Spectrum Analyzer

What is a Spectrum Analyzer

A spectrum analyzer is a measuring instrument that displays the power of an input signal as a function of frequency. It answers a different question from an oscilloscope: not "what is this signal doing over time," but "what frequencies are present, and how strong is each one." The horizontal axis is frequency (set by a center frequency and a span, or by start and stop frequencies); the vertical axis is amplitude, almost always logarithmic in dBm so that a window of more than 100 dB can be shown on one screen. The display itself is the deliverable: a stable, calibrated trace from which a user reads carrier power, harmonic levels, occupied bandwidth, spurious emissions, and noise floor.

Functionally, every analyzer performs the same chain: it takes a wideband electrical input, selects a slice of the spectrum, measures the energy in that slice, and plots it. How that slice is selected and measured is what separates the architectures discussed in Chapter 2. A modern instrument also adds an RF front end (attenuator, preamplifier, preselector), a reference oscillator that fixes frequency accuracy, a digitizer, and a signal-processing engine that runs detectors, averaging, and standard-specific measurement personalities.

The history of the instrument tracks the rise of radio. Early frequency-domain measurements in the 1930s and 1940s used tuned receivers and panoramic adapters. The swept superheterodyne analyzer, which downconverts the band of interest and slides it past a fixed resolution-bandwidth filter, became the dominant architecture from the 1960s onward and remains the textbook model. Digitization of the intermediate frequency in the 1990s allowed fast Fourier transform (FFT) processing of the captured block, improving filter shape and settling time. In the 2000s, continuous, gap-free processing produced the real-time spectrum analyzer, able to catch transients that a swept instrument slides past between sweeps.

The category also splits by signal type. A traditional spectrum analyzer measures scalar power versus frequency. A vector signal analyzer (VSA) digitizes the signal as complex in-phase and quadrature (I/Q) samples, preserving phase, so it can demodulate digitally modulated carriers and report metrics such as error vector magnitude. Because most bench instruments now ship with both capabilities, vendors market them as signal-and-spectrum analyzers, with the VSA features unlocked by software options.

In application scale, the category spans an enormous range. Entry-level bench units start near 9 kHz and reach a few GHz; performance instruments reach 50 GHz at the input connector and extend to 110 GHz and beyond with external smart mixers for millimeter-wave work. Instantaneous analysis bandwidth ranges from a few MHz on value instruments to several GHz on the highest-end signal analyzers used for wideband radar and satellite payloads. No single instrument is optimal across this whole space; selection is the act of matching one instrument's frequency reach, bandwidth, sensitivity, and modulation tools to a specific test.

Architectures and Types

Spectrum analyzers are grouped by how they select and measure the spectrum. There are three measurement architectures, swept-tuned superheterodyne, hybrid FFT, and real-time, plus several practical form factors (benchtop, handheld and field, and USB or PXI modular). The architecture decides speed and what kind of signal you can reliably catch; the form factor decides where you can use it. The table below compares the three core architectures on the metrics that matter for selection.

Swept superheterodyne is the classic architecture. The input is downconverted by a tunable local oscillator, then swept through the passband of a fixed resolution-bandwidth (RBW) filter; a detector calculates the amplitude at each frequency point in the span. It excels at wide spans and stable signals and is the most economical way to see a large frequency window. Its defining weakness is that only the frequency currently under the sweep is being measured, so a short pulse or a frequency hop that occurs while the sweep is elsewhere is simply not seen.

Hybrid FFT analyzers share the swept front end but digitize the intermediate frequency and apply a fast Fourier transform to the captured block. This yields sharper, more repeatable filter shapes and faster settling, so narrow-RBW measurements that would take a swept analyzer a long time finish far more quickly. The instrument may stitch several FFT blocks together to cover a wider span. The constraint is that each block can only be as wide as the digitizer's instantaneous bandwidth.

Real-time spectrum analyzers (RTSA) digitize a wide instantaneous bandwidth continuously and compute overlapping FFTs with no gaps, so every sample contributes to a transform. This gives a guaranteed minimum signal duration that is captured with 100 percent probability of intercept, plus frequency-mask triggering and density or persistence displays that reveal weak signals hidden under stronger ones. RTSAs are the right tool for intermittent interference, frequency-hopping protocols, and spectrum monitoring. Their real-time span is limited to the acquisition bandwidth, commonly 40 MHz on mid-range instruments and hundreds of MHz to several GHz at the high end.

By form factor, benchtop instruments offer the best performance and largest displays for the lab and production floor. Handheld and field analyzers (such as Anritsu Spectrum Master and Field Master families) trade some performance for battery operation, ruggedization, and built-in cable, antenna, and interference-hunting tools used in installation and maintenance. USB and PXI modular analyzers move the front end into a small box or card driven by a PC, suiting automated test systems, remote monitoring, and cost-sensitive real-time work.

Detectors, RBW, and Sweep Time

Two settings dominate how a measurement looks: the resolution bandwidth (RBW) and the detector. Get either wrong and a perfectly calibrated instrument will report a misleading number. This chapter explains both, plus the speed penalty that RBW imposes on a swept analyzer.

Resolution bandwidth is the width of the IF filter that the instrument sweeps across the span. It does two things at once. First, it sets frequency resolution: two tones closer together than the RBW merge into one bump, so to separate them you must narrow the RBW below their spacing. Second, it sets the displayed noise floor: thermal noise is broadband, so a narrower filter admits less of it. Reducing RBW by a factor of 10 lowers the noise floor by roughly 10 dB, which is the cheapest way to dig a weak tone out of the noise. Video bandwidth (VBW) is a separate low-pass filter on the detected signal that smooths the trace; setting VBW well below RBW averages noise to make a small signal easier to read but does not change the noise floor itself.

The cost of a narrow RBW on a swept analyzer is time. Sweep time rises approximately as the span divided by the square of the RBW, so narrowing the RBW from 10 kHz to 1 kHz over the same span lengthens the sweep by about 100 times. This is exactly why FFT and real-time architectures matter: they process the captured IF block digitally and largely break the RBW-versus-time trade-off, finishing narrow-RBW measurements far faster than a swept sweep can.

Detectors decide how the many samples that fall into one horizontal display bucket are reduced to the single value plotted for that bucket. Choosing the wrong detector is a frequent source of error: a peak detector overstates noise, while a sample detector understates a CW tone's peak. The table below summarizes the standard detectors and where each is correct.

For EMC and EMI work the detectors are not a free choice: CISPR 16-1-1 defines peak, quasi-peak, and average detectors with prescribed time constants and IF bandwidths. The standard EMI IF bandwidths are 200 Hz, 9 kHz, 120 kHz, and 1 MHz across the regulated bands. The quasi-peak detector weights an emission by its repetition rate using a defined charge and discharge time constant: in the 0.15 to 30 MHz conducted band it uses a 1 ms charge time, a 160 ms discharge time, and a 9 kHz IF bandwidth, so a continuous carrier reads at its full level while sparse, intermittent interference is weighted down. Because quasi-peak measurements are slow, the practical workflow is a fast positive-peak pre-scan over the whole band, then quasi-peak only on the frequencies that approach the limit line.

Standards and Application Domains

The right analyzer is the one that holds the right certifications and personalities for your domain. Different industries lean on different standards bodies, and the instrument's measurement applications must match. A spectrum analyzer used purely for R&D needs raw performance; one used for compliance must be traceable and run standard-defined measurements that a lab or regulator will accept.

EMC and EMI compliance is governed by CISPR (the International Special Committee on Radio Interference, under the IEC). CISPR 16-1-1 specifies the EMI receiver itself, including the detector time constants and IF bandwidths described in Chapter 3, while product-family standards such as CISPR 11 (industrial, scientific, and medical equipment), CISPR 22 / CISPR 32 (multimedia equipment), and the FCC Part 15 rules in the United States set the emission limits. A full-compliance EMI receiver is a regulated subset of the spectrum-analyzer family; a general spectrum analyzer with the right detectors is acceptable for pre-compliance scans that de-risk a design before a formal lab visit.

Wireless and cellular conformance relies on standards from 3GPP (LTE, 5G NR), IEEE (802.11 Wi-Fi, 802.15 Bluetooth and Zigbee), and Bluetooth SIG. Here the spectrum analyzer must run vector signal analysis personalities that measure error vector magnitude (EVM), adjacent channel power ratio (ACPR or ACLR), occupied bandwidth, and spectral mask compliance against the relevant release of the standard. The analysis bandwidth must exceed the channel bandwidth: a single 5G NR carrier can occupy 100 MHz or more, and carrier aggregation pushes the requirement higher still.

Frequency accuracy traces back to the instrument's reference oscillator and ultimately to national standards. Bench analyzers ship with a temperature-compensated crystal oscillator (TCXO) or an oven-controlled crystal oscillator (OCXO); the latter holds aging to a few parts in 10^7 per year and can be disciplined by an external 10 MHz reference or GPS for laboratory traceability. The table below maps common domains to their governing standards and the analyzer capability each one demands.

The practical takeaway: write down which standards your test report must cite before you shortlist instruments. A bench analyzer that lacks the certified EMI detectors cannot produce a CISPR-defensible report, and a unit without the 5G NR option cannot pass-fail a modulated carrier no matter how good its noise floor is. Standards drive the option list, and options drive the price more than the base hardware does.

Key Specification Parameters

A spectrum-analyzer datasheet lists dozens of numbers, but a handful decide whether the instrument can do your job. The most important are frequency range, displayed average noise level (DANL), phase noise, dynamic range (set by TOI and the noise floor), amplitude accuracy, resolution bandwidth range, and instantaneous or real-time bandwidth. Each is explained below, and the comparison table that follows shows representative figures from current manufacturer literature.

Frequency range is the span from the lowest to the highest frequency the instrument can tune. Entry instruments start near 9 kHz and reach a few GHz; performance instruments reach 50 GHz at the connector and extend to 110 GHz with external smart mixers. Buy enough range to cover your fundamental plus the harmonics you must verify, typically the second and third harmonic, so a 6 GHz fundamental measurement really wants 18 GHz of reach.

DANL (displayed average noise level) is the analyzer's own noise floor with the input terminated in a matched load, normalized to a 1 Hz RBW and expressed in dBm/Hz at 0 dB attenuation. It sets sensitivity: a signal must rise above DANL to be measured. Modern bench analyzers reach roughly -150 to -165 dBm/Hz with the internal preamplifier engaged. Lower (more negative) is better, but it always trades against dynamic range.

Phase noise describes the short-term frequency instability of the local oscillator, expressed in dBc/Hz at a stated offset from the carrier. It limits your ability to see small signals close to a strong one, for example a low-level spur near a clean carrier. Typical performance is on the order of -100 to -120 dBc/Hz at 10 kHz offset and -120 to -150 dBc/Hz at 1 MHz offset, improving on high-end instruments.

Dynamic range is the window between the largest and smallest signals measurable at the same time. The bottom is set by DANL and phase noise; the top by distortion the instrument generates when overdriven, characterized by the third-order intercept (TOI), typically +10 to +25 dBm. Spurious-free dynamic range for third-order products is approximately two-thirds of the gap between TOI and the noise floor. There is an optimum mixer level, often -20 to -30 dBm, where the noise and distortion contributions cross and dynamic range is maximized; the input attenuator is the knob used to find it.

Other decisive specs round out the picture:

• Amplitude accuracy: the absolute level uncertainty, commonly a fraction of a dB across most of the range, the figure your power measurements inherit.
• Resolution bandwidth range: how narrow and how wide the RBW filters go, for example 1 Hz to 10 MHz, plus the certified CISPR EMI bandwidths if you do EMC work.
• Instantaneous / real-time bandwidth: the widest band processed in one capture, which caps modulation analysis and real-time span; 40 MHz is common mid-range, with hundreds of MHz to several GHz at the high end.
• Frequency reference: TCXO or OCXO, with an external 10 MHz or GPS disciplining input for traceability.
• Inputs and protection: connector type (N or 3.5 mm), maximum safe input power, internal preamplifier, and preselector for harmonic and image rejection.

The table below compares representative bench and real-time instruments from current manufacturer literature. Use it to calibrate expectations, not as a substitute for the exact datasheet of the specific model and option configuration you intend to buy.

Selection Decision Factors

To turn the preceding chapters into a specific purchase, work through the decision sequence below. Most selection mistakes are not a single wrong number but a decision made at the wrong level, for example choosing on price before pinning the frequency reach and bandwidth the application actually needs. These eight steps double as a fixed RFQ template.

1. Frequency range: Cover your highest fundamental plus the harmonics and spurs you must verify, usually to the second or third harmonic. For millimeter-wave work, confirm whether external mixers extend the base instrument.
2. Architecture: Swept or FFT for stationary signals and general bench work; real-time (RTSA) if you must catch transients, hopping, or intermittent interference with 100 percent probability of intercept.
3. Instantaneous / analysis bandwidth: Must exceed the widest modulated channel you will analyze (a 5G NR carrier can be 100 MHz or more) and set the real-time span you can monitor at once.
4. Sensitivity and dynamic range: Read DANL and phase noise at your band of interest, and check TOI for distortion headroom. Confirm an internal preamplifier and preselector if you measure weak signals near strong ones.
5. Measurement personalities and standards: Verify the certified detectors for CISPR EMI, and the VSA options (EVM, ACPR) for the exact 3GPP or IEEE release your report must cite. Options often cost more than the base hardware.
6. Form factor and environment: Benchtop for lab and production; handheld and battery powered for field installation and interference hunting; USB or PXI modular for automated test racks and remote monitoring.
7. Reference and traceability: TCXO is adequate for general use; OCXO with an external 10 MHz or GPS disciplining input is needed for low aging and calibrated, traceable measurements.
8. Connectivity and automation: SCPI over LAN (LXI), USB, or GPIB for programmatic control; touchscreen and multi-window display for bench ergonomics; data export formats your workflow already uses.

One last commonly overlooked dimension is serviceability and calibration logistics. A spectrum analyzer must be recalibrated periodically (typically annually) to keep its amplitude and frequency accuracy traceable, and downtime during calibration costs production time. Check the maker's local calibration-lab coverage, loaner availability, firmware update cadence, and option upgradability so that a frequency or bandwidth limit added later does not force a new purchase. Keysight, Rohde & Schwarz, Tektronix, and Anritsu all operate calibration and service networks across major markets, while value vendors such as Siglent and Rigol compete on price and increasingly on local support, which makes them sensible for education, manufacturing test, and pre-compliance where the last decibel of performance is not required.