A practical approach to production test of UWB devices: Part 3
Adam Smith, Verigy -- Test & Measurement World, 12/21/2007 6:31:00 AM
Ed. Note: This article is part 3 of a 3-part series. Read part 1 and part 2.
Regarding test cost, we have a very good model, using Wi-Fi as an example, to provide test cost expectations for ultra-wideband (UWB). A reasonable amount of test cost per device in the wireless LAN market is in the $0.05 to $0.10—this is cost added to a device solely due to test. Using a simple calculation model on a mid-range RF tester at a rate of $125 per hour in an out-source assembly and test (OSAT) house, with 60% utilization and a 3.5-s test time (including mechanical indexing time), the approximate test cost is $0.12 per device. In order to reduce this cost towards the $0.05 range, either a tester with a lower hourly rate must be used (generally legacy testers with lower performance), or the test time per device must be reduced (which may mean reduced test coverage).
Cost-of-test targets for UWB
Using the numbers from the previous example, in order to achieve $0.05 per device, the test time either needs to be reduced to less than 1.5 s (including mechanical indexing time), or the hourly rate of the tester needs to be $50 per hour. Both of these conditions are extremely unlikely, so a combination of both will need to be required.
Another technique used to reduce cost in production test is to perform “multi-site” testing, which is the concept of testing multiple devices in parallel. This is a very common technique in the mobile-phone device market. However, this model only is cost effective for devices with high volume because the economic model only works if the test equipment is fully utilized. Since the entire global UWB market is projected to be relatively small through the next couple of years, multi-site testing is not an advised option as it will not lead to significant cost reduction for any individual device manufacturer.
Native testing
Testing a device as it is going to be used in the final application is called native testing. This gives the manufacturer of the device a high degree of confidence that when it ships to a customer, it will perform to specification. The cost model above shows another reason why a UWB-specific, native-test solution provided by the ATE manufacturer is not going to be a viable option for this market.
Native testing requirements for UWB fall into two categories: transmitter requirements and receiver requirements. On the transmitter side, a tester resource would need to stimulate a modulated base-band I/Q signal with approximately 264 MHz of bandwidth (half of the 528-MHz channel). Design validation would actually require beyond that for out-of-band measurements. Then, the test equipment would need to have the capability to down-convert a signal at up to 10.5 GHz down to an instrument that could acquire more than 528 MHz of bandwidth (preferably more than 1.5 GHz of bandwidth to be able to see the adjacent channels). The appropriate digital signal processing (DSP) would then be applied to demodulate and determine the quality of the transmitter.
On the receiver side, the test equipment needs to stimulate a modulated RF signal up to 10.5 GHz with a bandwidth of at least 528 MHz. Then, the tester would need to acquire a base-band I/Q signal with roughly 264 MHz of bandwidth. The appropriate DSP would then be applied to demodulate and determine the quality of the receiver.
Breaking this native solution into very general instrumentation, the following would be required:
1) RF stimulus with >500 MHz modulation capability up to 10.5 GHz,
2) RF receiver with >500 MHz demodulation capability up to 10.5 GHz,
3) base-band signal generator with >264 MHz modulation capability, and
4) Base-band signal digitizer with >264 MHz demodulation capability.
The immediate problem with this solution is that although items 1 and 2 are commercially available, the expense is prohibitive in a production environment, particularly because of the 528-MHz modulation/demodulation requirement. Most off-the-shelf signal generators can handle only up to 100-MHz modulation capability. Items 3 and 4 are more feasible in an ATE environment (though expensive), but are useless without 1 and 2.
Again, the solution is directed away from native-test techniques. If native testing is an absolute requirement for initial production lots, it is advisable to incorporate the instrumentation from the design validation testing into the ATE environment. This will not yield an inexpensive test solution; however, it will guarantee quality.
Design for Test
Because a native-test solution for UWB looks to be a very unlikely candidate for high-volume test, the UWB device companies can consider some steps to make test easier for these devices. In addition to digital scan techniques that have been largely adopted for testing the quality of the digital sections of the device, the following example techniques can be adopted into the design process to provide greater test coverage. These can allow the device manufacturer to take a more proactive position in finding a test solution for their silicon:
• Built-in self test (BIST). For devices with integrated baseband, a BIST technique can be used. An example of this would be a test mode that internally loops the UWB transmitter to the UWB receiver. This can be thought of as “RF scan” because it would allow the entire end-to-end chain to be functionally tested by sending in digital bits into the transmitter, and compare those to the digital bits that are received by the receiver. While this won’t tell anything about the quality of the RF signal, it will tell the manufacturer that the device is functionally viable.
• Analog test modes. For discrete transceivers that do not contain the base-band, an analog test mode can be added for additional test coverage on the transmitter. This would be accomplished by adding digital-to-analog converters (DACs) inside the device to generate the output spectrum. As mentioned above regarding native testing, generating the base-band signals with greater than 264 MHz of bandwidth is going to be expensive, due to the specialized equipment required. (Refer back to Figure 2 in part 2 to see how specialized the UWB market is). This test allows for spectral purity tests to be performed without the use of specialized baseband instrumentation.
• “Golden device” testing. Essentially, this technique tests the device with a known-good device. This is particularly useful on the receiver side because, as mentioned above regarding native testing, most commercial RF frequency generators do not allow for the modulation bandwidth required by UWB. While the signal generators that are used in design validation can be also used in production test, the cost makes it impractical for production test. Golden device testing is viable particularly in the first phase of UWB roll-out since only a very small number of test fixtures will need to be maintained. In order to use this technique, the “golden” device must be very well understood through characterization, which makes this technique difficult to apply across many test fixtures.
• Frequency conversion on the test fixture. As mentioned, the channel frequency of UWB exceeds the traditional frequencies supported by commercial production test equipment (UWB requires up to 10.5-GHz channel frequency). In order to provide test coverage for the high frequency requirement, the device manufacturer can place circuitry to up-convert lower frequencies to higher frequencies from the tester to the device or down-convert higher frequencies from the device to lower frequencies that the tester can natively measure. This technique, as in golden device testing, is viable in low-volume application. It becomes less manageable to maintain across many test fixtures, but certainly can be used when time to market is driving the requirement.
The remaining test coverage for UWB devices can be provided using traditional continuous-wave (CW) methodologies. Yield analysis will also be a useful tool in reducing the cost of UWB production test. This allows the manufacturer to remove tests based on the rate of failure, thereby effectively reducing the test time. If a test never fails, testing that particular parameter wastes valuable milliseconds.
With a practical approach to UWB production test, a combination of at least one of the techniques above combined with traditional methods can be used. By implementing these techniques, UWB can be tested on traditional production test equipment without the need for specialized test solutions. Requiring a “special” test solution can introduce risk both because of cost and availability.
Channel-frequency challenge in high volume
This discussion regarding a DFT or practical approach to testing UWB in lower volumes naturally leads to the question of how to approach the high-frequency requirement in a higher volume environment. As mentioned before, exotic test fixtures are acceptable in the early adoption phases of UWB, but the test-equipment manufacturer must provide a viable high-volume solution if UWB is to have market success. This solution must also be commercially viable for the test equipment manufacturer.
It is clear that a solution must be provided for the higher frequency UWB requirements; the approach to that solution must also match overall market conditions. Again, by using Figure 2 in part 2 as a wireless market guide, simply redesigning production test equipment to address greater than 10.5-GHz frequencies would solve the technical challenge for UWB customers. However, this would create an economic problem for all other wireless device markets.
Customers in the mobile-phone business, for example, would be purchasing capability that they could not possibly use. In addition, for the subcontractor test market, it is common to price hourly rates based on the overall configuration of the test equipment. This additional capability in the test equipment could increase hourly rates for all customers, not just UWB customers.
A more pragmatic approach to this challenge is to add an extension to the existing frequency capability of the ATE. This would essentially put the frequency up-conversion and down-conversion responsibility suggested in the DFT techniques into the ATE manufacturer’s hands. This is a significantly more robust technique for the device manufacturer because the test fixture is significantly simplified, and any “copy-exact” issues between test setups are now handled by the standard calibration of the tester. This means that device manufacturers can expect that from tester to tester they will achieve repeatable and reliable performance. This approach, though certainly not an insignificant challenge, is a conceptually simple extension of a tester’s existing capability while not adding significant cost to the test solution.
This allows the existing testers to intersect the market requirements of UWB as it progresses from the early adoption phase into accelerated growth with the drive on production repeatability and lower cost. It allows device manufacturers to use the test solutions that they have in place today for initial volumes, with a low risk upgrade path as the market develops. In addition, this allows for configuration flexibility since not all of the RF tester resources need to be upgraded to allow for test at frequencies greater than 10.5 GHz. The main key is that no specialized new test platform is required now or in the future.
The future of UWB and production test
The consumer market is ready for a high-speed personal networking solution, and UWB seeks to fill this need. There are still several obstacles between UWB and commercial success. Beside the technical challenges mentioned, it is imperative that UWB manufacturers arrive at a standard implementation in order to achieve wide success in applications like USB. Otherwise, it will be relegated to focused solutions in the marketplace. The test requirements, though challenging, are becoming well understood by ATE manufacturers. As UWB becomes a high-volume device technology, test equipment manufacturers seek to fill their customers’ needs with a low-risk and cost-effective solution. Because of UWB’s targeted entry in the consumer market, manufacturers must seek to use a combination of DFT and traditional methods on existing test platforms to get their devices to the end consumer—the gross margins of the UWB chipset will not allow for a unique and specialized test solution.
Adam Smith is a business development engineer at Verigy. He has 10 years of ATE industry experience, focused on RF/microwave device test technology. Adam holds a Bachelor of Science in Electrical Engineering from Cal Poly, San Luis Obispo. He can be reached at adam.smith@verigy.com.
Ed. note: This article is part 3 of a 3-part series. Part 1 described UWB technology, and part 2 discussed test challenges and how they will evolve over the three phases of UWB rollout.
Talkback
Related Content
Sponsored Links
TMW Resource Center
Browse Resources by Type:
- Access to EASE (Equipment Appraisal Search Engine)
DispoDog | Product Demo
VIEW NOW
National Instruments | White Paper
VIEW NOW
National Instruments | White Paper
VIEW NOW
View More
