High-Speed Testing Essential to Cellular-Phone Production

A power meter, a calibrated receiver, a spectrum analyzer, an oscilloscope, and a bit of technique used to be sufficient to completely test, tune, and certify virtually any receiver or transmitter. Even without these, you could perform a rudimentary performance test of audio wireless links by talking into the transmitter and listening to the receiver output.

Personal Communications Services (PCS)—especially those using CDMA techniques—have seriously reduced the usefulness of these test techniques and equipment.

Two factors virtually rule out classical RF/Modulation test setups. First, the driver logic for typical spread-spectrum communications equipment resides in LSI chip sets that do not provide internal test access during operation. Even the inputs and outputs are practically inaccessible due to the ever smaller and denser packaging of cellular phones and PCS sets.

Second, the modulation schemes used in spread-spectrum systems do not produce a constant carrier or even a constant carrier frequency. Instead, a sequence of brief transmissions occurs at several different frequencies.

Another characteristic of spread-spectrum techniques in communications—a great asset in operation, but a distinct problem in testing—is the error-correction capability inherent in the equipment. The design of this type of system allows for a significant number of missed communications elements without losing or even degrading the signal.

For this reason, the old “Testing, testing…can you hear me?” is not particularly useful in determining that one or even two of the carrier frequencies are not present or off frequency. Similarly, conventional signal-to-noise or background noise measurements, previously a prime indication of adequate signal strength or receiver sensitivity, have become virtually useless for testing.

Trying to use a spectrum analyzer to confirm that all frequencies are being utilized quickly shows you why spread-spectrum techniques are preferred for secure communications. There are so many frequencies which, combined with side lobes and harmonics caused by frequency changes, produce a display that looks more like lumpy noise than a valid transmission. With the display bobbing and flickering, counting frequencies would try the patience of a saint.

At one time, spread spectrum was the exclusive domain of the military, with production levels in the tens or hundreds per year. Tests were conducted by teams of engineers who methodically confirmed proper functioning of every circuit of each unit. Under those conditions, classical techniques could be applied effectively, albeit inefficiently.

Today, with millions of cellular telephones being produced yearly, fast, effective, and inexpensive device testing is a high priority. As a result, new test methods had to be devised.

New Techniques for New Systems

These test difficulties do not mean that the devices or equipment cannot be tested, just that old techniques are not readily applicable. In short, it took digital technology to create the problem—it takes digital technology to sort it out.

A good example is the vector signal analyzer, which provides a display for frequency hopping. In this display, each frequency is a point in a square matrix. Frequency hops generate a line from one frequency to the next producing a lace-like pattern that instantly highlights the ratio of hits for each visited coded frequency.

A missing point or an overly bright point indicates a frequency-hopping anomaly. This measurement is analogous to the old center frequency and side-lobe measurement.

A measurement that replaces selectivity, image rejection, and signal-to-noise measurements is defined for CDMA equipment in TIA/EIA IS-97 and IS-98A. The designation of this new parameter is Intermodulation Spurious Response Attenuation. It quantifies the capability of a spread-spectrum CDMA receiver to operate when two adjacent out-of-band signals mix to produce an interfering signal in the receiver. The test setup is explained in the standard using conventional equipment.

However, test setups consisting of several interconnected instruments can be unwieldy and constrain measurement repeatability. “Traditionally, test setups like the one shown in Figure 1 require multichannel RF stimuli that combine two or three RF sources using external isolators, combiners, and attenuators all connected by a network of cables,” said Chris Rix, senior applications engineer at IFR Americas.

“It is difficult to maintain calibration and repeatability with such a setup. Many tests also require that the RF outputs of the two sources have absolute level tracking to make a valid measurement. This is difficult to achieve through the combining network with two independent signal sources,” he said.

As you might expect, several suppliers have developed application-specific instruments combining several signal sources into one unit. For instance, the IFR 2026Q Multisource Signal Generator is designed to conduct IS-97 and IS-98A intermodulation interference tests on dual-mode CDMA receivers. It includes the signal sources and additional circuitry to provide an isolated duplex return path to a CDMA radio test set (Figure 2). “This type of implementation gives you absolute control of calibration aspects to make fully repeatable measurements with confidence,” said Mr. Rix.

A software-based approach suited for conducting various interference tests was outlined by Steve Stanton, product marketing manager at Tektronix: “To test how receivers operate with interference, it has been common practice to create two signals—the desired and the interferer. Using suitable software and an arbitrary waveform generator, such as WinIQSIM and AMIQ, this condition can be simulated and used for evaluating receiver response.

“Similarly, for evaluating the suitability of a local oscillator (LD) for a communications system, varying levels of phase-noise performance can be simulated in software and waveforms transferred to an arbitrary waveform generator. This phase- noise impairment can be added to an IQ modulated source, and the impaired signal can be used to test the effects of degenerated phase noise on system performance,” Mr. Stanton concluded.

The Role of Conventional Testers

While new test techniques and instrumentation are essential for many PCS tests, conventional equipment still plays a role. CW generators are a case in point.

“Signal generators which provide direct modulation signals at wireless RF transmission frequencies continue to be high-end units, primarily suited for R&D labs,” said Michael Lohrer, vice president of operations at Programmed Test Sources. “But for high-quantity production testing, a far more cost-effective solution uses a CW source to perform frequency conversion to a fixed IF.

“Central to the success of this heterodyned architecture is a CW source with low phase noise so that baseband information is not lost or distorted. Also, fast frequency switching of the source can allow fast coverage of an entire frequency band and reduce test time,” Mr. Lohrer concluded.

“CW sources are specifically useful as LOs in conjunction with a mixer to downconvert a high-frequency signal to a much lower IF or upconvert the IF to the desired RF,” concurred Kim Brown of Hewlett-Packard. “This mixing technique can take one of two forms: swept IF (fixed LO and swept RF) and fixed IF (both LO and RF are swept together. To generate a fixed IF, the DUT would receive signals from two LOs, each at a different frequency.”

The HP 8360L and 83750 series of microwave synthesizers are commonly used in these applications, according to Kurt Philipsen of Hewlett-Packard. Both series have a master-slave mode which allows one synthesizer to set the frequency of the other. This feature eliminates the need for an external controller, reducing the test-equipment cost and increasing the manufacturing throughput.

CW sources also are used sometimes as out-of-band interferers for mobile radios to test receiver desensitization in the presence of a large, out-of-band signal. “The critical specification for this test is the phase noise of the interfering generator. The out-of-band CW signal must not spill into the receive band and, as a result, must have low phase noise,” commented Mr. Stanton of Tektronix.

Equipment Selection

Unattended or automated testing is the key to providing consumer devices, such as cellular phones or PCS, at low cost. This requires that instruments be equipped with bidirectional communications capability. Such facilities allow device settings and operation to be controlled remotely and to forward test results to a central location. GPIB-controllable or VXI-based devices meet this requirement very nicely.

When rapid testing is required, instrument response or settling times may become critical. Each test adjustment requires some time for the equipment to stabilize or settle at the new adjustment.

This was rarely a concern when the adjustments were made manually. However, during automated testing, it is quite possible to collect data while sources are still settling on their assigned frequency or even before a tuning command is executed. Such situations easily result in inaccurate measurements, leading to rejecting good devices.

Avoiding such problems requires two steps. First, consider the settle and external command execution times when specifying the test equipment. Second, make sure the test program has sufficient delays to allow the equipment to fully respond before collecting data.

Some numerical data to illustrate this issue was offered by Mr. Stanton of Tektronix. “If testing is performed under GPIB control, the typical tuning response time of conventional high-performance RF signal generators has been on the order of 30 to 50 ms. Each tuning step requires that the ATE wait for the stimulus to stabilize before making a measurement. As measurement times with tools such as fast sweeping spectrum analyzers have decreased, signal generator tuning time has become a significant portion of test time.”

The speed, accuracy, and linearity with which signal generators can sweep through a frequency range of interest also can be critical for some tests. New microwave synthesizers address this issue by taking advantage of a digital processing architecture and provide submillisecond frequency switching and sweeping with digital accuracy.

“There are two ways to generate swept frequency in a signal source—analog sweep or digital sweep,” said Steve Reyes, marketing manager at Giga-tronics. “An analog sweep is phase continuous and provides 100% coverage of the frequency band. Frequency response characterization will be complete, producing no frequency holes. However, the linearity of typical analog sweeps is not as accurate as a phase-locked digital sweep.

“Digital sweep uses a phase-locked mode where the oscillator is locked at discrete points. This method provides crystal-based accuracy with excellent linearity. The drawback is speed,” Mr. Reyes continued. “To avoid missing narrowband frequency glitches, the operator must program very small frequency steps. This translates into a high number of phase-locked points within the digital frequency sweep and a very slow update rate.

“New digital synthesizers, such as the 12000A Series Microwave Synthesizers, use field-programmable gate arrays to perform logic operations plus digital signal processors as secondary processors for dynamic control and fast frequency setting. With switching speeds in the 500-µs range, digital step sweeps now approach the speed of analog sweeps,” Mr. Reyes concluded.

Flexibility is another aspect of selecting equipment for the cellular phone and PCS environment. It is unreasonable to expect test equipment to be perfectly matched to widely differing applications. For example, a factory might produce 20 models of cell phones, each with similar but slightly different RF specifications.

This concern was described by Mr. Rix of IFR Americas. “Many modern production lines are capable of multitasking because the product produced on any given day may depend on distribution demands. For this reason, the equipment in the test system should be able to generate several different formats of analog and digital modulation,” he explained.

Driven by customer demands, factories may even have to build a different model each day. A well-selected set of instruments and a well-designed test setup would be able to test any one of these models at almost a moment’s notice by simply loading the appropriate programs.

RF/Microwave Test Products
Digital RF Generator Provides

Multichannel CDMA Capability

The ESG-D Series of digital signal generators incorporates a dual arbitrary waveform generator and the capability to produce up to 256 Walsh-coded CDMA channels. TDMA emulation also is included. With ESG-D, PCS components can be tested in a realistic, multisignal environment. Frequency ranges of 250 kHz to 4 GHz are available. CDMA simulation is provided with single button access. From $14,600. Hewlett-Packard, (800) 452-4844, ext. 5770.

Microwave Synthesizer Offers

100- to 600-µs Switching

The 12000A Microwave Synthesizer uses a fast-tuned YIG oscillator tunable from 4 to 8 GHz with 0.1-Hz resolution. The YIG output then is scaled to provide a 10-MHz to 20-GHz output. DSP and FPGA augment a primary microprocessor to maximize tuning speed. A linear sweep mode and up to 12 frequency markers also are provided. FM and AM modulation capability is included. Leveled output power is +15 dBm from 10 MHz to 20 GHz with a resolution of 0.01 dB.. Under $25,000. Giga-tronics, (800) 726-4442.

CDMA Test Software

Simplifies Testing

The CATS-98A Software is designed to operate a base station simulator and the WIS-98A Wireless Impairment System to perform automated testing of mobile receivers and transmitters in accordance with the IS-98A standard. The program runs under Windows NT and uses a GPIB interface. It includes calibration to accommodate test- fixture cable losses. Provisions have been made to add test modules and upgrades. Noise Com, (201) 261-8797.

Frequency Synthesizer Targets

Automated Test Applications

The PTS-3200 Frequency Synthesizer offers 5- to 20-µs switching times over a 1-MHz to 3,200-MHz range with a 1-Hz resolution. Output power is adjustable from +3 to +13 dBm into 50 W . Phase noise is rated at -60 dBc, and the noise floor is -130 dBc/Hz. The unit is fully programmable through BCD or GPIB inputs. An internal 10- MHz crystal oscillator frequency standard has provision for an external frequency standard. $14,850. Programmed Test Sources, (978) 486-3400.

Multi-Source Generator

Provides Full CDMA Test

The 2026Q Generator provides outputs to test CDMA receivers in the presence of dual-tone interference over a frequency range of 800 MHz to 2.0 GHz in accordance with IS-95A. This is accomplished without additional combiners, switches, or cabling.

The output power levels are adjustable from -13 dBm to -127 dBm. The internal RF generators may be operated independently from 10 kHz to 2.4 GHz at -137 dBm to +24 dBm with internal or external modulation and separate output RF connectors. $22,995. IFR Americas, (800) 233-2955.

Modulation Generator

Addresses W-CDMA Needs

The AMIQ Modulation Generator provides wideband-CDMA (W-CDMA) and other signal formats for use in testing RF wireless components. It offers W-CDMA formats at bit rates up to 16.384 Mchip/s. The dual-channel arbitrary waveform generator has 14-bit vertical resolution and a 100-MHz sample clock rate. When paired with an SMIQ vector signal generator, it also handles TDMA, CDMA, and IS-95 formats. WinIQSIM, which ships with the AMIQ, is a Windows-compatible waveform editor that automatically calculates a range of I- and Q- baseband and IF signals. $14,950. Tektronix, (800) 426-2200 (press 3, code 1086).


 

Copyright 1998 Nelson Publishing Inc.

 

August 1998

Sponsored Recommendations

Comments

To join the conversation, and become an exclusive member of Electronic Design, create an account today!