|
It is now commonplace for consumers to use mobile devices in ways that were unheard of just 10 years ago. Whether it's browsing the Internet in their favorite coffee shop, listening to stereo music through wireless headsets, or downloading a large file in China after walking off a trans-Pacific flight, demand for advanced functionality is increasing pressure on mobile device Original Equipment Manufacturers (OEMs).
While this extra functionality means adding extra radios or bands, the added functionality also means extra cost. In order to maintain already slim profit margins, these costs are passed onto the consumer yet costs to the consumer must be controlled to effectively compete in an increasingly price-driven market. The only approach for OEMs to increase functionality and effectively compete is to make radical changes to the way their products are designed and manufactured.
Cost Contributors
The fundamental cost contributors for mobile devices are:
- Bill of Materials (BOM)
- Assembly
- Test
- Marketing
As mobile devices increasingly become commodity items, marketing costs willalso inevitably increase, so the focus should rest on the first three contributors for potential cost reductions.
BOM and assembly cost will likely increase with added functionality and resulting increased parts count. BOM cost reduction could be achieved by the integration of multiple functions onto a single semiconductor die or module, but this depends on technology development by the chipset supplier and is typically out of the OEM's control. Assembly cost reduction could be achieved by leveraging lower labor costs or increased automation, but these strategies may already be incorporated into an OEM's manufacturing process.
The remaining target for cost reduction is test. Since the fundamental reason for test is to guarantee mobile device quality, extreme care must be taken so that product quality is not diminished while lowering test costs.
Linkage Between Test Cost and Time
Cost of test to the OEM consists of capital costs for handlers, fixtures, and test equipment, as well as labor to provide necessary product "touches" not provided by the handlers. Cost allocated to the mobile device will vary in direct proportion to test time. An OEM that is successful in reducing test time will have the ability to reduce cost to the consumer, spend money on additional marketing, or keep the money as additional margin.
Test time is typically proportional to the number of bands and modes in the mobile device. Assuming all other things are equal, a quad-band GSM phone will take roughly 4X times as long to test as a single-band GSM phone. A dual-mode quad-band GSM/tri-band W-CDMA phone will take roughly 7X times as long to test as a single-band W-CDMA phone. As Bluetooth, WLAN, and even more global bands are incorporated into the RF and baseband processor, test time will only increase unless changes are made in the manufacturing process.
Historical vs. Modern Test
In the 80's and 90's, functional testing of mobile devices was accomplished using measurements made during a test call. The time needed for call processing during test added a significant amount of time to the overall sequence. For example, the initial registration and call setup of a CDMA mobile device prior to testing meant a penalty of over 30 seconds per call, resulting in a 25% increase in test time for a two minute functional test.
In addition to test time penalties for call-based functional test, limitations on processing power in test equipment meant that transmit and receive tests could not be performed at the same time in either calibration or functional test stages. Assuming a 10 second receive test time and a four second transmit test time, this serial testing resulted in a 40% penalty in test time over that required for parallel testing.
Loopback bit error rate (BER) measurements were originally used to determine whether a mobile device's receiver was functional. Because a BER measurement is statistical in nature, a large number of bits had to be sent and measured. These bits equated to a large amount of time to make this measurement, and posed an additional test time penalty. Modern test strategies to decrease test time include:
- Utilize parallel receive/transmit test strategies
- Eliminate the necessity for calls in the functional test stage
- Leverage new C/N-based receiver measurements in place of BER measurement
- Use fast calibration modes
All of these strategies require the presence of non-standardized test modes in the mobile device's chipset. The efficiency of test modes implemented in a chipset may actually affect capital investments needed in an OEM's factories, so product design using a more expensive chipset with better calibration and test modes could effectively save capital in later manufacturing stages.
Manufacturing Test Flow
In the first stage of a traditional mobile device manufacturing process, bare Printed Circuit Boards (PCBs) come into a manufacturing line, components and solder paste are mounted on the board using pick-and-place machines, solder paste is reflowed (repeated for the reverse side) and the populated PCB is calibrated. In the second stage, the keyboard, display, additional modules, and other components are mounted to the populated PCB, then assembly goes through final test. Sometimes, a voice call is used after assembly of the covers and battery to ensure operation prior to assembly of the consumer kit and shipment.
Calibration Methods
Mobile device calibration typically involves a series of frequency/power setup and power measurement operations by both the test equipment and mobile device.
For receiver calibration, a one-box tester (OBT) can provide a calibrated, modulated signal to the mobile device. The mobile device then makes a receive measurement, and a correction factor is generated and recorded in memory corresponding to the difference between OBT-generated level and mobile device-received level. This process is repeated at several frequencies in each band, and several power levels at each frequency.
For transmitter calibration, the mobile device is set to a specific frequency and power level, then the OBT measures the level, and a correction factor is generated and recorded in memory corresponding to the difference between mobile device-generated level and OBT-measured level. As with the receiver calibration, this process is repeated at several frequencies in each band, and at several power levels at each frequency.
Several fast calibration modes have been developed to decrease measurement cycle time and enable simultaneous measurement of the mobile device's transmitter and receiver. These methods depend on synchronization of the mobile device with the OBT to maximize test speed.
If the transmitter of a W-CDMA quad-band device is calibrated at 15 frequencies in each band and 20 levels per frequency, 1200 transmitter measurements are necessary. Along with 600 receiver measurements over the same bands, this totals 1800 measurements. Assuming these tests are done serially with a typical OBT or mobile device setup, and measurement time is 40 ms, total device calibration time is 72 seconds. Use of a parallel measurement approach with fast calibration modes reduces this substantially to 25 seconds (Figure 1).
|
