Cost pressures and ever-shortening production cycles call for fast, efficient manufacturing tests for mobile phones. Toward this end, it is helpful to take a look at two basic test phases during the production of a mobile phone: module test and final test.
The module test involves adjusting the phone, permitting an initial-but only partial-determination of the physical compliance with specifications. Next, a signaling test simulates actual operation of the phone. This second production test step verifies the results of the first step, while also ensuring the phone's compliance with specifications in those areas that cannot be adjusted.
The goal of a production test, along with any resulting test procedures, differs significantly from the further-reaching test objectives set during product development. Production tests are focused on Layer 1 of the OSI reference model; that is, on the physical characteristics of the mobile phone. Signaling, which addresses all layers above the physical layer, is the means to this end.
Proper implementation of signaling as it is used to set up calls, change channels and so on will have previously been verified during a mandatory conformance test. Unlike the physical characteristics of a mobile phone, signaling, like software in general, is not subject to variations. In other words, as long as a test procedure remains the same, it is reasonable to expect that the software will behave consistently. This, in turn, means that a production test's speed can be improved by testing only those characteristics that are subject to variation. It makes no sense to spend time and money on testing areas that have never before failed.
Production testing of a mobile phone starts with the module or nonsignaling test and adjustment. During this test, the phone is controlled via a service interface, which is also used to enter the adjustment values into the phone.
First, the local oscillator must be adjusted. The frequency errors are determined for an unsynchronized phone at an average temperature. A separate digital-to-analog converter can be used to easily detune the oscillator frequency. For example, GSM sets the requirements for frequency accuracy at 0.1 ppm of the carrier frequency. Without continuously synchronizing the oscillator, it would be impossible to meet this requirement.
In parallel to this frequency-error measurement, modern test equipment can also evaluate other relevant data, such as the I-Q imbalance and the origin offset. This data, which is used for evaluating modulator characteristics, makes it possible to optimize the suppression of the carrier and sideband.
Next, the output power of the phone is calculated. To achieve high efficiency, compromises that are necessary in the hardware design are compensated for with the use of certain adjustments. The number of test positions required depends on the mobile-radio standard being referenced.
The output power should be determined as quickly as possible using a test device, but it requires many iterations. Modern test concepts rely more heavily on the use of the hardware capabilities and on additional test options offered by the mobile-radio tester. In this case, the frame or slot structure of the mobile-radio standard is used. In a multiframe or multislot measurement, the dependency of the output power is determined as a function of the D/A control values.
In a procedure that is still considered part of a nonsignaling test, the phone sends a ramped output signal that corresponds to the dynamic range of the power output stage. The slot or frame timing is not derived from the control channels. It is generated based simply on a typical slot or frame length. This considerably decreases the number of control steps at the service interface, optimizing the measurement on the tester.
A disadvantage compared with conventional procedures is that this mode first must be implemented for the service interface, and it is impossible to reuse this specialized routine in the signaling state. On the tester side, the time required to set the analyzer must be kept brief. The requirements for the tester must also take into account the high dynamic range and the large bandwidth of the signal, such as is the case for wideband CDMA. At the very least, the output signal's compensated frequency dependency must be calculated at a specific output level for the phone.
Finally, the input field strength at the mobile phone's antenna input must be determined. The actual measurement takes place within the phone. The mobile-radio tester provides a generator signal with a defined output level. The effort for a CDMA-based system is higher than that for a GSM-based system. Depending on the system, the output power of the CDMA/W-CDMA phone depends on the received field strength of the basestation signal. The accuracy of the phone's output power, in turn, greatly influences the quality of the connection and the channel capacity of the entire cell.
To calculate the frequency response compensation, the received-signal strength indicator test is also carried out on various channels for at least one reference level. As in the output power test, it is conceivable that a sort of ramped signal with slot or frame timing could be defined on the tester side as well as on the phone side. This, however, is not a common practice.
Sophisticated module and signaling testers like this one evaluate data such as I-Q imbalance and origin offset, along with frequency-error measurements.
After the module test, which is frequently carried out on the printed board assemblies without housing, the fully assembled phone is tested. This “final test” covers characteristics that could not be checked during the module test. In the receiver test, for example, the phone is switched on and synchronized to a control channel. In principle, a service interface is no longer required at this point, but having one can help shorten the test time.
The software design of the phone has a major influence on the final test. After the phone is switched on, a marketing-defined animation is frequently displayed. Suppressing this type of gimmick during testing can easily save a few thousand dollars during production. Other time savings during this phase can be achieved by omitting the channel search or detection of synchronization on the phone, so that any other, usually necessary Layer 3 signaling steps can be skipped.
A similar strategy can be used for the connection setup. Instead of Layer 3 signaling, the phone could be brought to a defined state on the service interface while the mobile-radio tester is operated in a reduced signaling mode. Once the connection is verified, the potential for optimization lies primarily in simultaneously performing transmitter and receiver measurements.
Transmitter measurements include checking the modulation characteristics, power measurements, power ramp tests and determining the effect of the output signal on adjacent channels. The receiver test is handled by returning to the transmitter all data that is received in the test. The tester compares it with the previously transmitted data, in effect making a bit-error-rate measurement. The fastest and best assessment is achieved through a bit-by-bit comparison of data.
A basic requirement, however, is that faulty data also be returned. This might sound trivial, but data connections are not always implemented to allow for this, making only block-error measurements possible. This type of measurement requires a significantly longer time to achieve statistically comparable data. Audio measurements can round out the test suite because acoustic quality is an extremely important criterion for users. In contrast, tests on data connections are useful only if the physics of the connection have actually changed.
If, however, there has been no change to the phone, the additional test effort can be skipped or reduced to sampling. Signaling tests are typically carried out on two to three channels per band.
Thomas Lutz (firstname.lastname@example.org) is a senior application engineer for mobile-radio testers at Rohde & Schwarz GmbH & Co. KG (Munich, Germany).