Spectrum Analyzers: Introduction

An oscilloscope shows you what a signal looks like as a function of time — amplitude on the vertical axis, time on the horizontal axis. A spectrum analyzer shows you what a signal looks like as a function of frequency — amplitude on the vertical axis, frequency on the horizontal axis. These two views of the same signal are both valid and complementary, but they reveal completely different information. The oscilloscope shows you the shape of a waveform; the spectrum analyzer shows you all the frequency components hiding inside it.

A spectrum analyzer answers questions the oscilloscope simply cannot. Is this oscillator producing harmonics, and how strong are they? Is a switching power supply radiating interference, and at what frequencies? Is there a spurious oscillation somewhere unexpected in an amplifier? How does a band-pass filter actually perform, and how much attenuation does it give outside its passband? Is a wireless module transmitting on the right channel, and is it splattering into its neighbors? The spectrum analyzer shows signals in the frequency domain — amplitude against frequency — and turns all of these questions into a glance at the screen.

Time Domain vs Frequency Domain

To understand what a spectrum analyzer does, you first need to clearly grasp the difference between the time domain and the frequency domain. These are two different mathematical representations of the same physical signal, and each reveals different information.

In the time domain, a signal is described as a value that changes over time. An oscilloscope plots voltage on the vertical axis and time on the horizontal axis. If you connect a 7 MHz sine wave to an oscilloscope, you see a smooth sine wave cycling about once every 143 nanoseconds. If you also have a second harmonic at 14 MHz, you see the waveform become slightly distorted — the two components add together in time, producing a more complex shape.

In the frequency domain, a signal is described as a collection of discrete (or continuous) components, each with a specific frequency, amplitude, and phase. A spectrum analyzer plots amplitude on the vertical axis and frequency on the horizontal axis. That same 7 MHz signal with a 14 MHz harmonic now appears as two separate vertical spikes: a tall one at 7 MHz representing the fundamental, and a shorter one at 14 MHz representing the harmonic. The distance between the spikes and their height relative to each other are immediately obvious at a glance, whereas they would require careful waveform analysis to extract from an oscilloscope display.

This transformation between the time and frequency domains is described mathematically by the Fourier transform. Any periodic signal — no matter how complex — can be decomposed into a sum of sinusoids of different frequencies, amplitudes, and phases. The spectrum analyzer performs this decomposition (using either an analog swept filter or a digital FFT) and displays the result. What you see is essentially a bar chart of the frequencies present in your signal and how strong each one is.

The same signal containing a 7 MHz fundamental and 14 MHz second harmonic, shown on an oscilloscope (time domain, left) and a spectrum analyzer (frequency domain, right). The oscilloscope shows a distorted waveform; the spectrum analyzer immediately reveals two separate frequency components and their relative amplitudes.

View Larger

What a Spectrum Analyzer Shows

The vertical axis of a spectrum analyzer display is almost always calibrated in dBm — the power level in decibels relative to one milliwatt. The horizontal axis spans a frequency range set by the user. Each vertical "spike" or "peak" you see on the display represents a frequency component in the signal, and its height tells you how much power is present at that frequency.

A spectrum analyzer does not show phase information. It shows only amplitude as a function of frequency. This is an important distinction from a vector network analyzer (covered in the Test Equipment: Advanced module), which measures both amplitude and phase. For many measurements — measuring transmitter harmonics, checking oscillator spectral purity, characterizing filter stopband attenuation — amplitude information alone is exactly what you need.

The "floor" of the spectrum analyzer display is the noise floor — the minimum signal level the instrument can detect. Any signal below the noise floor is hidden in the noise and cannot be seen. The noise floor is typically expressed in dBm and depends on the instrument's internal noise figure and the resolution bandwidth setting. Understanding the noise floor is essential for making valid measurements — you cannot measure a signal weaker than the noise floor, no matter how carefully you set up the instrument.

Reading Signal Levels

The vertical axis is calibrated in dB per division. Common settings are 10 dB/division or 2 dB/division. At 10 dB/division, the entire vertical range might span from −10 dBm at the top to −110 dBm at the bottom, giving 100 dB of dynamic range on screen. Markers can be placed on peaks to read exact dBm values without estimating from the scale. Most spectrum analyzers allow you to add a second marker and read the delta (difference) between two peaks, which is extremely useful for measuring harmonic levels.

For example: if the fundamental of a 7 MHz RF amplifier measures +43 dBm (20 watts) and the second harmonic measures −17 dBm, the harmonic is 43 − (−17) = 60 dBc below the fundamental, where dBc means “dB below the carrier.” Radio regulations typically require harmonics and spurious emissions to be at least 40–60 dB below the fundamental for transmitting equipment, so this amplifier would comfortably pass.

Types of Spectrum Analyzers

There are three main architectures used in spectrum analyzers. Each has different strengths, limitations, and cost implications.

Swept-Tuned Superheterodyne Analyzer

The original and still the most common professional architecture. A voltage-controlled oscillator (the local oscillator) sweeps across the frequency range of interest. At each point in the sweep, the input signal is mixed with the local oscillator to produce an intermediate frequency (IF). A bandpass filter at the IF — called the resolution bandwidth (RBW) filter — allows only a narrow slice of the spectrum to pass through at each instant. A detector then measures the amplitude of this narrow slice, and the result is plotted on the display. By the time the sweep has crossed the entire span, the amplitude at each frequency has been sampled, and the display shows the full spectrum.

The key advantage of this architecture is extremely wide frequency range — professional swept analyzers can reach 40 GHz or higher. The limitation is that it measures only one frequency slice at a time. If a brief, intermittent signal appears while the sweep is not pointing at that frequency, the analyzer misses it entirely. This architecture also suffers from the fact that sweep speed and resolution bandwidth are linked: a narrow RBW requires a slow sweep to allow the filter time to respond.

FFT-Based Analyzer

An analog-to-digital converter samples the input signal at high speed, digitizing a block of time-domain data. A digital signal processor then computes the Fast Fourier Transform (FFT) of this data, converting it to the frequency domain. The result is a snapshot of the spectrum at that instant, computed digitally.

FFT analyzers have the advantage of processing all frequencies simultaneously — they cannot miss a brief signal as long as it occurs during the capture block. They are also extremely flexible in terms of resolution and speed. The limitation is analog bandwidth: the ADC can only digitize up to half the sample rate (the Nyquist limit). Modern high-speed ADCs have pushed this into the GHz range, but swept analyzers still hold the advantage at very high frequencies.

The NanoVNA-based TinySA and similar low-cost instruments use this architecture for their lower frequency ranges (below a few hundred MHz). This is why they can display spectrum in near-real-time on small hardware.

Real-Time Spectrum Analyzer (RTSA)

A real-time spectrum analyzer combines the FFT approach with continuous, gapless processing. It captures and processes time-domain samples fast enough to compute overlapping FFT blocks with no gaps — every moment in time is analyzed. This enables detection of intermittent signals and transient events that would be missed by either swept or standard FFT analyzers. RTSAs can display a persistence display showing how often each frequency/amplitude combination has been observed, making them invaluable for finding intermittent interference or identifying modulated signals.

RTSAs are high-end instruments typically costing tens of thousands of dollars. The affordable options most hobbyists use (TinySA and similar USB analyzers) are effectively FFT analyzers with varying degrees of gap-free performance depending on hardware speed.

Key Specifications

When evaluating or using a spectrum analyzer, these are the specifications that matter most.

Frequency Range

The minimum and maximum frequencies the instrument can display. The TinySA covers 100 kHz to 960 MHz in its basic mode, which covers broadcast, most RF circuit work and the 433 and 868/915 MHz bands. The TinySA Ultra extends this to 6 GHz, covering Wi-Fi and Bluetooth. For higher frequencies you need an instrument rated to at least that frequency or a downconverter.

Displayed Average Noise Level (DANL)

Also called the noise floor, DANL is the minimum signal level the analyzer can display, expressed in dBm. It is typically specified at a specific resolution bandwidth — narrowing the RBW lowers the DANL (improves sensitivity) because less noise passes through the filter. A typical DANL might be stated as −150 dBm/Hz, which means −150 dBm when measured in a 1 Hz bandwidth. At a 1 kHz RBW, the DANL rises to −150 + 10·log10(1000) = −120 dBm.

This relationship is important: every time you multiply the RBW by 10, the noise floor rises by 10 dB. Narrow RBW = better sensitivity but slower sweep. Wide RBW = faster sweep but higher noise floor.

Dynamic Range

The ratio between the largest and smallest signals the analyzer can accurately display simultaneously. A dynamic range of 70 dB means you can see a signal 70 dB weaker than the strongest signal on screen at the same time. Dynamic range is limited by internal noise at the low end and by compression or distortion in the input stage at the high end. For harmonic measurements on an oscillator or amplifier, you need enough dynamic range to see harmonics 43–60 dB below the fundamental — most good spectrum analyzers easily achieve this, but very cheap USB analyzers may struggle.

Resolution Bandwidth (RBW)

The width of the IF filter used to measure amplitude at each frequency. A 3 kHz RBW means the analyzer measures all energy within ±1.5 kHz of each displayed frequency point and adds it together. Narrow RBW gives better frequency resolution (can distinguish two closely spaced signals) and lower noise floor, but requires longer sweep times. Wide RBW gives faster sweeps but cannot separate closely spaced signals and has a higher noise floor.

A practical example: to distinguish two signals 2 kHz apart, you need an RBW of 1 kHz or less. To measure an amplifier's harmonic at the same time as its fundamental, you might use a 10 kHz or 30 kHz RBW for speed, since the harmonic is megahertz away from the fundamental anyway.

Video Bandwidth (VBW)

A low-pass filter applied to the detector output after the RBW filter, which smooths the display trace. Narrowing the VBW averages out noise fluctuations, making weak signals more visible by revealing them above the smoothed noise floor. Setting VBW = RBW/10 typically produces a smooth, averaged display with about 10 dB improvement in apparent noise. The tradeoff is increased sweep time because the smoothing filter needs time to settle at each frequency point.

Phase Noise

The short-term frequency instability of the analyzer's local oscillator. Phase noise appears on the display as a "skirt" of noise around strong signals. If the LO has poor phase noise, a strong signal at 14.000 MHz will show spectral energy spreading several kHz to either side, potentially masking a weaker signal nearby. Phase noise is specified in dBc/Hz at a given offset from the carrier. A specification of −100 dBc/Hz at 10 kHz offset means the phase noise 10 kHz away from a strong signal is 100 dB below the carrier power.

Input Impedance and Maximum Input Level

Almost all RF spectrum analyzers have a 50 Ω input impedance. The maximum safe input level is typically +20 dBm (100 mW) without a preamplifier, or +30 dBm (1 watt) with the input attenuator set correctly. Connecting a transmitter running 100 watts (50 dBm) directly to a spectrum analyzer will destroy the input stage instantly. You must always use a calibrated attenuator pad or a directional coupler to reduce the signal to a safe level before it reaches the analyzer input. This critical safety precaution is covered in detail in the Test Equipment: Advanced module.

Internal block diagram of a swept-tuned superheterodyne spectrum analyzer. The voltage-controlled local oscillator sweeps in frequency, downconverting successive slices of the input spectrum through the IF chain. The RBW filter determines frequency resolution; the VBW filter smooths the detector output before display.

View Larger

Practical Applications

Spectrum analyzers are used across a wide range of measurements.

Harmonic and Spurious Output of Transmitters and RF Modules

Radio regulations typically require harmonics and spurious emissions to be at least 40–60 dB below the fundamental, and anything sold must pass electromagnetic compatibility (EMC) tests. A spectrum analyzer is the only practical tool for checking this on the bench: connect the output through a suitable attenuator to the analyzer input, then measure the level of each harmonic relative to the fundamental.

Oscillator Spectral Purity

An oscillator or synthesizer that produces a “clean” output on an oscilloscope may still have measurable harmonics, phase noise sidebands or spurious products that degrade a receiver, a test signal or an ADC clock. The spectrum analyzer reveals these hidden impurities, and checking the spectrum of an oscillator should be a standard step before committing a design.

Filter Characterization

By injecting a swept signal from a tracking generator through a filter into the analyzer's input, you can measure the filter's insertion loss and stopband attenuation across frequency. Many spectrum analyzers include a built-in tracking generator, and the TinySA's built-in signal generator can do the same job.

Finding Interference Sources

A portable spectrum analyzer with a small antenna or a near-field probe is invaluable for hunting interference. Moving around a room, or around a circuit board, while watching the display reveals which frequencies carry interference energy and — from how the level changes with position — where it is coming from.

Pre-Compliance EMC Testing

Before sending a product for formal EMC testing, designers use a spectrum analyzer with a line impedance stabilization network (LISN) and near-field probes to find and fix conducted and radiated emissions. This is covered in the EMC Pre-Compliance Testing lesson.

Affordable Options

Until the mid-2010s, even used spectrum analyzers cost thousands of dollars, placing them out of reach for most hobbyists and small companies. Today, several affordable options exist.

The TinySA ($65–$120) covers 100 kHz to 960 MHz (or 100 kHz to 6 GHz for the Ultra version). It has a built-in signal generator, making it a two-in-one instrument for basic measurements. The noise floor is around −95 dBm at a 3 kHz RBW, adequate for harmonic measurements if you use proper attenuators.

Handheld analyzers such as the TinySA (around $100) cover roughly 100 kHz to several GHz and are ideal for hobby harmonic and interference checks. The noise floor and dynamic range are inferior to dedicated analyzers, but they are sufficient for band monitoring, signal identification, and preliminary harmonic checks. The TinySA is a better choice for precision measurements.

Used HP/Agilent/Keysight swept analyzers — the HP 8560A, HP 8590L, or Agilent E4402B — can be found on eBay for $200–$800 and are professional instruments offering excellent performance. They require patience to set up and may need minor service, but they are full-featured instruments capable of measurements from 9 kHz to 3 or 6.5 GHz with dynamic ranges exceeding 90 dB.

For this module's practical exercises, the TinySA is recommended as it represents the right balance of cost, portability and capability for bench work.

dBm to Watts Converter

Spectrum analyzers display signal levels in dBm. It is useful to be able to convert quickly between dBm and watts — for example, when checking whether a measured harmonic meets a limit expressed in absolute terms, or when choosing an attenuator before connecting a power amplifier.

The conversion formulas are: P(W) = 10^(dBm/10) / 1000, and dBm = 10 × log₁₀(P(W) × 1000).