Backpanels Present PCB Test Challenges

Nevertheless, these advantages come at a price. With traditional backplanes, you could use a wire-wrapping tool to repair a broken connection or modify signal routing, but PCB backpanels don’t afford that flexibility. To avoid the time and expense of replacing a bad PCB backpanel after product assembly, you must comprehensively and accurately test each backpanel before attaching daughter boards and cables.

Backpanels (Fig. 1) pose unique challenges for test-system designers. Whereas most daughter boards require several hundred to a few thousand test points, the average system backpanel contains from 5000 to 15,000 points. One midplane board (a backpanel with daughter cards plugging into both sides) we examined required more than 35,000 points per side—75,000 test points altogether.

Backpanels generally require contact on both sides for adequate fault coverage. Therefore, some form of hinged clamshell fixture represents the norm for backpanel test. About 80% of backpanels contain, in addition to daughter-board and cable connectors, components such as resistors, capacitors, inductors, and switches. Some backpanels include EEPROMs or other complex digital devices that require testing, further complicating fixturing.

Testing backpanels often involves probing at different heights on the same board. Although some nodes reside on the board surface as they do with daughter cards, backpanels can include a variety of connector types with profiles as high as 1.5 in., requiring fixture pins of different lengths for access.

Making Connections
To ensure that the board works properly, a test should emulate the contact environment in the target system. Therefore, the tester should access the board through the connectors themselves, rather than from corresponding test nodes on the board’s underside. In addition, the test must verify mating with the related daughter card, which means contacting the top of the connector.

Two different connectors that permit relatively easy probing individually can pose significant problems when they reside on the same board, mandating different probe configurations and lengths on one test fixture. An isolated connector can also present test challenges: To ensure that it’s actually present, your tester may need to probe the top and bottom of the board at the same point. Figure 2 shows a backpanel test fixture with different-length probes.

A backpanel’s design may demand a fixture that concentrates many probes in small areas. During testing, these probes exert unbalanced pressure on the board surface. The resulting forces may deform the board, reducing fixture probing accuracies, and in extreme cases causing solder-bond breakage and other mechanical failures. Therefore, fixtures should include spacers, supports, and stops to balance the pressures across the board, minimizing deformation and its accompanying problems.

Backpanel fixtures can cost as much as the systems to which they attach. Incorporating rigid rather than spring-loaded pins can alleviate some of the fixture’s cost. When a backpanel allows probes of uniform height, a single-side, rigid-pin fixture for a universal-grid test system (Fig. 3) serves well. You can assemble this type of low-cost fixture quickly because it does not require any wiring, and you don’t have to cut probe pins to different lengths.   

Most bare-board testers use this fixture construction. But if you use this type of fixture and don’t trim the pins, you risk damaging the board under test. In addition, the probes usually do not contact the board perpendicularly, but usually at a slight angle. The resulting side-loading on a connector can cause poor contact at the connector/test-probe interface and can damage the backpanel.

Dealing with High Profiles
A universal single-side, differential-probe-height fixture, shown in Figure 4, offers similar advantages and drawbacks, but lets you access even high-profile connectors. Assembly takes longer because you must trim pins to different and specific lengths. Accurate pin trimming is even more important here than with universal-length pins.

If you require probe-head styles that are unavailable as rigid probes, or if you want to minimize the chance of backpanel damage, you can choose a universal single-side fixture with a wired clamshell (Fig. 5). Here, spring probes on the top-side access plate accommodate a variety of connector heights. The higher-priced spring probes, the additional wiring, and the need for transfer probes that carry signals from the bottom test interface to the top test fixture make this an expensive alternative that is more difficult to assemble. Still, spring probes provide more reliable performance than rigid pins do. Also, the perpendicular spring-probe contact and the buffer of mechanical compression reducing pressure on the board minimize the chance of backpanel damage.

Figure 5. This universal single-sided fixture with a wired clamshell incorporates spring clips to accommodate a variety of connector heights on the backpanel under test.

The most reliable fixture type features spring-probes and wire-wrap technology on both top and bottom fixture plates; this type of fixture provides easy access to the connectors and is least likely to damage the product. This design may feature a hinged construction, or the top plate may move separately in conjunction with the press unit of the test system. Hinged fixtures are generally more expensive, but may be more compatible with older testers.

Merging Solid and Spring Pins
One fixture option forges a compromise between the ease of assembly offered by the rigid-pin types and the reliability, probe-head selection, and freedom of movement available with spring-probe fixtures. This technique, which merges a solid-pin test fixture with a spring-probe-filled top plate, works well on universal-grid test systems that provide grids for both the top and bottom of the board. Top and bottom fixtures connect separately to the test system through independent interfaces, avoiding the need to wire transfer pins and thereby speeding fixture construction and lowering its cost. In addition, the solid-pin portion of this fixture can independently test the bare board.

As fixtures become more complex, you need to pa;y greater attention to how you handle, store, and maintain them. Backpanel fixtures are often large (as large as 3x4 ft and weighing 60 lbs. or more), making them awkward for an operator to cart to and from a storage area.

As a result of bumps along the way, a fixture put away in working order may emerge for next use exhibiting numerous faults. Even a few failures among 20,000 wires can be difficult to find. The variety of probe sizes and styles can mean widely differing probe lives, complicating probe-replacement scheduling. You should therefore take extra care in handling these fixtures, and you should monitor reported failure patterns to promptly uncover bent or broken probes as well as other fixture problems.

For high-volume backpanel production, a probe-pin-based conventional fixture design is the logical choice. Test systems employing this design can generally test a PCB backpanel in from 30 s to 5 min.—including setup times.

For low-volume applications, perhaps a few panels per day, a collection of paddle cards wired or cabled to the test system and plugged directly into connectors on the backpanel offers numerous advantages. The cards themselves are usually less expensive than fixtures.

Also, because each paddle card plugs into a specific connector type, the same paddle card can test any backpanel containing that connector, alleviating both the cost and time required to build individual fixtures. In fact, once you have paddle cards for a particular collection of connector types, building a fixture for the next board containing those connector types requires you to simply mix-and-match the cards.

Still, someone must design, build, and test the paddle cards—usually a custom manual operation. Maintenance can prove difficult, and because paddle cards require more direct handling (repeated manual insertions) than standard fixtures, they are more prone to breakage. The enormous number of connections required for each test makes ensuring wiring integrity or verifying broken connections much more difficult.

More significantly, setting up a backpanel test with paddle cards can prove a long and painful process. A setup time of several hours to test a single board is common. Throughput often cannot exceed two or three backpanels per shift. Nevertheless, where this solution is practical, it can reduce fixturing costs. And because paddle cards can be used for several board types, they require less storage space.

To address particularly pernicious connector configurations, you can attach paddle cards to conventional fixtures. In this arrangement, the fixture includes cabling to the appropriate mating connectors. During test, the operator manually attaches these paddle cards to the backpanel under test just as with the paddle-card-only solution.

Another type of compromise can help if you have only paddle-card test equipment and suddenly face a high-volume test requirement. You can build a special fixture and then use a standalone press unit to bring the fixture pins in contact with the backpanel under test. The fixture mates to the test system through connectors on its back side. An operator plugs the paddle-card into the fixture only once, essentially eliminating per-board setup time during production test. The operator may actuate the press unit manually or (for higher point counts) with pneumatic pressure.

Planning the Test
Forging a connection between backpanels and test equipment represents only part of the battle. Testing these boards presents its own challenges. You can achieve the best results from a four- or five-level test process. Note that definitions of these levels vary from backpanel manufacturer to backpanel manufacturer, but the concepts remain constant.

Level 1 consists of a thorough isolation (shorts) and continuity (opens) test of the bare (unassembled) backpanel. You can create a test program from a CAD-generated (often in Gerber format) netlist. One alternative—a self-learned test from an allegedly known-good board—can miss repeating faults, such as those caused by scratches on the artwork used during fabrication. It may also incorporate into the test missing or faulty connections within the test system or fixture, leaving some areas untested.

Although the shorts test seems familiar to anyone who performs this test on conventional boards, the opens test differs dramatically from its counterpart at in-circuit test. In testing backpanels for continuity, the fixture must probe each end of every trace (network) or trace branch on the board.
Because in-circuit testers require only one test point per net, they do not include enough test points to execute such an elaborate test. Continuity in this case generally means a trace resistance of 20 V or less. Shorts testing looks for thresholds often set at 10 MV or higher.

Level 2 tests for shorts and opens on the assembled backpanel. This single-threshold test usually involves only low-voltage stimuli. In fact, a bare-board-tester’s high-voltage capability could damage some of the backpanel’s components. Note that Gerber files can furnish complete netlists (test programs) for Level 1, but obtaining comparable information for loaded backpanels is more difficult. Fortunately, new fixture-design software is emerging to compensate for onboard components.

In some cases, a Level 2 backpanel fixture incorporates special probes called switch probes. These probes target keyed locations on the board to determine if certain connectors are present or oriented properly. A misoriented connector causes a switch probe to compress differently from a correct one, allowing the test system to identify the problem.

Level 3 examines passive components (usually resistors and capacitors) for value and tolerance, generally on an in-circuit tester or manufacturing-defects analyzer (MDA). This level usually also tests any relays on the board.

Some companies combine Levels 2 and 3 into a conventional in-circuit test. If you perform Level 2 testing separately, however, Level 3 need not repeat the shorts-and-opens test, thereby significantly simplifying fixture construction and reducing required test-point count.  Level 4 is an in-circuit test of the board’s active components, if any, and is therefore a power-on test.  Level 5 performs a functional test through the backpanel’s inputs and outputs and often a few internal points.

Cut Unneeded Levels
Not everyone includes all levels in a test strategy. If a board contains only connectors, for example, testing may be complete after Level 2. You might substitute Level 5 for Level 4 on certain boards. Consider a company that builds a backpanel containing passive and a very few active components. Rather than commit to a full in-circuit test merely to verify components left unchecked by Level 3, the board proceeds straight into the low-level functional (Level 5) test.

You may reorganize the levels to reflect test-equipment availability. You might have an adequate Level 2 test system, for example, but not a Level 3 system with a sufficient number of test points. If the board contains components at all—passive or active—strict adherence to the levels strategy would require following the Level 2 shorts-and-opens test with a full in-circuit Level 3 test.

Some manufacturers dislike this approach because it requires constructing an in-circuit fixture for all boards that include at least passive components. They prefer to acquire a test system designed to perform Level 2 and Level 3 tests with a single fixture. If a backpanel also contains active components, you can follow Level 3 with a Level 4 in-circuit test, or you can modify the strategy to perform both Level 3 and Level 4 tests simultaneously on the in-circuit tester.

Some backpanels permit automated-optical inspection (AOI). AOI does not require fixturing and its associated costs. Inspection may also look for incorrect silk-screen legends and similar quality-control problems that electrical techniques cannot find. Optical inspection, however, does not pass current through traces, components, or connectors on the backpanel and therefore will likely complement rather than replace conventional backpanel testing for the foreseeable future. T&MW

Bill Maillet is a test engineer at ECT Test Services in San Jose, CA. He previously founded and ran TQC Testlabs. He received his undergraduate degree in engineering from University of California at Berkeley.

Kevin Wheel is operations manager at ECT Test Services in Nashua, NH. He has more than 15 years experience in the test industry and holds an ASEE degree from Vermont Technical College.