Spectrum Analyzers
While an oscilloscope visualizes electrical signals in the time domain (amplitude versus time), a spectrum analyzer visualizes signals in the frequency domain (amplitude versus frequency). This fundamental shift in perspective makes spectrum analyzers indispensable for characterizing RF, microwave, and optical systems, identifying interference, analyzing signal modulation, and measuring phase noise.
Operating Principles
Spectrum analyzers primarily fall into two architectural categories, although modern high-performance instruments often combine both techniques into a hybrid approach.
- Swept-Tuned Analyzers: The traditional architecture, functioning essentially as a highly sophisticated superheterodyne receiver. The input signal is mixed with a local oscillator (LO) that sweeps across a defined frequency range. The resulting intermediate frequency (IF) is filtered by the Resolution Bandwidth (RBW) filter, detected, and displayed. This architecture is excellent for measuring extremely wide frequency spans with high dynamic range.
- Real-Time (FFT) Analyzers: These analyzers digitize the incoming signal and use Fast Fourier Transform (FFT) algorithms to compute the frequency spectrum mathematically. Unlike swept analyzers, which have "dead time" as the LO resets its sweep, real-time analyzers can capture transient or intermittent signals without missing data, providing an instantaneous snapshot of the spectrum.
Key Specifications & Parameters
The RBW determines the analyzer's ability to resolve closely spaced signals. It represents the width of the final IF filter. A narrower RBW provides higher frequency resolution and lowers the Displayed Average Noise Level (DANL), allowing weaker signals to be seen, but it significantly increases the sweep time.
The VBW filter is a low-pass filter applied after the envelope detector (before the display). It smooths the trace by averaging the noise. Unlike RBW, changing the VBW does not affect the frequency resolution or the actual signal-to-noise ratio at the detector; it only cleans up the visual representation.
The DANL is the intrinsic noise floor of the instrument itself, typically specified in dBm/Hz. It defines the absolute minimum signal level the analyzer can detect. It is highly dependent on the RBW setting and whether an internal preamplifier is active.
No local oscillator is perfectly stable; there are always rapid, short-term random fluctuations in its phase. This phase noise appears as a "skirt" spreading out from the base of the carrier signal. Excessive phase noise in the analyzer limits its ability to measure signals close to the carrier.
Calibration Considerations
Calibrating a spectrum analyzer is a complex process, typically requiring a synthesized sweep generator, a power meter with precision power sensors, and a step attenuator.
- Frequency Accuracy: Verifying the timebase (internal reference oscillator) and checking for sweep span linearity.
- Amplitude Accuracy (Absolute): Verifying the internal 50 MHz (or similar) reference signal level.
- Frequency Response (Flatness): Measuring the relative amplitude variation across the entire frequency range of the instrument. This is often the most time-consuming test.
- Scale Fidelity (Log Linearity): Verifying that the logarithmic amplifier correctly represents changes in signal amplitude (e.g., verifying that a 10 dB drop in input power registers exactly as a 10 dB drop on screen).
- Input Attenuator Switching Uncertainty: Verifying the accuracy of the internal RF step attenuator at different settings.