Reducing Test Time with RF List Mode

Publish Date: Mar 08, 2010 | 9 Ratings | 4.89 out of 5 |  PDF

Overview

All phases of RF development are time sensitive, from design to validation and verification to production test. Because a large number of measurements are inevitably performed on the production line, RF engineers are constantly concerned with decreasing test time without compromising the needed test integrity to ensure compliance with stringent wireless specifications. In addition, the growing number of multiband wireless protocols added to devices each year makes this task even more challenging.

This paper addresses two factors that can slow down multiband measurements: instrument tuning time and overhead associated with changing RF configurations. However, with features such as list mode and configurable loop bandwidth, you can greatly reduce measurement time and deterministically advance through a preloaded list of RF configurations.

Table of Contents

  1. Benefits of RF List Mode
  2. Using RF List Mode
  3. Decreasing Tuning Time
  4. Perform Fast and Repeatable Tests with RF List Mode

1. Benefits of RF List Mode

In the most general sense, RF list mode provides an analyzer or generator the ability to perform RF configuration changes deterministically from a predefined test sequence. This is typically managed through the interactive or programmatic creation of a list of settings such as center frequency, power level, step duration, trigger signal, and others. Traditionally, instruments receive configuration updates through GPIB or other nondeterministic communication to the device. RF list mode provides you with the ability to preload these configuration changes prior to test execution, which results in fast and deterministic advancement from one RF configuration to another.

When testing RF components, such as power amplifiers or filters, for a cellular standard, such as GSM/EDGE, the measurements made during production test often include:

• Power added efficiency (PAE)
• Gain versus frequency at nominal power level
• Adjacent channel leakage ratio (ACLR) or adjacent channel power ratio (ACPR)
• Third-order intercept point (IP3)
• Modulation error ratio (MER) or error vector magnitude (EVM)

When testing individual RF components, you can perform these measurements in any order and conduct many tests in parallel. Increased tuning times become a vital part of making these tests as efficient as possible.

When testing a complete RF device, however, it will generally be placed into a test mode where a variety of tests are performed in sequence. Typically, RF devices are not designed with a trigger input or output to allow synchronization with your test system. Instead, the tests are cycled through with a set time allotted for each test. For this type of test setup, the determinism that results from using RF list mode is essential.

Back to Top

2. Using RF List Mode

When automating tests, using GPIB or Ethernet to communicate a change in RF configuration introduces considerable delay. You can use a PXI instrument to remove much of this overhead. The fast communication between the controller and the RF hardware allows for typical RF configuration changes of 2–4 ms. With the addition of RF list mode for the NI PXIe-5663E vector signal analyzer (VSA) and NI PXIe-5673E vector signal generator (VSG), you can dramatically reduce nominal tuning time between steps to less than 500 µs.

Increasing Test Speed

With RF list mode, you create a list of RF configurations before testing begins, allowing the instrument to progress rapidly through the list without the overhead associated with updating the RF configuration. Figure 1 shows a visual representation of this.


Figure 1. Performing deterministic RF configurations changes with RF list mode.


In a system without RF list mode, there are nondeterministic software calls to the hardware each time a new RF configuration is required. Additionally, the dwell time or measurement time is software timed, and is therefore subject to system jitter. With RF list mode, the RF configurations are downloaded to the hardware before testing and test steps advance deterministically by a user-specified trigger or step duration.

The new NI-RFSA and NI-RFSG drivers contain a list mode palette for NI LabVIEW. Figure 2 shows this palette as well as a portion of the block diagram used to perform a simple frequency sweep with a VSG. A list is created and steps for frequency are supplied programmatically. Later, in the block diagram, the list is sent to the VSG and the test sequence begins.

Figure 2. Use the Configuration List palette in LabVIEW to create a new list of RF frequencies.

In RF component testing, it is common to perform a frequency sweep at fixed power and measure the response through the component. To compare the speed improvements associated with RF list mode, you can perform such a test using a band-pass filter with a center frequency of 520 MHz.

Figure 3. This graph shows a measured 520 MHz band-pass filter response.

 

Table 1 summarizes the results of a frequency sweep performed with RF list mode and the NI PXIe-5663E and NI PXIe-5673E. Compare these results to the sweep performed with the same hardware operating without list mode. Both sweeps were performed from 400 MHz to 630 MHz at 0 dBm in 116 steps (2 MHz/step). A 1 MHz acquisition was made with the VSA with a 10 µs acquisition time.


Table 1. Measurement times significantly reduced with RF list mode enabled.


Even over the short period of this test, you see a large variance in the time each mode requires for reconfiguration of the VSA and VSG. When performing the test without list mode, the reconfiguration time is made up of nondeterministic software calls to the hardware to download the next configuration in the sequence. The data acquisition must be performed after each tune and is uploaded back to the controller prior to the list continuing execution. In contrast, with RF list mode enabled, the hardware advances through configurations until complete, not needing to wait for additional software tunes or for the data to be uploaded. Additionally, you can even upload and process data from completed list steps while list progression continues, allowing for parallel test system execution and increased throughput.

Although initiating a device using RF list mode can take longer due to all the configurations being downloaded up front, it need occur only once prior to tests beginning, after which, each filter would be characterized in a mere 56 ms compared to the 280 ms result without list mode (five times faster).

Making Deterministic Measurements

When performing tests on an RF device that uses fixed timing to cycle through test modes automatically, deterministic changes in instrument configuration are required for consistent and repeatable testing. Wireless devices commonly use this method for testing in lieu of designing an input or output trigger into the device to be solely used for production test. Due to the lack of software overhead, the needed level of determinism is made possible by RF list mode.

Figure 4 shows the results of using the NI PXIe-5673E VSG in standard operation to generate a single tone at 1 GHz and step through six power levels in 7 dB steps starting with -10 dBm and ending with -45 dBm with a 500 µs dwell time for each step. This was measured using the NI PXIe-5663E VSA.

Figure 4. This chart shows nondeterministic power steps in with 500 µs specified dwell time in standard operation.

You can see the nondeterministic dwell time when compared with Figure 5, which shows the same VSG with list mode stepping through these power levels deterministically advancing from one power level to the next every 500 µs.

Figure 5. Use RF list mode for deterministic 500 µs power steps.

Back to Top

3. Decreasing Tuning Time

The NI VSA and VSG are combinations of individual modular PXI instruments. Figure 6 shows a simplified diagram of the NI PXIe-5663E with the individual modular instruments highlighted.

Figure 6. This is a simplified diagram of the NI PXIe-5663E modular instrument.

A number of factors, including local oscillator tuning time, filter settling times, and even the digital communication speed to the instrument, limits the tuning time of a VSA or VSG. Generally, however, optimizing the local oscillator used for upconversion to or downconversion from RF will yield the highest returns to overall instrument tuning time. In the case of the NI PXIe-5663E, the NI PXIe-5652 acts as the local oscillator.

The NI PXIe-5652 uses a phase-locked loop (PLL) to maintain the frequency accuracy of the voltage-controlled oscillator (VCO) output. The most basic PLL consists of a phase detector, a low-pass filter, and a VCO connected to a frequency divider in a feedback loop with the phase detector as shown in Figure 7.

Figure 7. This is a simplified diagram of a PLL.

In the NI PXIe-5652, a phase detector is used to compare the generated signal to a reference oscillator and the difference is used to then adjust the generated signal. The frequency divider on the feedback path is used to divide down the generated signal to the same frequency as the reference signal. If, for example, the generated signal begins to fall behind in phase, an error signal (proportional to the difference) is output from the phase detector and the VCO frequency is increased in order to catch up.

The function of a PLL is two-fold: tune quickly to a specified frequency and once locked, monitor and maintain that lock. Unfortunately, these are conflicting functions that relate to the loop bandwidth or the bandwidth of the loop filter. The loop filter controls the bandwidth of phase (frequency) error that the phase detector will detect and adjust for. Narrow (low) loop bandwidth is ideal for generating a clean signal with low phase noise, as the most subtle changes in phase will be corrected for. Wide (high) loop bandwidth, however, makes it possible to track much wider swings in frequency, leading to faster frequency changes, but lowering the VCO phase noise performance.

The NI-RFSA and NI-RFSG drivers used to program the VSA and VSG give you the power to adjust this bandwidth to optimize the signal for a given test. By using a high PLL bandwidth, you decrease the tuning time at the expense of additional phase noise; if you require lower phase noise over faster tuning times for a particular measurement, specify a narrow PLL bandwidth for best performance.

Back to Top

4. Perform Fast and Repeatable Tests with RF List Mode

Today’s RF engineers face the challenge of performing fast, accurate, and repeatable tests for wireless components. Engineers often perform these measurements over multiple bands, and changing RF configurations between measurements can be costly to overall test time. With the addition of a reconfigurable loop bandwidth and RF list mode, you can increase measurement speed with fast and deterministic changes in RF configuration, which significantly reduces test time.

Back to Top

Bookmark & Share

Ratings

Rate this document

Answered Your Question?
Yes No

Submit