Flying Probers Circumvent Loss of Test Access

As a result, current bed-of-nails test practice — using manufacturing defects analysis (MDA) or in-circuit test (ICT) — has become a compromise. The major failing of these methods is the requirement for test access to pads or filled vias at every node or network on a board. With the latest generation boards, 100% test access for a bed-of-nails system can be almost impossible to achieve and, in consequence, test software can now analyse the performance of only networks, or component clusters. In addition, an ongoing drawback of these methods is that you can’t start testing until you have a fixture and until you develop a test programme.

The need to reclaim earlier high levels of fault coverage has generated a range of alternative test methods such as built-in-self-test (BIST), flying probers, optical inspection, and X-ray inspection.

No single method satisfies all of today’s test requirements, but flying probe systems are becoming preferred. Bed-of-nails test techniques still form the basis of the test method, except that flying probe systems use a low number of moving probes rather than the high number of fixed probes in ICT. Test times may be slower due to probe movements, but the method has compensating benefits. In practice, a flying prober can provide close to 100% test coverage on a board with thousands of nets of passive components and hundreds of digital devices. In addition, you can usually commence testing within a day of receiving CAD data and a prototype board.

Today’s generation of flying probe test systems are light years ahead of systems available even five years ago; top-end systems conduct up to fifty tests per second and provide true ICT and powered test.

Process Errors Top Fault List
Most engineers have an idea of the fault spectrum generated by the manufacturing process for a printed-circuit-board-based product. Faults can be due to design parameters, human error, placement or glue failures, component faults, or solder problems. Figure 1 gives you an idea of the order of magnitude for different faults.

Figure 1. A typical fault spectrum for printed circuit boards.

Solder shorts and opens are the most common faults and can occur at any stage in a production life cycle. Placement faults are more rare but, again, they can occur throughout the production cycle, and then probably in batches due to machine set-up variations.

Component value problems may be due to human error in the supply chain and they can be catastrophic if you don’t locate them quickly.

Shorts Can Go Undetected
Short circuits can be solder splashes, bleeds between device legs, surface mount links fitted by mistake, or misplaced devices.

In a conventional bed-of-nails test system, a short circuit test is quick and easy if access is available to all networks via probes. Problems arise if there is limited access because complex networks don’t have test pads. An impedance comparison technique (that analyses clusters of components) can detect gross errors. But, the technique is ineffective if you want to check low impedance circuits, in particular if the cluster includes low value resistors and high value capacitors.

A further problem, which many testers overlook, is the effect of a short circuit between used and unused legs of a complex IC. It’s rare to find test pads for access to unused IC legs. It’s too easy to overlook the possibility of unexpected logic levels appearing on an unused pin that may, in turn, enable unused gates, facilities, or paths. This situation can and does occur and leads to unexpected and difficult-to-diagnose results at a functional test stage. A good in-circuit test on a board should reduce functional test to a verification phase only.

Using flying probe techniques that include the ability to contact on adjacent pads of an IC — irrespective of pitch — solder bridges cannot escape. To fully understand the shorts testing on a flying prober you need to examine the topography of the board under test. If tracks are not adjacent, then you can assume they cannot be shorted. The software algorithms for flying probe testers need to access and evaluate track, via, and component proximity provided by the CAD system to decide a maximum solder splash or bleed likely. Then, the only check is for short circuits across gaps less than the maximum predetermined dimension, which you can program in microns.

The test program generator automatically includes tests between adjacent IC legs, BGA and connector pads, then optimises the test to reduce test time. Test optimisation consists of sequencing all, or selected, types of test to produce the minimum number of probe movements and distance travelled. Top-end systems facilitate this optimisation by hardware switching that, in turn, allows every probe to become a stimulus, measurement, or guard connection on any test.

Fixtureless Method Spots Opens
ICT employs a variety of fixtureless test methods to detect opens, and SPEA flying probers use a technique called ElectroScan. The technique injects and detects signals in complex components such as BGA devices, and automatically selects resistance, impedance, or current flow with proximity detection.

Probers detect open-circuited passive components with ease, but discrete semiconductors or ICs may mask a fault unless your prober gives you access to specialised techniques. These techniques may use protection components inherent in the device design itself, as in a TTL device. Or, the technique may try to pass current into a device and sense that same current using a proximity sensor (see Figure 2).

Figure 2. SPEA’s ElectroScan probing technique can detect opens on large pin-count ICs. You can move the ElectroScan probe during debug to locate the optimal test position.

This fixtureless technique is useful for detecting opens on large pin-count devices because the system can move the proximity detector during debug to locate the strongest signal as it checks each pin. The test program will then store and use this position for subsequent testing.

Using this technique on a conventional bed-of-nails ICT system introduces a reliability problem. The board flexes as the vacuum pulls it onto the bed-of-nails and this, in turn, can mask unreliable contact if dry joints exist between the IC and the board. SPEA flying probers overcome this problem by allowing you to adjust probe pressure from a few grammes up to 200 gm. You can also position probes with 10-micron accuracy in order to eliminate board distortion during test.

Micron Accuracy on Pads
In general, ICT of components is becoming difficult due to the lack of test pad access on any bed-of-nails system. Top-end flying probe systems allow better and reliable access by probing to within microns of the edge of component pads, or on vias. The systems can contact solder fillets away from the component itself. Test software allows you to program an offset, from 10 microns upwards, from the component CAD datum either globally or by individual component. This feature removes the need for manual intervention, although the software still allows you to do this during debug.

Component measurement is extremely good on top-end flying probers. The SPEA 4040 system, for example, quantifies capacitance down to a few pF, resistance down to 10 mV, and inductance down to mH.

Flying probe systems with four probes can appear to have limitations because each test can have a maximum of only two guard points (assuming no edge connector access). In many cases, though, this arrangement improves capacitor measurements! Think of trying to measure component values on a circuit board with hundreds of wires connected to it, each one providing stray capacitance. This is exactly the situation with a bed of nails connected to a board. A flying probe test avoids this situation and measures only the component.

Access Via Edge Connectors
To enable you to make more connections between your test system and a board, some flying probers facilitate connections via edge connectors using a ZIF interface built into the test system.

This facility also allows you to do double-sided testing using a simple bed of nails, which provides access only those networks inaccessible from the topside. The program generator automatically accepts the information in this arrangement. You can use the arrangement to test a board in a single pass of the test programme and, therefore, reduce handling to a minimum.

Board edge connectors provide a number of other advantages. First, test speed increases because a probe does not need to visit any network that is accessible via the edge connector. The program generator automatically deduces this accessibility. Second, you can increase the number of guarded test points and reduce still more the number of probe placements.

As a further advantage, in almost all cases an edge connector contacts with a board’s power rails, which in practice connects to around 40% of all analogue components. This situation enables the test generator to reduce probe movements by around 25%. Furthermore, you can arrange to apply power to the power rails of a board under test as an initial pulse. This test checks that input current is not excessive prior to allowing glue logic and analogue devices to be continuously powered and thoroughly checked functionally.

For any product incorporating built-in-self-test, power is a necessity anyway and the ability to run self-test and monitor points using the flying probe provides a powerful fault location tool.

John Chubb is technical director with SPEA UK. John has been involved with ATE since the early 1970s and has been involved with flying probe test for five years.