Take a slice out of test time
Automated Test Summit {Part 2 of 3}: A variety of techniques help you speed up tests and enhance their analysis.
Martin Rowe, Senior Technical Editor -- Test & Measurement World, 5/1/2004
|
Finding ways to cut test time and effectively analyze test data, often while subjecting DUTs to adverse environmental conditions, are among the subjects being covered by the 2004 Automated Test Summit (see "Summit details," below). This second of three articles discussing topics from the summit (Ref. 1) presents some pointers that attendees are learning. Simultaneous testing offers one method for shortening test time without greatly increasing costs. You can further improve a simultaneous test setup by adding a test scheduler that will make efficient use of your test resources. To uncover manufacturing defects that might otherwise escape notice, you should perform tests while stressing your products (see "Test under stress ").
What is simultaneous testing?With simultaneous testing, you use a single test system to test more than one UUT at a time. Obviously, you could perform simultaneous testing by building a duplicate test system, but high equipment costs, limited floor space, equipment obsolescence, and staffing requirements often make this an impractical option. Fortunately, most simultaneous test setups don't require two testers because UUTs can often share instruments.
"You may be underutilizing your test equipment," says Richard McDonell, product manager at National Instruments, "In the majority of today's test systems, instrument utilization rates average 30% to 50%. That's because many test systems perform sequential testing, performing one test at a time on one UUT."
A traditional tester often works on one unit at a time, performing all the required tests before testing the next unit. By switching to a parallel process, your system will test two or more UUTs independently.
In contrast, a batch process lets the tester perform the same tests on multiple UUTs at once with the UUTs beginning and ending the tests in unison. You can use the batch model to control a set of test sockets that test multiple UUTs as a group, such as when you have a set of circuit boards attached to a common carrier.
![]() |
|
Figure 1. Simultaneous testing lets you run more than one test at a time on multiple UUTs. |
Engineers at Flextronics Test used both processes when they developed a Gigabit Ethernet transceiver test system. The system has two test fixtures, each of which tests four transceivers in batch form. While one fixture runs its tests, the technicians unload and reload the other fixture—a parallel approach.
The tester also uses redundant test equipment. "We chose a redundant system architecture to eliminate idle test time and eliminate sharing of test equipment," notes Pat Cupo, director of marketing at Flextronics Test. "Redundant optical spectrum analyzers decrease test times and minimize optical switching."
You can also use simultaneous testing to minimize equipment idle time. Consider the parallel test illustrated in Figure 1. An RF test, which might use a signal generator and an RF spectrum analyzer, is the only test running at the start of the cycle. The audio test, which might use a digitizer, remains idle as does the digital test, which might use a protocol analyzer. Six test periods are required to complete all of the tests on four UUTs.
To make more efficient use of the equipment, you can use a test executive such as TestStand from National Instruments to schedule the tests based on idle instrumentation. In the scheduling scenario shown in Figure 2, the four UUTs require only four test periods. All test equipment is in use all the time and all tests run simultaneously on different UUTs.
More than timeTime isn't the only factor to consider when organizing your tests. Test results play an important role, too. By collecting test data from production lines, you can analyze test data to look for trends. In many cases, detailed data analysis can uncover causes of failures and help you determine if failures come from product designs or manufacturing processes.
"By looking at time between failures, you might decide either to alter the order of your tests or eliminate tests that hardly fail," says Nader Fathi, CEO of SigmaQuest. "If a particular test fails more often than others, you should organize your tests to run that test first." Doing so will let you focus on problems and get failures out of production earlier.
![]() |
|
Figure 2. Scheduling maximizes test equipment usage and minimizes overall test time. |
Because they're serial buses, Ethernet and USB lack the hardware triggers that you get with the parallel VXI, PXI, and IEEE 488 buses. VXI and PXI add signals to industry-standard buses that provide the timing and synchronization necessary for many test applications. The backplane of each bus contains a 10-MHz system reference clock that lets you synchronize measurements across multiple instruments. A local bus runs between any two adjacent modules in the backplane, letting you pass analog or digital signals between modules.
An eight-line trigger bus transfers any type of digital signal from a system controller to any or all instrument modules. Signals may include clocks, digital data streams, or single-shot triggers. Finally, a star trigger can synchronize modules or provide variable-frequency clocks.
Simultaneous testing, scheduling, and test architecture all improve test throughout and minimize costs in design verification or in production testing. Not all of these tools will work in every application, so evaluate them thoroughly.
| Bus | Throughput (Mbytes/s) |
| IEEE 488 | 1 |
| 8 (HS488) | |
| VXI | 40 |
| 80 (VME64) | |
| 160 (VXI 3.0) | |
| PXI/CompactPCI | 132–264 |
| 10/100/1000 Ethernet | 1.25 /12.5 /125 |
| USB 2.0 | 60 |
| IEEE 1394 | 100 |
| Summit details | ||
| The Automated Test Summit is a series of full-day conferences, including technical papers and seminars, presented by the companies listed below. To learn more about the Automated Test Summit—including conference dates and locations—visit www.ni.com/testsummit. | ||
| Flextronics Test, www.flextest.com | JTAG Technologies, www.jtag.com | National Instruments, www.ni.com |
| SigmaQuest, www.sigmaquest.com | Tektronix, www.tektronix.com | Thermotron Industries, www.thermotron.com |
| Virginia Panel, www.virginiapanel.com | ||
| Reference |
|
|





















