Your First Spectrum

Why this is worth doing before more reading

Every abstraction in the signals and propagation modules — bandwidth, modulation, adjacent-channel interference, the noise floor — is a picture. Until you have seen it, it is vocabulary. After twenty minutes with a receiver it is something you recognise.

An RTL-SDR costs about thirty dollars and receives from below the AM broadcast band to 1.7 GHz. It cannot transmit, which means nothing in this lab requires a licence.

Setting up

1. Plug in the dongle before starting the software. Most SDR programs enumerate devices at launch.
2. Choose your software. SDR++ is the current cross-platform default; SDRangel does more; GQRX suits Linux and macOS; SDR# is Windows-only and still common in tutorials.
3. Select the device and set the sample rate to about 2.4 MHz. That is the width of spectrum you will see at once.
4. Turn the gain up until the noise floor rises, then back off a little. Too little gain and weak signals vanish; too much and strong ones create spurious images across the display.

If the device does not appear on Windows, the driver is the usual culprit — the generic DVB driver claims it, and Zadig replaces that with the SDR driver.

The first signal to find

Tune to a local FM broadcast station. Not because it is interesting, but because you can verify it independently: you know its frequency, you know what it should sound like, and if it is not there something is wrong with the setup rather than with your understanding.

Set the demodulator to WFM (wide FM) with about 200 kHz of bandwidth. You should hear the station and see a broad, flat-topped signal on the display.

Once that works, the receiver is proven and everything else is exploration.

Reading the display

Two panels, showing the same data two ways.

The spectrum plot — frequency across, amplitude up. A snapshot of now.

• Horizontal position is frequency. The centre is where you tuned; the edges are half your sample rate either side.
• Vertical position is amplitude, almost always in dBFS — decibels relative to the converter’s full scale, so the numbers are negative and only differences are meaningful. A signal 30 dB above the noise floor is 30 dB above it regardless of what the absolute numbers read.
• The flat band along the bottom is the noise floor. Everything you can receive must poke above it, which is the entire game.

The waterfall — frequency across, time down. Each row is one spectrum snapshot, colour-coded by amplitude, scrolling downward.

The waterfall is the more useful of the two, because it shows behaviour. A carrier is a vertical line. A transmission that starts and stops is a dashed line. A signal drifting in frequency slopes. A CW transmission is Morse code written down the screen, legible by eye.

Four things to look at, in order

1. The FM broadcast band, 88–108 MHz. Set the sample rate high and look at the whole band. Note how evenly spaced the stations are, how wide each is (200 kHz), and how the strong local ones dwarf the distant ones.

2. The 2 metre band, 144–148 MHz. Mostly quiet, with narrow bursts when someone transmits. Compare the width of an FM repeater signal here with a broadcast signal — 15 kHz against 200 kHz, an obvious difference on screen. This is the bandwidth table made visible.

3. Your own GMRS handheld. Transmit a short test on a low-power channel with the SDR nearby and watch the trace appear. You are now seeing your own signal. Note how wide it is, and look either side for anything that should not be there.

4. The noise floor with the antenna disconnected. Everything vanishes except the receiver’s own noise. That baseline is what every signal you ever receive must exceed, and knowing what it looks like tells you immediately when something is wrong with an antenna.

What “gain” is doing

The gain control does not make weak signals better. It moves the whole picture up relative to the receiver’s own noise, and past a point it adds distortion that manufactures signals which are not there.

The test for too much gain: tune to a strong station and look for copies of it at regular intervals across the spectrum. Those are intermodulation products created inside your receiver. Back the gain off until they disappear.

This is the selectivity-versus-sensitivity trade from the practical circuits module, in front of you, with a slider.

Check yourself

1. You see a vertical line on the waterfall that never changes. What is it?
2. Your noise floor is at −70 dBFS and a signal peaks at −40. How strong is the signal relative to the noise?
3. You tune to a strong local station and see identical weaker copies every few hundred kilohertz. Real or not?