The recent proliferation of new RF applications has brought about a dramatic increase in spectrum usage, particularly in the unlicensed bands used by many of the new devices. Crowding virtually guarantees the presence of RF interference. This has led to adaptive techniques that can change power, data rate and modulation format in response to events in the signal environment.


Many of the new applications also require high data rates and high RF signal bandwidths as well as complex protocols for error correction, authentication, handovers and data security. Power, frequency, modulation, spectral occupancy and other parameters all change dynamically on ever decreasing time scales requiring designers and technical professionals in the RF field to characterize and measure RF signals as their systems respond to various events that trigger changes in the signal being measured. These emerging RF issues share one characteristic — they change dynamically in time, often in response to trigger events. The time axis along with these events and the system responses that occur can no longer be ignored.

Fig. 1 Typical frequency-domain display of an RF signal.

Consider the typical frequency-domain display, as shown in Figure 1. The image is a plot of signal amplitude (vertical axis) vs. frequency (horizontal axis). The graph makes no reference to events that precede or follow the instant of measurement. However, the system’s events to these transient events might be a critical parameter.

Where once it was enough to understand signals purely in terms of their frequency-domain behavior, it is growing ever more critical to evaluate a series of dynamic events over time. The time dimension is the new frontier in RF signal acquisition and analysis.

Acquisition and Analysis Challenges Evolve With Changing RF Signal Attributes

RF signals have evolved over the years, and acquisition methodologies have changed with them. For decades, most RF signals were continuous-wave analog phenomena, usually modulated using predictable AM, FM, or PM techniques. Bandwidth and dynamic range were the big challenges. Swept spectrum acquisition became the method of choice for frequency-domain analysis. It swept repeatedly across the frequency band of interest, recording the amplitude at various frequency points, and displaying the results of multiple sweeps in sequence. The quality of the acquisition relied on relatively stable, unchanging input signals.

Fig. 2 Sweeping in steps across a series of frequency segments showing important missed transients.

In today’s world of dynamic, modulated RF signals, the swept approach can miss critical details. As shown in Figure 2, the sweep steps across the frequency band in steps. The sweep is looking at frequency Fa while a momentary aberration is occurring at Fb. By the time the sweep arrives at frequency Fb, the error has vanished. Without some means of capturing the whole frequency band at once the glitch goes undetected.

Each of the three acquisition approaches has its strengths:

With the growth of cellular communications, digitally modulated RF signals became the new acquisition and analysis challenge. An alternative approach, vector signal analysis, emerged to handle digitally modulated signals. In effect, the signal is sampled in the time domain and processed to extract the modulation information. This method provides readings of key modulation parameters such as error vector magnitude (EVM).

A third methodology, real-time spectrum analysis, captures a time record of data for the passband of interest but takes the process a step further; it seamlessly captures data and stores a span of RF frequencies at once, then continues to capture these spans of frequencies, appending them into the memory, thus building a seamless history of the signal over time. Frequency-domain triggering complements this by providing the means to capture changes in the signal’s frequency-domain characteristics.

  • Currently swept acquisition offers the best dynamic range. It is a powerful, accurate method for the analysis of simple, predictable RF signals.
  • Vector signal analysis is optimized for modulation analysis, though with less capability on the RF side.
  • Real-time spectrum analysis triggers, captures and analyzes time-varying and transient RF signals by seamlessly capturing and storing a span of RF frequencies at once, enabling in-depth analysis with time-correlated multi-domain views (time, frequency, modulation and code).

Unique among these choices, the real-time spectrum analysis architecture supports practical bench top solutions that can acquire the all-important time dimension. Because it can characterize dynamic and/or unpredictable signal behavior, it is a pragmatic solution for developing forward-looking RF applications such as third and fourth generation (3G and 4G) technologies, wireless data networking and collision control, and other radars.

Meeting the Needs of Time-varying RF Signals — Real-time Spectrum Analysis

The key unique capabilities of real-time spectrum analysis are:

  • Triggering: Many signals are made up of unpredictable pulsed events. It is necessary to trigger on these transient phenomena and the only way to do this is by triggering on changes in both the frequency and amplitude characteristics.
  • Capturing (acquire): The only way to ensure that all time-varying and transient RF signals are captured is to continuously capture a seamless time record of a span of RF frequencies over a long period of time.
  • Analyzing: Complete analysis of an RF signal relies on the ability to understand the complex relationships between the different domains — time, frequency, phase, modulation, code and more. Only real-time spectrum analysis offers the power to show time-correlated multi-domain views, providing a unique and powerful insight into the signal. Because the time location (relative to the trigger) of every stored frame is known, it is possible to build time-correlated views in which the error vectors on a IQ display can be equated to specific points in the frequency and time-domain displays.

Triggering in the Frequency Domain

As shown in Figure 3, the real-time architecture makes it possible to define, detect and respond to trigger events in the frequency as well as the time domain. A frequency-domain trigger allows the user to set up both amplitude and frequency conditions for a capture.

Fig. 3 Conceptual block diagram of a real-time spectrum analyzer.

In the time domain world, the triggering capabilities of the oscilloscope are the key to meeting many of the toughest measurement challenges. Triggering allows the instrument to monitor an incoming signal indefinitely without consuming valuable memory space. Only when a specific condition occurs — a glitch or a logic state, perhaps — does the oscilloscope begin recording signal data.

Fig. 4 Frequency mask trigger set-up concept.

Real-time spectrum analysis frequency-domain triggering works much the same way. It always captures the signal of interest, thus ensuring that the signal can be completely analyzed. To understand frequency-domain triggering, consider the transient in Figure 4. The real-time acquisition tool is observing the signal continuously, as explained earlier. The frequency-domain trigger can be set to ignore the “normal” signal characteristics, ignore low level noise and trigger only when an anomaly exceeds a specific power level.

Capturing Signals and Events in Real-time

Upon encountering the trigger event, real-time spectrum analysis proceeds with its seamless, continuous signal capture unless instructed to do otherwise. The acquisitions accumulate potentially thousands of frames per second. These frames accumulate in a dedicated internal memory.

The most fundamental expression of frequency activity over time is the spectrogram. Using the stored data, a real-time acquisition tool can plot a cascading series of frames representing all of the “time” that elapsed while they were being captured. Figure 5, a simplified conceptual view, summarizes the relationship between stored frames and the spectrogram.

Fig. 5 The spectrogram cascading a series of full-span frames to produce a view that reveals spectral activity over time.

Looking downward from the viewer’s perspective, a succession of real-time bandwidth frames contains the underlying information for the spectrogram. The light blue area indicates the full width of the span (for example, 15 MHz). The trace color signifies amplitude; here, the red trace’s curvature indicates a shift in the frequency of the peaks. The green trace documents transients in some of the frames. These transients would very likely be missed if not for the seamless capture.

A real-world example illustrating this concept is shown in Figure 6. In this case, the designer has set up a frequency mask in the left window in order to trigger the analyzer if the oscillator under test does not meet the desired specifications for frequency undershoot or overshoot. The spectrogram display in the right window shows how the oscillator violates the mask as it turns on and provides the ability to measure the settling time and overshoot of the oscillator by showing how the frequency of oscillation changes over time.

Fig. 6 Using the frequency mask trigger and spectrogram to characterize the turn-on behavior of an RF oscillator.

Analyzing Results Across Multiple Domains

Moving to an example from the digital modulation world, the benefits of multi-domain analysis can be further explored. Once stored, the signal information is available to support analysis in the frequency, time, or modulation domains, all time-correlated. Points on the spectrogram have their equivalents on the frequency-domain view; EVM and constellation diagram vectors have their equivalents on the spectrogram, and so on. This allows the user to expand the analysis from one domain to another and improve the detailed view of the captured signals. Figure 7 is an analytical screen showing the time domain, frequency spectrum, error vector magnitude and constellation display for a 16QAM signal with a significant level of amplitude distortion.

Fig. 7 Example of correlated multi-domain analysis.

Saving Time with Real-time Spectrum Analysis

Like so many innovative testing methodologies that came before it, real-time spectrum analysis is all about saving time and effort. It can deliver relevant results quickly in applications that defeat other acquisition approaches.

For the designer working with advanced technologies, seamless real-time capture is a tremendous help in understanding complete transmit/receive sequences. Unlike before, the engineer can see both processes on one view — the spectrogram. From there, it is a simple matter to hone in on individual events in other analytical views displaying the frequency or time domains.

For those conducting surveillance, frequency-domain triggering aids the capture of unpredictable transmissions anywhere in the monitoring band. Rather than continuously acquiring a background signal and storing vast amounts of information into memory, the acquisition does not begin until the transmission occurs. Then its entire duration is stored for detailed examination in both the frequency and the modulation domains.

For troubleshooting emerging technologies, such as high speed downlink packet access (HSDPA), multi-domain analysis combined with seamless, uninterrupted acquisition can help designers observe a history of their signals’ modulation scheme changes over time, with detailed views of the critical transitions between schemes.

In each of these cases, the ability to find elusive signals and perform a thorough time-correlated analysis saves time while enhancing effectiveness. Designers can detect problems early. As many engineers have experienced, transient or intermittent phenomena can escape early design verification. Fixing problems after a device goes into production (or worse yet, customer delivery) is very costly, if indeed the problem can be rectified at all.

Troubleshooting and verification done with real-time spectrum acquisition eliminate the need for careful design and suitable performance margins, but it certainly increases confidence in the design’s performance.

Real-time Spectrum Analysis at Work

RFID Device Design Validation

Radio frequency identification (RFID) is gaining momentum in manufacturing, retail, banking (smart credit cards) and many more applications. RFID involves a “reader” device that transmits an RF pulse to transponders known as “tags” attached to merchandise items, shipping containers, personal ID badges, or even currency. The tag may be active or passive; both types transmit RF energy to the receiver, which records the data it receives.

Verifying a RFID device design poses a host of new challenges. Signal characteristics include:

  • Dynamic range with timing resolution: Some RFID devices transmit at a relatively high power level and receive the return pulses at a much lower level. It is essential to distinguish between the two.
  • Modulation within a burst: RFID burst signals carry modulation within the burst. Diverse RFID variants use differing modulation formats: On/off keying (OOK), pulse code modulation (PCM), amplitude shift keying (ASK), frequency shift keying (FSK) and more.
  • Frequency hopping: Some RFID readers send out signals that hop in an unlicensed band to avoid interference with signals from other devices.

It is clear that RFID validation calls for a solution that goes beyond the conventional static approach to frequency-domain acquisition to trigger on bursty, hopping signals, capture long transmit/receive sequences in their entirety, and analyze signal details in several domains.

Figure 8 is a real-time capture of an RFID transmission. The measurement instrument is triggered when the first burst of RF energy breaks the frequency mask shown in the top half of the display. The entire pulse is stored in memory, including the transmission of the RFID reader and the reflection of the RFID tag.

Fig. 8 Triggering on a transient RFID signal and storing it in memory.

The time-domain plot of power vs. time is the critical view for the first analysis step. The plot at the bottom of Figure 9 provides the needed resolution to see the basic signal characteristics at a glance. The initial part of the signal has very deep amplitude modulation, while the actual data is encoded in lower amplitude variations that occur later in the pulse. Note that a spectral view appears concurrently in the upper right portion of the screen. The center frequency is 13.56 MHz, which verifies that the RFID system is working within its specified nominal operating frequency.

Fig. 9 Multi-domain analysis of power vs. time, frequency spectrum and analog demodulation of the RFID signal.

Yet another cursor definition produces a detailed view of the first data packet. The location of the data was apparent from the AM demodulation plot, but the actual encoding is done with phase modulation. Figure 10 shows another plot where the lower display has been zoomed in on the first packet and set to perform the phase demodulation necessary to show the data that has been transmitted.

Fig. 10 Zooming-in on the phase encoded information.

Using these analysis tools, it is possible to examine the entire transaction between the reader and the tag in great detail and observe the changing frequency-domain characteristics of the RF signal. Figure 11 shows two different interactions between the pair of devices. The top half of the screen is the record of a “read” command, where the reader queries the tag for information and the tag’s response. The bottom half of the screen is the record of a “write” command, where the reader sends information for the tag to store and the tag sends an acknowledgment. At the start of each interaction, the transmission from the reader exhibits spectral splatter due to pulse modulation and the return signal from the tag exhibits discrete frequency sidebands resulting from FSK modulation. With the spectrogram display, it is easy to see the difference between these two interactions that would be difficult to see and impossible to differentiate using a swept spectrum analyzer.

Fig. 11 Spectrogram showing RFID “read” and “write” transactions between the reader and tag.

RFID requirements are not unique; there are many other applications that have similar needs. These include keyless entry systems, wireless LANs, automotive applications and consumer electronics items such as game controllers. In each case, proper validation and troubleshooting measurements will require triggering, capturing long sequences continuously and analyzing results in multiple domains.

HSDPA Research and Development

In an ever-noisier RF environment, it is necessary to adapt constantly. That is exactly what adaptive radio systems, particularly HSDPA, do to ensure quality transmission and reception in a world of competing RF signals. In an adaptive cellular phone system, the modulation format changes from one instant to the next, depending on link quality and the application in use. The system might use 64QAM when the phone is close to a base station; farther away, it may switch to 16QAM and eventually QPSK. Similarly, it may change to a more efficient modulation format when the user switches from voice transmission to data reception (downloading from a Web site, for example).

These fast modulation changes are not without complications. Most importantly, troublesome intermittent transients can accompany the transition from one format to the next. These can impact more than just the immediate user; they can impact other network elements attempting to interact with other phones.

Figure 12 shows the multi-domain analysis of a 16QAM signal with phase error caused by interference. The time-domain constellation and error vector magnitude are displayed.

Fig. 12 Multi-domain analysis of a 16QAM signal with phase error caused by interference.

Conclusion

RF signals are becoming the medium of choice for an increasing number of communication and control applications. A host of emerging RF-based applications brings with it the need to understand signal behavior over time. To do so, engineers are turning to real-time spectrum analysis to characterize the dynamic signals they encounter. This method makes it possible to trigger on frequency characteristics, to capture a long, seamless record of the signal over time, and to analyze signals in multiple domains based on the stored data. Real-time spectrum analysis improves the engineers’ view of signal characteristics that eludes other acquisition methodologies.

Marcus da Silva holds BS and MS degrees in electrical engineering from the University of Missouri–Rolla. He is currently a principal engineer with Tektronix’s Communications and Video Business Unit (CVBU) with more than 25 years of experience in the test and measurement and telecommunications fields. Before joining Tektronix, he was vice president of engineering and chief technical officer at Vivato where he assembled and managed the team that developed the industry’s first Wi-Fi switch, a device that dramatically extends the range of wireless LANs with the unique combination of phased-array antennas, Ethernet switching and WLAN technology. Previously, da Silva held various engineering, management and marketing positions at Hewlett-Packard and Agilent over a 23-year period. He was R&D section manager where he developed Agilent’s wireless communications test instruments, and served as a business development manager in HP’s Communications Test Systems group. He also had several engineering, marketing and manufacturing assignments where he made notable contributions in frequency synthesis, test methodologies, device modeling, microwave component design and metrology. In addition to his management and engineering work, da Silva was a key technical contributor for the Federal Communications Commission (FCC). He has also contributed to the IEEE 802 Wireless LAN standards and the TIA Cellular standards committees.