $300 each—$456,900—adds to costs.

One justification for using an x-ray system is its ability to find the faults that cause you to scrap PCBs. Assume by applying an x-ray system to the simple PCB, you eliminate the 1523 scrap PCBs, as shown in the bottom half of Table 3. (These are the same values shown in the bottom half of Table 1.) Remember, there’s no scrap in this table. Now the one-year benefit for the simple PCB amounts to almost $546,000, perhaps enough to justify purchase of an x-ray system.

When you take scrap into account for the more troublesome SMT PCBs, as shown in Table 4a, the gain totals $756,885. And if you use information from x-ray inspections to reduce manufacturing faults, the gain increases to $1,006,885, as shown in Table 4b.

False failures cause problems

Of course, your economic evaluation should also consider false failures—reported problems that don’t actually exist. These failures reduce the benefits of using an x-ray system. Most x-ray system users report a higher false-failure rate from x-ray inspection than from traditional electrical tests. These false failures add costs by requiring repair and retest steps for PCBs that aren’t actually bad. You should also account for the cost of the false-failure PCBs that circulate in the repair loop. (Generally the cost of this “inventory” falls below the level of uncertainty in the overall analysis, so I’ll ignore it.)

Repair-and-retest costs spring from the practice of looping the repaired PCB back through the manufacturing-and-test process. If one of these PCBs fails again, the cycle repeats. At a certain point, you assume the “fault” does not exist, and pass the PCB on to the next step. If company policy dictates passing the PCB on after three such cycles, every false failure adds three in-circuit-type “repairs” to the overall cost.

To be complete, your models should show the economics of assuming a reasonable false-failure rate. You’ll have to determine—or make an educated guess at—this rate for your production line. I’ll assume that typical testing flags 2% of all PCBs as defective, even though they are good. When an analysis includes an x-ray system, I’ll assume the false-failures rate rises to 6%, a fairly high number, but it’s realistic, and it makes the analysis more conservative. (Additional tables available on the T&MW Web site analyze false-failure data for the simple PCB and the complex SMT PCB. These examples all include a 1.5% scrap rate, too. www.tmworld.com/articles/2001/02_xray_table.htm)

Note that even when you include the additional false-failure cost attributed to an x-ray system, the cost can be more than offset by reducing the amount of scrap. These examples demonstrate the importance of both estimating the cost of scrap and then eliminating that scrap as a key justification for using an x-ray system.

You can use these quantitative examples as starting points for your own analyses. Factors such as yield, test effectiveness, false-failure rates, and scrap figure prominently in predicting the benefits of x-ray inspection. But your situation may look different. You already may experience high yields and little scrap. Perhaps you already ensure high manufacturing quality by inspecting solder paste after printing. On the other hand, you may have accumulated a scrap pile large enough to damage your company’s financial health.

In addition to the costs of x-ray equipment, what other costs will you have to cover? You must include costs for labor, programming, maintenance, facility improvements, and any costs needed to change your repair-and-retest loop. If an x-ray system will replace an existing in-circuit tester, the labor costs of moving PCBs from place to place may be equivalent, but x-ray equipment operators may cost more than visual inspectors—especially if they make pass/fail decisions. To keep models simple, consider only differences in labor cost. At this point, you need not consider benefits, training, and other costs.

If you plan to share an x-ray system among several products or production lines, amortize the costs over all of them. But you must consider the savings from each product or production line individually. Two lines that produce the same number of comparable-value PCBs per year may not benefit equally from adding x-ray inspection. An analysis may suggest adding x-ray inspection to one product line, but foregoing it on another. If you end up with excess inspection capacity, you can assign it elsewhere.

For an inexpensive x-ray system, infrastructure, labor, repair-station, and other costs represent a higher percentage of overall costs than they would for a more expensive system. Consider just operator costs. One operator who earns $15 per hour (including vacations, health insurance, and other benefits) over 2080 paid hours per year costs $31,200. This amount adds more than 60% to the base cost of a $50,000 x-ray system, but it adds only 6% if the x-ray system costs $500,000. So, if you look only at the equipment costs as a first approximation of overall costs, you’ll greatly overestimate the cost advantages of the less-expensive system.

How do you get some realistic estimates for costs? Although you shouldn’t rely on vendors to perform an overall economic analysis (by their nature they’re biased), they can supply accurate estimates of their equipment’s likely cost. Also, ask vendors to estimate costs for competitors’ types of products. Those estimates may help you better evaluate the competitors’ estimates.

Deciding that you need x-ray inspection represents the beginning rather than the end of the planning process. You still have to consider the many alternatives and examine how each would affect your operation’s economic performance. Only then can you construct an adequate justification with which to approach your company managers. T&MW

For more information

Davis, Brendan, The Economics of Automatic Testing, 2nd ed., McGraw-Hill, New York, NY, 1994.

Scheiber, Stephen F., Economically Justifying Functional Test, Quale Press, Florence, MA, 1999.

Titus, Jon, “X-Rays Expose Hidden Connections ,” Test & Measurement World, October 1999. p. 28.

For up-to-date information about companies, visit the Inspection Equipment portion of our Buyer's Guide.

Stephen F. Scheiber has spent more than 22 years in the test industry. He consults and writes extensively about such topics as test strategies and test economics. E-mail: sscheiber@aol.com.