The roadmap for wireless chipsets shows that devices are becoming more sophisticated. Digital modulations are replacing analog modulation in wireless markets such as cordless phones, cellular phones and television.
According to Dataquest, 45 million wireless end units will be sold in 1995. In 1998, the forecast shows an increase to 68 million wireless end units.
On the other side of the road is the decreasing number of ICs needed to encrypt, modulate, transmit, receive, demodulate and decrypt. New wireless IC designs show standard IF and RF functional blocks merged with conversion and digital processing functions.
Signals encountered in wireless ICs vary from pure digital to digitized analog to pure analog at baseband, intermediate and radio frequencies. The trend is toward a single-chip transceiver with digital baseband, modulated RF and possibly signal taps in the converted baseband and IF stages (Figure 1).
The increased integration of these different signal types into fewer devices presents a host of challenges for conventional test equipment. Testing such devices requires the capability to source and measure any of the waveforms seen in Figure 1 with strict timing relationships between the digital and analog.
Wireless Chipsets
Wireless communications chipsets have two high-level tasks:
Receive a message. The chipset must detect an RF signal, then downconvert it to a lower frequency where it can digitize the signal and apply digital signal processing (DSP) to demodulate it to a digital message.
Transmit a message. The chipset must translate the message into a standard modulation signal and upconvert this signal to a higher frequency, where it can be detected through the air by an external receiver.
To fully test these mixed-signal microwave chipsets, it is necessary to source and capture signals at any point in either chain. Different vendors allocate the functional blocks differently across their chipset, although all are heading in the direction of a single device that incorporates digital I/O and RF transmit/receive.
Integrated Test System
The architecture of a mixed-signal microwave test system that incorporates all these solutions must be more than a collection of isolated instruments. Timing and event relationships must be maintained and the system must contain powerful, programmable DSP capability (Figure 2).
The microwave front end consists of eight bidirectional RF ports, providing vector network analyzer (VNA) functionality with both through and reflective measurements. Microwave continuous wave (CW) sources can be sourced directly or combined for two-tone measurements. Modulation is provided by upconversion of a high-speed arbitrary waveform generator (AWG) with a microwave CW.
Signals from the DUT are routed via couplers to the system’s downconverter. The downconverted signal can be accessed by a high-frequency digitizer for wideband analysis and/or demodulation. This can be achieved by a precision low-frequency digitizer for greater dynamic range or by a time-measurement instrument for frequency counting and multiple interval measurements. This architecture allows fast spectral measurement without the need to sweep frequencies.
Baseband source and measure capabilities are provided with precision low-frequency instruments which, with event control, can source and measure in-phase (I) and quadrature-phase (Q) signals. Digital I/O follows the arbitrary waveform architecture, with data processed by DSP. All pins, including the RF ports, provide DC parametric measurement.
Baseband Tests
A mixed-signal microwave tester must function as flexibly with a digitized analog version of a waveform as it does with a pure analog version. Depending upon device design, the tester might have access only to the input or output of an ADC and DAC. This doesn’t present a challenge for a mixed-signal tester with digital I/O, digitizers and AWGs.
A typical sourced analog waveform might be a pair of sinusoids or fully modulated signals. Delivery of the signal involves clocking a digital representation of the I and Q signals through an AWG. To maintain the 90-degree phase difference between the two paths, it is necessary to have repeatable and exact triggering of both AWGs.
To determine the optimal phase balance between I and Q signals internal to a device, it may be necessary to independently adjust the phase of one analog signal while holding the other fixed and measuring a device parameter, such as the signal-to-noise ratio.
To make meaningful measurements on captured baseband signals, the capability to separate time error, offset error and scaling error from other signal distortions is crucial. The tester must process I and Q independently and process routines to make digital modulation measurements such as error vector magnitude or phase trajectory error. Phase synchronization of the digitizers on the tester becomes important or the mismatches measured would be that of the instruments.
Modulated IF and RF Tests
The IF signals contain a baseband I and Q pair that has been multiplied by an orthogonal carrier pair and summed. When device integration begins at the IF stage, this modulation must be provided in one signal from the tester.
Traditionally, these operations have been done using analog mixers and summers, giving the possibility for I/Q imbalance inaccuracies. DSP techniques, however, can provide digitized modulation, converted to AC by an arbitrary waveform source. Further upconversion of the IF signal by a microwave CW will not introduce any phase imbalance.
Other devices require direct digital IF data. Parts are beginning to use DSP and high-speed converters to source and capture signals directly at IF.
To create these waveforms, use DSP to create the digitized baseband signals and then perform the mixing and summing digitally. Devices which demodulate RF signals without access at the intermediate stages will require RF modulation, either pulsed in sync with digital capture occurring at baseband or modulated in phase, frequency or amplitude.
A captured IF signal can be used to evaluate the performance of a receiver downconversion path from RF to IF. For useful measurements, the tester must have the flexibility to digitally downconvert directly or combine analog downconversion and digital downconversion to baseband signals. As in the case of a captured analog signal, sophisticated processing routines may be necessary to determine the vital measurements from the captured signal.
Unmodulated RF Tests
Traditional VNA instrumentation measures one frequency at a given point in time. The microwave instruments provide this as well, allowing for scalar and vector (magnitude and phase) measurements, including gains, isolation and S-parameters.
The use of a high-frequency digitizer gives the additional capability to analyze the spectral content of a time-domain capture without repetitive measurements at each frequency. The spectral information yields results for many RF tests where product frequencies can be compared to fundamental tones, such as intermodulation products, spurious frequencies and device-mixer products. Spectral components are used in calculating phase noise and frequency conversion gains.
High Level of Device Integration
Due to the high level of device integration, the desired functional block being tested may reside deep within the part, inaccessible to device pins. This prevents the delivery of optimal test signals to the block and direct tester measurement of the block output.
It is difficult to determine the performance of a block without controlling the input stimulus and allowing direct access to the block output. As a result, a higher-level parameter is better aligned with the block-level observability and may be measured.
Alternately, techniques may be used that allow traditional measurements to be made despite the limited observability of the block. In both cases, sophisticated processing algorithms provide a solution.
Challenges With Device Integration
Nontraditional Converter Testing
As mentioned before, one step in transmitting a digital message is converting the digital message bitstream into baseband analog I and Q signals which are then mixed in quadrature up to an IF analog signal. This conversion involves two DACs and two analog filters.
To provide a conventional test stimulus to the DAC, such as a linear ramp or a multitone, it’s necessary to have pin access to the DAC input. Providing more functional blocks within an IC, however, runs counter to the needs of conventional component testing. In this example, the DAC might receive its input directly from a ROM lookup table or a fixed DSP block internal to the device. The only way to influence the input to the DAC is by changing the digital message.
In this case, the DAC linearity test can be replaced with a more appropriate digital modulation test, such as error vector magnitude or phase trajectory error, which is faster and provides more insight into how the DAC affects system-level performance of the IC. Also, the digital message might be manipulated to provide a single tone at the DAC input and a sine wave histogramming technique may be used to measure the DAC linearity.
Asynchronous Device Frequencies
Varying device frequencies lead to asynchronous test conditions. Measuring RF signals involves downconversion from RF to a known IF frequency and digitizing. The output frequency of a voltage-controlled oscillator (VCO) internal to the device can make synchronous capture difficult which will, in turn, lead to inaccuracy in the measurements. One solution uses a time measurement subsystem or DSP routines to determine the VCO frequency and then adjusts the VCO or the tester downconverter frequency to ensure coherent capture.
Unknown Capture Phase
Digital modulation test, such as error vector magnitude, requires knowledge of the signal at exact time instants. Variations within devices prevent time alignment of the digitizer sample instant to the device on the first pass. Signal processing can be used to time-align the samples or the digitizer sample instant can be adjusted and another capture performed.
Integrated Calibration
Time and level accuracy must be calibrated over time intervals and temperature drift. At telecom and baseband frequencies, passive paths to the DUT deliver signals accurately. Active circuitry between the tester and the DUT can be calibrated with simple magnitude-gain measurements.
Switched calibration paths are made inside the test system, allowing for automated instrument calibration between the device test runs without stopping production. Even high-speed digital pins are calibrated internally, with the exception of time domain reflective measurements to the device pins, to compensate for digital phase skews.
RF frequencies require complex signal calibration in the form of S-parameters so source and measure signals are calibrated for the impedance mismatches of the transmission paths. The device interface hardware can be characterized for the S-parameters on the tester, the bench or simulator. Since the paths are passive, they are not subject to temperature drifts and the error terms remain constant.
The difficulty of making calibration measurements of the instrumentation itself lies in the practicalities of the production environment. Conventional methods of RF instrument calibration rely upon putting standards on the tester in place of the device interface.
This forces a choice between poor accuracy and undocking test heads from handlers/probers to insert the calibration fixture whenever ambient temperature changes. The use of internal cal standards in the microwave instruments results in no standards on the device interface plane. All tester instrument calibration is performed automatically, transparent to the production environment.
Conclusion
The integration of mixed-signal functions with RF and microwave capabilities requires an integrated test solution. A test-system architecture must be flexible and configurable for different combinations of functional blocks while allowing for end-to-end test of the device. The control of timing and events throughout the system allows for consistent and accurate measurement of the DUT.
About the Authors
Dave Derian is the Team Leader of Wireless and Video Test Applications at Teradyne. Before joining the staff, he was a Senior Engineer in telecom applications at LTX Corp. Mr. Derian received a B.S.E.E. degree from Worcester Polytechnic Institute.
John Moore is an Applications Specialist in the Teradyne Technology Applications Group. Previously, he was a Device Applications Engineer at the company and a Design Engineer at GE and Hughes. Mr. Moore earned B.S.E.E. and M.S.E.E. degrees from the University of Pennsylvania.
Teradyne, Inc., 321 Harrison Ave., Boston, MA 02118-2238, (617) 422-2567.
Copyright 1995 Nelson Publishing Inc.
June 1995
|
|