A $30 RTL-SDR dongle and a Raspberry Pi give you a wideband radio receiver that covers 500 kHz to 1.7 GHz. It hears FM broadcasts, air traffic, weather sensors, pagers, and satellites. The dongle digitizes raw radio and the Pi processes it in software. This guide covers the hardware, driver install, first reception, a Python IQ capture script, and two working projects: an ADS-B aircraft tracker and a 433 MHz sensor-to-MQTT bridge.
Quick Takeaways
- An RTL-SDR is a USB receiver built on the RTL2832U chip plus a tuner (R820T2 or R828D). It samples up to 2.4 MSPS stably.
- Any Raspberry Pi 3, 4, or 5 runs it. Use the RTL-SDR Blog V4 driver fork if you own a V4 dongle.
- Install
rtl-sdr, blacklist the dvb_usb_rtl28xxu kernel module, then verify withrtl_test. - Antenna quality and USB noise matter more than the Pi model.
Component Overview: How an RTL-SDR Works
The signal path has four stages:
- The antenna converts radio waves to a tiny AC voltage.
- The tuner (R820T2/R828D) mixes the chosen frequency down to a low intermediate frequency.
- The RTL2832U contains an 8-bit ADC that samples the signal and outputs IQ samples (in-phase and quadrature pairs).
- The Pi receives the IQ stream over USB 2.0 and demodulates it in software (FM, AM, SSB, ADS-B, and others).
The RTL2832U was designed for DVB-T television dongles. Hobbyists found that its test mode exposes raw IQ data. That discovery is why these dongles cost so little.
Which Dongle Should You Buy?
| Feature | Generic DVB-T Stick | RTL-SDR Blog V3 | RTL-SDR Blog V4 |
|---|---|---|---|
| Tuner | R820T2 | R820T2 | R828D |
| Frequency range | 24–1766 MHz | 500 kHz–1766 MHz | 500 kHz–1766 MHz |
| HF reception | No | Direct sampling mode | Built-in upconverter |
| Oscillator | Basic crystal | 1 PPM TCXO | 1 PPM TCXO |
| Bias tee | No | Yes | Yes |
| Broadcast FM notch | No | No | Yes |
| Approx. cost | $10–15 | $30 | $35 |
| Driver | Standard | Standard | Blog fork required |
Buy the V4 if you want HF and a cleaner FM band. Buy the V3 if you want broad software compatibility. Skip generic sticks for anything involving frequency accuracy, because their drift makes narrowband decoding unreliable.
Hardware Setup and Wiring
Use this layout:
Antenna → SMA/MCX cable → RTL-SDR → USB extension cable → Raspberry Pi
- Connect the dongle through a USB extension cable of 0.5 to 1 m. This moves it away from the Pi’s noisy USB 3.0 and HDMI circuitry.
- Add a ferrite choke on the USB cable if you see spurious spikes on the waterfall.
- The dongle draws about 270–300 mA at 5V. A Pi 4 handles it directly. On a Pi Zero or an overloaded Pi 3, use a powered USB hub.
- Do not hang a bias tee load (an LNA, say) off a Pi that is already near its USB current limit.
Antenna Length Reference
Dipole arm length is roughly 71.2 / frequency in MHz meters.
| Target | Frequency | Per-Arm Length |
|---|---|---|
| FM broadcast | 100 MHz | ~71 cm |
| Airband (voice) | 121–137 MHz | ~55 cm |
| NOAA weather satellites | 137 MHz | ~52 cm |
| ISM sensors | 433.92 MHz | ~16 cm |
| ADS-B | 1090 MHz | ~6.5 cm (quarter-wave) |
The telescopic dipole in the V3/V4 kits covers most of these by adjusting the element length.
Install the RTL-SDR Drivers on Raspberry Pi OS
Start from a fully updated Raspberry Pi OS (Bookworm or newer).
sudo apt update && sudo apt full-upgrade -y
sudo apt install -y rtl-sdr librtlsdr-dev
Blacklist the DVB-T Kernel Driver
Linux loads a TV driver that claims the dongle before the SDR tools can. Block it:
echo 'blacklist dvb_usb_rtl28xxu' | sudo tee /etc/modprobe.d/blacklist-rtl.conf
sudo reboot
RTL-SDR Blog V4 Driver Fork
The V4’s R828D tuner needs the Blog fork of librtlsdr. Remove the apt package first, then build:
sudo apt purge -y ^librtlsdr
sudo apt install -y git cmake build-essential libusb-1.0-0-dev
git clone https://github.com/rtlsdrblog/rtl-sdr-blog.git
cd rtl-sdr-blog && mkdir build && cd build
cmake .. -DINSTALL_UDEV_RULES=ON
make -j4
sudo make install
sudo cp ../rtl-sdr.rules /etc/udev/rules.d/
sudo ldconfig
sudo reboot
Verify the Install
rtl_test -t
A healthy result lists the tuner type (for example Found Rafael Micro R820T tuner) and the supported gain values. To check for dropped samples at full speed:
rtl_test -s 2400000
Run it for a minute. Repeated lost at least N bytes messages point to a weak USB cable or insufficient power.
First Reception: Listen to FM Radio
Replace 99.5M with a local station:
rtl_fm -f 99.5M -M wfm -s 200000 -r 48000 - | aplay -r 48000 -f S16_LE
| Flag | Function |
|---|---|
-f 99.5M |
Center frequency |
-M wfm |
Wideband FM demodulation |
-s 200000 |
200 kHz sample rate, matching FM channel width |
-r 48000 |
Audio output resample to 48 kHz |
-g 40 |
Optional manual gain in dB |
A headless Pi has no audio output by default over SSH. Stream the raw IQ data to your PC instead with rtl_tcp:
rtl_tcp -a 0.0.0.0 -p 1234
Then connect SDR++, GQRX, or SDR# on your desktop to the Pi’s IP address on port 1234. The Pi does the capture and your PC does the heavy DSP.
Run a GUI Receiver Directly on the Pi
sudo apt install -y gqrx-sdr
Set the device string to rtl=0, sample rate to 1.8 MSPS (the safest rate on a Pi), and mode to WFM (stereo).
Software Comparison
| Tool | Interface | Best For | Pi Load |
|---|---|---|---|
| rtl_fm | CLI | Quick demodulation, scripting | Low |
| GQRX | GUI | Waterfall and general listening | Medium |
| SDR++ | GUI | Lightweight modern UI | Medium |
| rtl_433 | CLI | 433/868/915 MHz sensors | Low |
| readsb | Service | ADS-B aircraft tracking | Low |
| GNU Radio | GUI/Python | Custom DSP flowgraphs | High |
Firmware/Code: Capture IQ Samples with Python
The pyrtlsdr library wraps librtlsdr. Raspberry Pi OS Bookworm blocks system-wide pip installs, so use a virtual environment:
sudo apt install -y python3-venv python3-numpy
python3 -m venv --system-site-packages ~/sdr-env
source ~/sdr-env/bin/activate
pip install pyrtlsdr
This script captures samples and finds the strongest frequency in the band:
from rtlsdr import RtlSdr
import numpy as np
sdr = RtlSdr()
sdr.sample_rate = 2.4e6 # 2.4 MSPS: max stable rate over USB 2.0
sdr.center_freq = 99.5e6 # Tune to 99.5 MHz
sdr.gain = 40 # Manual gain in dB; use 'auto' to let the tuner decide
# Read 256k complex samples (IQ pairs)
samples = sdr.read_samples(256 * 1024)
sdr.close()
# Apply a Hann window to reduce spectral leakage, then take the FFT
n = 4096
window = np.hanning(n)
spectrum = np.fft.fftshift(np.fft.fft(samples[:n] * window))
power_db = 20 * np.log10(np.abs(spectrum) + 1e-12)
# Map FFT bins to absolute frequency in MHz
freqs = np.fft.fftshift(np.fft.fftfreq(n, 1 / 2.4e6)) + 99.5e6
peak = freqs[np.argmax(power_db)] / 1e6
print(f"Strongest signal near {peak:.3f} MHz ({power_db.max():.1f} dB)")
The FFT turns time-domain IQ samples into a frequency-domain spectrum. Averaging several FFT frames before taking the peak gives a steadier result.
Scan a Whole Band with rtl_power
rtl_power -f 88M:108M:10k -g 40 -i 10 -e 1h fm_scan.csv
This sweeps 88–108 MHz in 10 kHz bins, integrates for 10 seconds per pass, and runs for one hour. Plot the CSV with heatmap.py or any spreadsheet tool to find active channels.
Gain, Sample Rate, and Frequency Correction
| Setting | Typical Value | Effect of Too High | Effect of Too Low |
|---|---|---|---|
| Gain | 30–45 dB | Overload, phantom signals | Weak signals vanish |
| Sample rate | 1.8–2.4 MSPS | Dropped samples on Pi | Narrow visible bandwidth |
| PPM correction | 0 on TCXO models | n/a | n/a |
Start with a manual gain of about 40 dB. Raise it until the noise floor lifts, then back off a few dB. Generic dongles without a TCXO drift by 20–80 PPM. Find the offset with rtl_test -p and pass it using -p in your tools.
Real-World Project 1: ADS-B Aircraft Tracker
Aircraft broadcast position, altitude, and speed on 1090 MHz using Mode S ADS-B. A Pi with a short antenna can track planes within 100+ km in open terrain.
Install the maintained readsb decoder with its web map:
sudo bash -c "$(wget -nv -O - https://github.com/wiedehopf/adsb-scripts/raw/master/readsb-install.sh)"
Then open http://<pi-ip>/tar1090 in a browser. For a bare-bones test without the full stack:
rtl_adsb -V
Build tips:
- Mount a 1090 MHz quarter-wave ground-plane antenna (a roughly 6.5 cm element) outdoors with a clear sky view.
- Use low-loss coax such as RG-6 or LMR-240. Every extra meter of cheap coax costs signal at 1 GHz.
- A 1090 MHz bandpass filter with an LNA at the antenna raises range sharply in noisy locations.
Real-World Project 2: 433 MHz Sensors to MQTT
Many weather stations, door sensors, and tire-pressure monitors transmit on 433.92 MHz with simple OOK/FSK modulation. The rtl_433 tool decodes hundreds of protocols without extra code.
sudo apt install -y rtl-433
rtl_433 -f 433.92M
Decoded packets show model, ID, temperature, and humidity. Publish them to a broker for Home Assistant or Node-RED:
rtl_433 -f 433.92M -F mqtt://localhost:1883
Run it as a systemd service so it starts on boot. Place the antenna near your sensors, since 433 MHz transmitters are low power and walls attenuate them quickly.
Troubleshooting Common Errors
| Symptom | Likely Cause | Fix |
|---|---|---|
usb_claim_interface error -6 |
DVB-T kernel driver loaded | Blacklist dvb_usb_rtl28xxu, reboot |
No supported devices found |
Missing udev rules or bad cable | Install udev rules, try another cable/port |
Failed to open rtlsdr device #0 |
Permissions | Add your user to the plugdev group, or run with sudo |
| Dropped samples | Weak USB power or high sample rate | Use a powered hub, drop to 1.8 MSPS |
| Spikes at fixed intervals | Pi USB/HDMI noise | Extension cable, ferrite choke, move antenna |
| V4 tunes wrong or fails | Stock librtlsdr installed |
Rebuild the Blog fork driver |
| Constant overload | Gain too high or strong FM station | Lower gain, add an FM notch filter |
Limits to Know Before You Start
The 8-bit ADC gives roughly 48 dB of dynamic range. A strong local FM or pager transmitter can desensitize the whole band, so use band-specific filters. The usable bandwidth is about 2.4 MHz at once, which rules out wideband captures like Wi-Fi. There is no transmit capability, so the dongle is receive-only. That also keeps it legal to experiment with in most countries, though local rules on decoding certain traffic (such as encrypted or cellular communications) vary. Check your national regulations.
FAQ
Can a Raspberry Pi run an RTL-SDR?
Yes. A Raspberry Pi 3, 4, or 5 runs an RTL-SDR at 1.8–2.4 MSPS for FM listening, ADS-B decoding, and rtl_433 sensor logging. A Pi Zero works for lightweight decoders, but not for GUI receivers like GQRX.
What can you receive with an RTL-SDR on a Raspberry Pi?
You can receive FM and AM broadcasts, airband voice, ADS-B aircraft, NOAA weather satellite images (137 MHz), 433/868/915 MHz ISM sensors, amateur radio, and pager traffic. The V3/V4 also reach HF bands such as shortwave through direct sampling or the V4’s built-in upconverter.
Why does my RTL-SDR not work on Raspberry Pi?
The usual cause is the Linux dvb_usb_rtl28xxu driver claiming the dongle. Blacklist it, reboot, and run rtl_test -t. For an RTL-SDR Blog V4, also install the Blog fork of the driver. Insufficient USB power is the other common cause.
Is the RTL-SDR V4 better than the V3?
The V4 adds a built-in HF upconverter, a broadcast FM notch filter, and improved filtering. The V3 uses the older R820T2 tuner and works with stock drivers. Choose the V4 for HF and strong-FM areas. Choose the V3 for the simplest software setup.



