Real-time spectrum analysis basics

Chapter 1: Introduction and Overview

As RF signals become ubiquitous in the modern world, the problem of interference between the devices that generate them has grown. Products such as mobile phones operating in licensed spectrum must be designed not to send RF energy into adjacent frequency channels, which is particularly challenging for complex multi-standard devices that switch between different transmission modes and maintain synchronized links to different network elements. Simpler devices operating in unlicensed bands must also be designed to operate correctly in the presence of interfering signals, and government regulations often dictate that they are only allowed to transmit signals in short bursts at low power. These new digital RF technologies require the use of computers and RF, including wireless LANs, mobile phones, digital TV, RFID, and more. These technologies, combined with the latest technologies in software-defined radio (SDR) and cognitive radio (CR), provide a new path forward that will fundamentally change the way spectrum analysis is conducted and improve the efficiency of one of the scarcest commodities - RF spectrum. To overcome these evolving challenges, it is critical that today's engineers and scientists can reliably detect and characterize time-varying RF signals, which is not easily accomplished using traditional measurement tools. To address these issues, Tektronix developed the Real-Time Spectrum Analyzer (RSA), an instrument that can discover elusive effects in RF signals, trigger on those effects, seamlessly capture them into memory, and analyze them in the frequency, time, modulation, statistical and code domains.

This article explains how the RSA works and provides you with a basic understanding of how to use the RSA to solve many measurement problems associated with capturing and analyzing modern RF signals.

Modern RF Measurement Challenges

Given the challenges of characterizing the behavior of today's RF devices, it is imperative to understand how frequency, amplitude, and modulation parameters behave in the short and long term. In these cases, using traditional tools such as swept spectrum analyzers (SA) and vector signal analyzers (VSA) may provide an overview of the signal in the frequency and modulation domains, but they typically do not provide enough information for the user to confidently characterize the dynamic RF signals generated by the device.

Consider the following challenging measurement task:

• Discover rare short-term events

• See weaker signals masked by stronger signals

• Observing the signal masked by noise

• Find and analyze transient and dynamic signals

• Capture burst transmission, glitches, switching transient events

• Verify PLL settling time, frequency drift, micro-amplification, and capture spread spectrum and frequency hopping signals

• Monitor spectrum usage, detect rogue transmissions, and diagnose transient EMI effects

• Characterize modulation schemes over time to isolate software and hardware interactions

Each measurement involves time-varying RF signals that are often unpredictable. To effectively characterize these signals, engineers need a tool that can find elusive events, effectively trigger on them, isolate the events into memory, and analyze signal behavior in the frequency, time, modulation, statistical, and code domains.

Figure 1-1. A swept spectrum analyzer steps through a series of frequency bands, often missing important transient events that occur outside the currently scanned frequency band, such as the tan segment Fb highlighted on the right.

Simple inspection of the instrument structure

To understand how an RSA works and appreciate the value of the measurements it provides, it is necessary to first examine two other traditional types of RF signal analyzers: the swept spectrum analyzer (SA) and the vector signal analyzer (VSA).

Swept spectrum analyzer

The swept tuned superheterodyne is the traditional architecture that first enabled engineers to make frequency domain measurements decades ago. Originally built using purely analog components, swept spectrum analyzers have continued to evolve with the applications they serve. The current generation of swept spectrum analyzers includes a variety of digital elements such as ADCs, DSPs, and microprocessors. However, the basic swept approach remains largely unchanged and is best suited for observing controlled, static signals. Swept spectrum analyzers measure power versus frequency by downconverting the signal of interest and sweeping it through a passband of a resolution bandwidth (RBW) filter. The RBW filter is followed by a detector that calculates the amplitude of each frequency point within the selected bandwidth. Although this approach provides a high dynamic range, it has the disadvantage that it can only calculate amplitude data for one frequency point at a time. This approach is based on the assumption that the analyzer can complete multiple sweeps without the measured signal changing significantly. As a result, this approach requires the input signal to be relatively stable and unchanging. If the signal changes rapidly, the change may be statistically missed. As shown in Figure 1-1, the sweep is looking at frequency band Fa, and a transient spectral event occurs at Fb (left). By the time the sweep reaches band Fb, the event has disappeared and is not detected (right). Swept spectrum analyzer architectures do not provide a reliable way to detect the presence of such transient signals, so debugging many modern RF signals requires a very long time and a lot of work. In addition to missing transients, the spectrum of pulsed signals used in modern communications and radar can be incorrectly represented. Swept spectrum analyzer architectures cannot represent the spectrum occupied by pulses without repeated sweeps. Special attention must also be paid to sweep rate and resolution bandwidth.

Figure 1-2 a, b, c: Simplified block diagrams of a swept spectrum analyzer (a), a vector signal analyzer (b), and a real-time spectrum analyzer (c).

Figure 1-2 illustrates a typical modern swept spectrum analyzer architecture. Although modern swept spectrum analyzers have replaced the analog functions with digital signal processing (DSP), the basic architecture and limitations remain the same.

Vector Signal Analyzer

Analyzing the digital modulation of a transmitted signal requires vector measurements that provide both amplitude and phase information. Figure 1-2b is a simplified VSA block diagram.

The VSA digitizes all RF power within the transmit band and places the digitized waveform into memory. The waveform in memory contains both amplitude and phase information, which the DSP can use for demodulation, measurement, or display processing. Inside the VSA, the ADC digitizes the wideband IF signal, and downconversion, filtering, and detection are done digitally. The transformation from the time domain to the frequency domain is done using an FFT algorithm. The VSA measures modulation parameters such as FM deviation, code domain power, and error vector magnitude (EVM and constellation diagrams). It also provides other displays such as channel power, power versus time, and spectrum diagrams.

Although VSAs have added the ability to store waveforms in memory, their ability to analyze transient events is limited. In the typical VSA free-run mode, acquired signals must be stored in memory before they can be processed. The serial nature of this batch processing means that the instrument is blind to events that occur between acquisitions. It cannot reliably find single-shot or infrequent events. These events can be isolated in memory using the infrequent event trigger. Unfortunately, VSA triggering capabilities are limited. External triggering requires prior knowledge of the event in question, which may not be practical. IF level triggering requires a measurable change in total IF power and cannot isolate weak signals in the presence of large signals or signals that change in frequency but not in amplitude. Both of these situations often occur in today's dynamic RF environments.

Real-time spectrum analyzer

The term "real-time" comes from early work on digital simulations of physical systems. A digital system simulation is said to be working in real time if it works at the same speed as the actual system being simulated.

Analyzing a signal in real time means that the analysis operations must be performed fast enough to accurately process all signal components in the frequency band of interest. This definition states that we must:

• Sample the input signal fast enough to meet the Nyquist criterion. This means the sampling frequency must exceed twice the bandwidth of interest.

• Perform all calculations continuously and quickly enough so that the analysis output keeps up with changes in the input signal.

Discover, Trigger, Capture, Analyze

Real-time spectrum analyzers (RSAs) are designed to address the measurement challenges associated with instantaneous, dynamic RF signals described in the previous section. RSAs perform signal analysis using real-time digital signal processing (DSP), which is done before storage in memory, whereas VSA architectures typically use post-acquisition processing. Real-time processing allows the user to discover events that would not be visible to other architectures and trigger on those events, optionally capturing them in memory. Batch processing can then be used to fully analyze the data in memory in multiple domains. Signal conditioning, calibration, and certain types of analysis can also be performed using the real-time DSP engine.

Figure 1-3. VSA processing compared to real-time spectrum analyzer real-time engine processing.

The heart of the RSA is the real-time processing block, shown in Figure 1-2c (page 6). Like the VSA, it digitizes a wide capture bandwidth. Unlike the VSA, the real-time engine works fast enough to process every sample point without blanking, as shown in Figure 1-3. Amplitude and phase corrections are applied continuously to compensate for the analog IF and RF responses. Not only is the data stored in memory fully corrected, but all subsequent real-time processing is performed, operating on the corrected data. The real-time engine supports the following features to meet the needs of modern RF analysis: