Test Strategy Selection for Low Volume PCB Assembly

Choosing how to test a low volume assembly is a different problem from choosing how to test a high volume one, because the fixture cost dominates the calculation and because the product may change before the fixture is worn out. A test strategy for low volume has to trade coverage against flexibility, and the right answer is usually a combination rather than a single method. A test strategy is a plan for the whole product life, and it should be documented rather than decided one order at a time.

Why Low Volume Changes the Decision

On a high volume line a bed of nails fixture is amortised over hundreds of thousands of boards and it pays for itself quickly. On a low volume build the same fixture may never be used enough to justify its cost, especially when a design change makes it obsolete part way through the year. The fixture also occupies storage and requires maintenance, so its cost continues even in the months when it is not used.

The decision therefore hinges on whether the value of the faults caught exceeds the cost of the fixture and of the programming. Where the product is expensive or safety related, the answer is often yes; where it is a simple board with a short production life, it is usually no. The decision should be recorded with the reasoning, because the same question will be asked again on the next similar product.

Fault Spectrum and Coverage

A test strategy should be built from the faults that are actually likely to occur. Opens, shorts, missing components, wrong components, reversed parts, misaligned parts and defective components each need a different detection method, and no single test finds all of them. Mapping the fault spectrum against the available methods is the step that most often gets skipped, and it is the step that determines the value of the whole program.

Coverage is normally quoted as a percentage of the possible fault spectrum, and the figure has to be understood rather than accepted. A method that detects every short but no parametric fault gives a high number for the wrong reason. Our short circuit guide describes the specific methods used for that class of defect. Where a fault class is not covered by any method, that gap should be accepted deliberately rather than discovered later.

<img src="https://www.gopcba.com/wp-content/uploads/2026/06/SMT车间一角.jpg" alt="Flying probe tester measuring a low volume PCB assembly” />

Flying Probe Testing

A flying probe tester moves a small number of probes to the points it has to measure, so it needs no fixture at all. That makes it ideal for low volume, for prototypes and for products whose layout changes frequently, and the programming can be generated directly from the CAD data. That also means the test can be updated in an afternoon when the layout changes, which is a decisive advantage on a product that is still being modified.

The trade off is speed. Measuring thousands of points sequentially takes far longer than pressing a board onto a bed of nails, so the cost per board is higher and the test time may become the bottleneck on a larger order. Our flying probe guide explains how the equipment is set up and programmed. For an order of a few hundred boards the slower cycle time is rarely the constraint, and the saving in fixture cost is real.

In Circuit Test with a Fixture

An in circuit test presses the board onto a bed of nails, so every measurement is made in parallel and the cycle time is short. It measures component values, checks continuity and can power a small part of the circuit to test a functional block.

The fixture is the problem for low volume work. It has to be designed and built for a specific layout, it has a lead time, and any change to the board afterwards may require it to be rebuilt. Probe access also has to be designed into the layout, which is a constraint on the designer. On a low volume program the designer often has more freedom than on a high volume one, so probe access can be negotiated rather than imposed.

Functional Test and System Test

A functional test powers the assembly and exercises it as a working device, checking the outputs against expected behaviour. It gives the highest confidence per board because it tests the product rather than its parts, but it needs a test harness, a program and a fixture or a bed of nails to connect it. The development effort for a functional test is significant and it is repeated for every product variant, which is why many low volume builds start with structural test only.

System test goes a step further and tests the assembly inside the product it will be fitted to. It is the most realistic test of all and the most expensive to set up, which is why it is normally reserved for safety related or high value products. Where the volume is low and the value is high, the extra effort is often easier to justify than it would be on a consumer product.

Boundary Scan and Built In Test

Boundary scan uses the test access port on a complex device to drive and observe the pins around it, which gives good coverage of interconnect faults with no mechanical access at all. It also tests the connection between the device and the board, which is exactly the fault class that a visual inspection cannot see. It is particularly valuable on a dense board where probe points are scarce.

Built in test moves the testing into the product itself, using firmware to check the hardware and report a result. It is inexpensive per unit once the firmware exists, and it can be run at any point in the life of the product rather than only at the factory. The limitation is that it can only test what the firmware has been written to check, so the coverage has to be analysed rather than assumed.

Combining Methods

Most low volume programs combine two or three methods. A flying probe test covers the structural faults and the component values, boundary scan covers the dense interconnect, and a functional test confirms that the assembly works as a system.

The combination should be designed so that the methods do not duplicate each other and do not leave a gap between them. Mapping the fault spectrum against the methods on a single sheet is a simple exercise that usually reveals both the overlap and the missing coverage. Our test coupon guide covers the bare board side of the same planning. The aim is a program in which every fault class has an owner and every method has a defined purpose.

Cost of Test Versus Cost of Escape

Every test method has a cost per board, and every escape has a cost that depends on when it is found. A fault found at in circuit test costs a repair, a fault found at system test costs a disassembly, and a fault found by the customer costs a return and a reputation. Putting those figures side by side is what turns a technical discussion into a decision that a project manager can support.

The comparison should be made with realistic numbers. A cheap test that misses one percent of faults is expensive if those faults are found late, and an expensive test is justified if it catches a fault class that would otherwise reach the field. The calculation should be revisited when the volume or the product life changes, because the answer moves with them.

Process Control Points

The test program itself has to be controlled: the revision of the test software, the fixture or the probe program, the calibration of the measuring instruments and the limits applied to each measurement. A program that is not under revision control will produce results that cannot be compared over time. The limits should also be reviewed after a process change, since a gap that was acceptable on the old process may be a defect on the new one.

Results should be analysed rather than only acted on. The first pass yield, the failure pareto and the repair record together show whether the testing is finding real problems or generating false calls. Our quality documentation describes how these results are recorded at gopcb.

FAQ

Is flying probe a replacement for in circuit test? For low volume it often is, because it removes the fixture cost and the lead time. For high volume the bed of nails fixture is usually faster and cheaper per board.

What coverage should a test strategy aim for? Enough to catch the faults the product can realistically suffer, with the residual risk accepted in writing. A coverage figure should always be quoted with the fault classes it includes.

Can functional test replace structural test? It can find many faults but it locates them poorly, which makes repair slower. The two are complementary, and the usual arrangement is structural test first and functional test afterwards.