In-Process Functional Test Cuts Cost of RF Products
A protocol-independent strategy isolates RF-component defects early in the manufacture of high-volume cable modems, set-top boxes, wireless phones, and PBXs.
Mike Dewey, GenRad, Westford, MA -- Test & Measurement World, 12/1/1998
|
A version of this article ran in Test & Measurement Europe in June-July 1999. Read article in PDF format. |
||||
| For more information, Read The Case for In-Process RF Test, below. | ||||
A typical production line for today’s complex PCB assemblies includes in-process and final test (Fig. 1). For RF products, however, many manufacturers forgo in-process testing because of limited access to nodes and the presence of small-value components that in-circuit test equipment cannot reliably measure. (For instance, in-circuit testers cannot reliably measure the 1-mH chokes and 1-pF capacitors commonly used in 900- and 1800-MHz products.) |
Figure 1. A typical manufacturing process begins with pick and place operations, continues through in-process test, and ends with product assembly and final test. Makers of RF products often rely on a “build& pray” approach that skips in-process functional test, to the detriment of final-test yields and overall manufacturing cost-effectiveness.
 |
| Figure 2. Although individual RF components resist in-circuit test, you can divide a GSM cell-phone handset into functional blocks amenable to in-process testing. |
Table 2 lists in-process functional tests that can verify the functionality of the handset’s receiver and transmitter.
Note that by combining failure results from both groups of tests, you can indite or verify operation of the components in certain functional blocks. For example, correct frequency operation for receive and transmit tests would suggest that the product’s synthesizer is functioning properly.
For this example, the test system (Fig. 3) consists of the following instruments:
 |
| Figure 3. This in-process functional test system evaluates a GSM handset PCB. The UUT connects to a down converter/switching module, which in turn routes the signal to the VXIbus digitizer or generator. The system can be expanded to test multiple UUTs. |
• a VXIbus digitizer, which measures output power (to verify power-amplifier and synthesizer operation), verifies transmitter ramping functions, and analyzes the transmitter’s phase error by employing a post acquisition algorithm.
• a custom RF interface module comprising a down converter, power splitter, and RF switching. The down converter provides an IF compatible with the VXIbus digitizer’s bandwidth and sampling capabilities.
• a VXIbus RF generator; the RF generator provides the local oscillator (LO) for the down converter and also provides a CW and FM signal for evaluating the handset receiver’s receive signal strength indicator (RSSI) as well as its automatic-gain-control AGC vs. RF-level and IF bandwidth/filter
performance.
To perform the in-process tests, you must have some simple control that lets your tester program the handset’s synthesizer and initiate the transmitter’s operation. You also need the requisite connection to the UUT’s antenna port and, in this GSM example, access to the I/Q and RSSI/AGC voltage nodes. A tester also can perform calibration or flash programming at this test stage.
As noted earlier, you may be able to move some end-of-line tests to in-process test. For example, ETSI-defined GSM measurements such as power ramping, power level, power-level-setting accuracy, transmitter phase error, and frequency error can occur in process, eliminating several end-of-line tests.
Although this article has concentrated on cellular phone technology, you can use similar measurement techniques for other types of RF products. The key to a successful deployment of this strategy is to have a product with some test access and control along with a compelling economic case. Adoption of this test strategy will help solve the manufacturing costs and quality issues not only today, but also in the future, because the protocol-independent strategy provides the means to re-deploy the test system and ramp up production for next-generation products quickly and easily. T&MW
Mike Dewey is product line manager for GenRad’s VXIbus-based Geneva functional test system.
Although a few manufacturers reach the 6s quality level at final test, many manufacturers routinely exhibit failure rates of 4%, with failure rates of more than 10% not uncommon for some second-tier manufacturers.
For manufacturers producing even moderate volumes (for example, 1 million products per year), a failure rate of 4% results in significant rework costs—an aggregate labor cost can reach $250 per hour when you include ammortized costs of repair station equipment, scrappage, and time costs for the skilled RF engineers required to troubleshoot a plastic-encased product that’s not designed to be disassembled.
 |
| Reinstating in-process test in the manufacture of RF products can cut manufacturing costs, for a range of in-process-test failure-detection rates. |
To understand the economics of deploying an in-process, functional test strategy, first consider the following sample statistics for manufacturing GSM handsets with final functional test only:
• 1 million units/year production volume,
• 2 min. final test,
• 4% failure rate at final test,
• 40,000 defective units,
• $250/hr., 30-min./unit repair costs, and
• $5 million annual repair cost.
Now consider the economics of using an in-process test strategy:
• 1 million units/year production volume,
• 1 min. in-process test,
• 4% failure rate at in-process test,
• $90/hr., 30-min./unit post-in-process-test repair costs,
• 1.5 min. final test (assumes some previous final tests occur during in-process test),
• <1.5% failure rate at final test,
• $250/hour, 30-min./unit post-final-test repair costs,
• $3,675 million annual repair costs, and
• 2.5 min. per unit total test time.
The in-process test strategy yields $1.325 million in annual savings, at the cost of a 0.5-min./device increase in test time and an investment of about $300,000 in an in-process test station with handlers. If the increased test time slows the production-line beat rate to an unacceptable level, you could add a second in-process test station and still save $725,000 the first year.
This analysis does not account for the cost of the floor space required for the in-circuit test stations, but those costs are offset by reducing the scrappage that results when disassembly of a handset’s plastic case irreparably damages internal PCBs. Figure A illustrates annual savings vs. percent failure at in-process test by instituting an in-process functional test strategy.—Bob Stasonis, GenRad
