The adoption of intensive automation—offshore manufacturing where labor rates are lower and outsourcing goes to OEMs— is now recognized as a given for any major player in the mobile phone industry. However, further reducing costs will require a more subtle approach. A closer examination of the cost vs. benefit of quality assurance testing is an obvious target. Over time, as other costs have been reduced, product testing has become an increasingly significant proportion of the total cost. While efforts to reduce costs associated with test have been successful, they are countered by increased product complexity and time-to-market pressures, requiring more rather than less testing as well as higher priced capital equipment. Where possible, manufacturers have adopted reduced test scenarios backed by batch sampling, but this approach risks product quality. The era in which quality is assured by using functional based test methodologies is rapidly coming to an end. This test methodology treats the device as a mobile phone and tests it by making and receiving phone calls during which parametric performance is measured. This approach might have been relatively efficient when terminals were just single-band mobile phones, but this is no longer the case. Making and receiving calls requires the test system to emulate a radio base station supporting various protocols rendering the test equipment overly complicated and therefore expensive. Moreover, the testing of terminal performance parameters requires the device to be placed under the control of a system-specific signal protocol (designed to be a robust communications protocol and not a high speed manufacturing process control protocol), which is not conducive to the needs of manufacturing. This type of testing has served the industry well, especially as the market proliferated and manufacturers raced to gain market share and brand value. As cost leadership is now key to future success, one needs to take a more informedapproach to quality assurance testing. 
The ‘open’ approach
One emerging method is to treat the device as ‘open’. A mobile phone device can be viewed as a computer to which a number of RF and other peripherals can be added. The quality assurance process is no different. Its purpose is to provide assurance that no errors have occurred in the supply chain, component handling or assembly process and it is not a means of re-verifying the design. The difference is that by considering the device as open, one can implement a test process that is optimized for speed. In this approach, the test requirements of the device are an intrinsic part of the design, to the point where a large measure of self-testing can be accomplished. However, designers of mobile phones must consider many factors, only one of which is design for test. Adding components that consume energy, space and cost over what is the bare minimum is hard to justify purely for the benefit of design for test. Designers are encouraged to use as few, common, lower grade components as possible. RF component tolerances compound to give rise to variations of performance sample-to-sample that must be compensated for by alignment procedures. RF circuit designs behave differently as frequency, level and temperature varies. Techniques for compensating for these effects varies between manufacturers, but is largely influenced by the choice of chipset used. The number of measurement pointsrequired over which correction data is calculated can be very high. The time associated with calibration is thenanother source of cost that can be influenced by choice oftest equipment and test philosophy. Suppliers of mobileterminal chipsets have recognized this and have introducedspecial alignment modes that accelerate test time bysequencing through a predefined number of device statesautomatically rather than as a series of static conditions.This approach requires test equipment that can performsynchronously. The equipment is configured in advance tomirror the change in terminal condition, e.g. as it changestransmit and receive frequency or level. This approachcertainly helps to speed the test process, but can still resultin test instrument hardware being relatively underutilized. Closer examination of the manufacturing process reveals conventional non-signaling based test equipment has a one-to-one relationship with the device under test. This means that for the duration of the test process, the test resource is available only for that device. Moreover, the device is only actively using the test equipment hardware portion while acquiring a signal from the device or while stimulating the device. Between each test or while processing acquired data the hardware element is idle. Test equipment hardware is made up of a forward path providing stimulus signals for receiver alignment as well as verification and a reverse path for transmitter alignment and verification. In most practical cases, these resources are not used simultaneously and the relative difference in complexity for Rx and Tx test generally means that hardware used for Rx is very under utilized and therefore redundant most of the time. This accounts for the poor utilization of conventional non-signaling-based test equipment. By decoupling the one-to-one relationship between device under test and the test equipment as well as decoupling the relationship between the test equipment hardware transducers and their signal processing parts, far greater levels of hardware utilization can be achieved, which lowers the cost of test through more efficient use of capital equipment. Flexible solution
One solution from companies such as Aeroflex is a high speed modular platform based on an industry standard, PXI. The platform is differentiated by its ability to maintain both open hardware and open software frameworks. This allows it to retain the maximum flexibility offered by the PXI standard and ensures hardware utilization is maximized. Not only is the equipment intrinsically fast responding, but also its software can be used in application development that optimizes test sequences. This ensures terminal and test equipment idle time is minimized and CPU processing time is optimally utilized. This flexible solution can be deployed in new ways, so that multiple devices undergo test together each seizing access to the hardware as it becomes available while multicore processors deliver high speed parallel processing. About the author
Tim Carey is a Product Manager for Aeroflex International Ltd with responsibilityfor PXI modular instruments. Tim has a 14 year history within T&M product marketing and prior to that spent 13 years in airborne radar system development. Tim has a Bachelor of Science degree in electronic engineering and CIM diploma inMarketing. He can be reached at tim.carey@aeroflex.com | 
