Using a Spectrum Analyzer

You now know what a spectrum analyzer is and what its controls do. This lesson covers the practical side: how to connect one safely to real signals, how to read the display confidently, how to measure signal levels accurately, and how to avoid the common errors that produce misleading results. Theory becomes useful only when you can apply it to real equipment without destroying either the instrument or your confidence.

The focus is on the measurements you will actually make as a ham radio operator: checking your transmitter's spectral purity, measuring the output of an oscillator, characterizing a signal from a local oscillator, and monitoring band activity. Each of these has specific setup requirements and common pitfalls.

Connecting Safely — Attenuators and Power Limits

The input stage of a spectrum analyzer — the first mixer, typically — is a delicate, expensive, and easily damaged component. Virtually all spectrum analyzers are designed to handle a maximum continuous input power of +20 to +30 dBm (100 mW to 1 watt). Exceeding this limit, even briefly, will destroy the input mixer and require expensive repair.

Amateur radio transmitters produce powers from milliwatts (QRP) to hundreds of watts. Before connecting any transmitter to a spectrum analyzer, you must calculate the power level at the analyzer input and verify it is within the instrument's limits.

Choosing an Attenuator

The attenuator must satisfy two requirements: it must reduce the signal to a safe level, and it must be rated to handle the transmitter's output power. Both conditions are independently necessary.

Requirement 1 — Signal level: For a 100 W (+50 dBm) transmitter and a +20 dBm maximum input, you need at least 30 dB of attenuation. A 30 dB or 40 dB attenuator is appropriate.

Requirement 2 — Power rating: A 30 dB attenuator dissipates most of the transmitter's power as heat. At 100 W, a 30 dB attenuator must dissipate 99.9% of the input power = about 99.9 watts continuously. You need a power attenuator rated for at least 100 watts, not a signal-level attenuator. Cheap SMA-connector attenuators rated at 1 or 2 watts will be destroyed instantly.

Directional Couplers as Attenuators

A better approach for high-power transmitter measurements is a directional coupler. This device samples a small fraction of the transmitted power — typically −20 dB or −30 dB — from a through-line without affecting the transmitter or dummy load. The transmitter drives into a 50 Ω dummy load (which must be rated for the full transmitter power), and the −20 or −30 dBm sample port feeds the spectrum analyzer. A directional coupler can handle the full transmitter power on the through-line since it does not absorb the transmitter's power — only samples it.

Reading the Display

A spectrum analyzer display looks intimidating at first but becomes intuitive quickly. The key elements are the frequency axis, the amplitude axis, the noise floor, and the signal peaks.

Identifying True Signals vs Noise

The noise floor appears as a fuzzy horizontal line near the bottom of the display when no signals are present. It fluctuates randomly because it represents actual thermal noise. True signals appear as peaks rising above this floor. A signal is real and reliably measurable if it rises at least 10 dB above the noise floor. If a signal is only 3–6 dB above the floor, it is marginally detectable but not accurately measurable.

A practical test: turn off the signal source and observe the display. Everything remaining is noise (or interference). Turn the signal back on and observe what peaks appear that were not there before. Those are your signals.

Impulse Noise and Non-Stationary Signals

Some interference sources produce brief bursts rather than continuous signals — power line interference, switching supplies, and digital devices often fall into this category. In peak detector mode, a single brief burst leaves a peak on the display that persists until the analyzer sweeps that frequency again. A slow sweep with peak detector makes brief events visible by capturing them; a fast sweep may miss the same event. If you suspect brief bursts, slow the sweep time and watch the display for "blobs" that appear and disappear.

Noise Floor in Practice

The noise floor determines the weakest signal you can see. In practice, the measurement noise floor is set by three factors: the analyzer's DANL, the RBW setting, and any external noise sources (interference, the signal source's own noise floor, etc.).

For a typical TinySA at 3 kHz RBW, the noise floor is around −115 dBm. For a harmonic measurement on a 20 W transmitter whose fundamental reads +43 dBm at the transmitter output (before the 30 dB attenuator), the fundamental at the analyzer input is +13 dBm. The harmonic is at most −50 dBc from the fundamental, placing it at +13 − 50 = −37 dBm at the analyzer — well above the −115 dBm noise floor. You have plenty of headroom. But if the transmitter were only 100 mW (+20 dBm) and the harmonic were 60 dBc below, the harmonic would be at −40 dBm before the attenuator, and −70 dBm at the analyzer input (after 30 dB attenuation) — still above the −115 dBm floor, so still measurable.

Dynamic Range and Its Limits

Dynamic range is the ratio between the strongest and weakest signals that can be measured accurately at the same time. It is limited at the top by the analyzer's input stage compression point, and at the bottom by the noise floor. For most analyzers, the spurious-free dynamic range (SFDR) is 60–80 dB at typical settings.

For harmonic measurements, you need enough dynamic range to see harmonics 43–60 dBc below the fundamental. If the analyzer has 70 dB of SFDR, this is comfortable. If you are trying to see a harmonic 80 dBc below the fundamental, you need an analyzer with higher dynamic range or must use different measurement techniques.

Overload and Compression

Input overload occurs when the signal at the analyzer's first stage exceeds its linear operating range. The result is not simply a wrong reading — the analyzer will generate its own harmonic and intermodulation products, which appear on the display as if they were real signals from the source. This is extremely dangerous for harmonic measurements: you might conclude your transmitter has a harmonic when the "harmonic" was actually created by the analyzer's own overloaded input.

To check for overload: insert additional attenuation (e.g. 10 dB) and observe the display. If the fundamental and all other peaks each drop by exactly 10 dB, the measurement is linear and valid. If some peaks drop by more or less than 10 dB, or if peaks appear or disappear, the system was overloaded at the original attenuation setting — use more external attenuation.

Instrument-Generated Intermodulation Products

When two or more strong signals are present simultaneously, the analyzer's own input stage can generate third-order intermodulation products — false signals at frequencies 2f₁−f₂ and 2f₂−f₁. These appear on the display at plausible frequencies and can be mistaken for real signals from the source.

The same test used to detect overload also detects instrument-generated IMD: insert 10 dB of additional attenuation. If a signal is real, it drops 10 dB. If a signal is a third-order IMD product generated by the analyzer, it drops 30 dB (because third-order intermod products scale as three times the input power change in dB). If a displayed peak drops more than 10 dB when you add 10 dB of input attenuation, it is not a real signal from the source — it was generated inside the analyzer.

Attenuator Reference Table

Correct setup for measuring a 20 W transmitter with a spectrum analyzer. The transmitter feeds into a 50 Ω dummy load through a 30 dB directional coupler (or through a 30 dB, 25 W attenuator as shown). The sample port delivers +13 dBm to the analyzer — well within the safe input range.

View Larger