Headless SDR Server

Building a dedicated remote SDR system for monitoring amateur radio.

Status: Operational
Started: September 2026

Project Overview
What you wanted the machine to accomplish and why you built a dedicated N100 Debian box rather than continuing to run SDRs from the desktop.

Hardware

  • Intel N100, 4-core/4-thread mini PC
  • Operating system: Debian, headless
  • Primary software: SDR++ Server
  • Two RTL2838UHIDIR / RTL-SDR receivers

Software

  • Debian
  • SDR++ Server
  • SSH
  • separate configurations for each receiver

Project Overview

The goal of this project was to take two existing RTL-SDR receivers and move them off the main desktop onto a dedicated, low-power computer that could remain online continuously.

An Intel N100 mini PC was repurposed for the job and installed with Debian without a desktop environment. Administration is performed remotely over SSH. The final design runs separate SDR++ Server instances for the two receivers, allowing each radio to be independently selected and used by SDR++ clients elsewhere on the network.

The two receivers were given permanent identities instead of relying on the identical/default RTL-SDR serial numbers that can cause devices to swap positions after rebooting. One receiver is dedicated primarily to 2-meter amateur-radio monitoring and is identified as 2METER. The second receiver, intended for more general-purpose listening and experimentation, is identified as DISCO.

Building the Server

Debian was installed as a headless system, with SSH being the primary management interface rather than installing and maintaining an unnecessary graphical desktop.

Once Debian was running, the first step was making sure the operating system could actually see both SDR receivers. lsusb showed both RTL2838UHIDIR devices along with the machine’s Bluetooth controller and USB hubs, confirming that this was not initially a USB-detection problem.

The RTL-SDR utilities were then used to validate the hardware. A long rtl_test run continued for more than an hour without developing recurring errors. Early in testing there was a “PLL not locked” message and a report of 84 lost bytes, but the receiver continued running normally afterward. That gave enough confidence that the USB path and SDR hardware were stable before spending time configuring SDR++.

Giving the SDRs permanent identities

With two otherwise similar RTL-SDRs connected, relying on USB enumeration order would have been asking for trouble. The EEPROM serials were therefore changed with the RTL-SDR utilities.

The receivers were named:

2METER — dedicated amateur-radio / 2-meter receiver
DISCO — general-purpose “discovery” receiver

That meant SDR++ could select a receiver by a meaningful permanent identifier instead of “device 0” and “device 1,” which could potentially change following a reboot or USB reconnection.

Separate SDR++ Instances

Rather than have both receivers fight over one SDR++ configuration, each was given its own working directory.

The two configuration roots are:

/home/xxxx/sdrpp-2meter
/home/xxxx/sdrpp-disco

Each directory has its own SDR++ configuration, including:

config.json
rtl_sdr_config.json

The respective RTL-SDR device name is stored in the configuration so that the 2METER instance always opens 2METER and the DISCO instance always opens DISCO.

The 2-meter SDR++ Server instance is listening on port 5259. We also configured a separate port for DISCO, although the exact second port number is not preserved in the notes I can retrieve now, so I would rather leave that blank on the public page than guess.

One interesting version oddity appeared during setup: the running SDR++ build reported v1.3.0, while Debian’s package information showed 1.2.1-1517. The working executable itself was the newer build, so this was treated as a package/version-label discrepancy rather than a functional problem.

Networking

The server was given stable addressing through OPNsense rather than relying on whatever DHCP address happened to be available.

Reservations ending in:

.141
.142

were established for its network interfaces.

That makes the SDR services predictable for clients and avoids having to rediscover the machine whenever DHCP leases change.

Wi-Fi was also investigated. The minimal Debian installation initially did not contain nmcli, so NetworkManager had to be installed. After that, the machine could display nearby access points, although one scan attempt produced a “not authorized” error. Since the primary objective was getting the SDR server operational rather than turning Wi-Fi troubleshooting into a separate project, that issue was not allowed to hold up the SDR configuration.

Problems and Fixes

Two identical RTL-SDR receivers

Problem: Two similar RTL-SDR devices could not safely be treated as “first dongle” and “second dongle.” USB enumeration can change, particularly after rebooting or unplugging equipment.

Fix: Each receiver’s EEPROM serial was changed to a meaningful unique identifier: 2METER and DISCO. SDR++ was then configured against those identities.

This is one of the changes I would consider essential to any permanent multi-RTL-SDR server.

Making one SDR++ installation behave like two independent radios

Problem: A single configuration would make it unnecessarily easy for changes to one radio to affect the other.

Fix: Separate working directories were created for each SDR++ Server instance. Each gets its own configuration files and server port, effectively making the N100 behave like two independent network receivers.

Initial RTL-SDR test warnings

Problem: Early receiver testing produced a PLL not locked message and a small lost-byte count.

Fix: Testing was allowed to continue rather than immediately treating the first warning as a hardware failure. The extended rtl_test run continued for more than an hour without accumulating continuing errors, indicating that the receiver and USB connection were stable.

That turned out to be a useful lesson in not diagnosing a radio based on one startup warning.

Debian did not have nmcli

Problem: The minimal headless Debian installation did not include NetworkManager’s command-line tools.

Fix: NetworkManager was installed, after which nearby Wi-Fi networks could be enumerated. A later authorization issue appeared during scanning, but Wi-Fi was not required for the core SDR server to function.

SDR++ configuration changes did not appear to take effect

This became one of the more interesting SDR++ lessons from the project.

Problem: Editing the primary config.json did not completely reproduce or lock down the desired receiver/VFO behavior.

Fix: SDR++ maintains radio-specific settings separately as well. The corresponding radio_config.json also had to be matched to the intended configuration. Once both sides of the configuration were aligned, the saved VFO/radio setup behaved as expected.

We subsequently discussed making the finalized configuration files read-only as a possible safeguard against accidental changes, but I would describe that on the site as an experiment/possible guardrail rather than claim it became the final locking mechanism.

The saved client configuration confirms that SDR++ maintains the overall layout/VFO information in config.json while modulation, bandwidth, squelch, filtering, and other individual-radio values reside in the radio configuration.

Current Server Configuration

The server currently operates as two independent SDR++ Server instances on the same N100 Debian machine.

FunctionCurrent configuration
HostIntel N100 mini PC
OSDebian, headless
ManagementSSH
Receiver 1RTL-SDR — 2METER
Receiver 2RTL-SDR — DISCO
2METER directory/home/matt/sdrpp-2meter
DISCO directory/home/matt/sdrpp-disco
2METER server port5259
DISCO server portSeparate port configured; number needs to be rechecked on server
Receiver selectionPermanent RTL-SDR EEPROM serial/name
SDR++ build observedv1.3.0
Debian package metadata observed1.2.1-1517
NetworkStable OPNsense DHCP reservations
Reserved addresses.141 / .142
Current stateOperational

SDR++ Client Configuration

The SDR++ client that connects to the server is currently set to use SDR++ Server as its source. The saved configuration has a 146.885 MHz center frequency, a dark theme, a 65,536-bin FFT, and the amateur band plan enabled.

The client configuration contains saved Hampton Roads-area repeater entries for:

FrequencySaved identification
145.410 MHzNewport News 4 / WC4R
146.610 MHzChesapeake Bowers Hill / W4CAR
146.790 MHzSuffolk / N4SD
146.820 MHzChesapeake / W4CAR
146.850 MHzPortsmouth / W4POX
146.895 MHzVirginia Beach / W4KXV
146.970 MHzVirginia Beach / W4KXV
147.000 MHzNewport News / WN4HRT
147.075 MHzNorfolk waterfront / KC2HTT
147.195 MHzSmithfield / WT4RA
147.300 MHzVirginia Beach 4 / WN4HRT
147.375 MHzNorfolk / W4VB

Those names and frequencies are present in the saved SDR++ configuration rather than reconstructed from memory.

The matched radio configuration also stores narrow-FM settings for most of these repeaters. The commonly used profile is approximately 9.95 kHz bandwidth, 75 µs de-emphasis, 2.5 kHz tuning steps, Voice IF noise reduction where enabled, and power-based squelch settings.

For example, the saved 146.820 W4CAR receiver profile uses 9,950 Hz bandwidth, 75 µs de-emphasis, Voice IF noise reduction, high-pass filtering, and a 2.5 kHz snap interval. The 146.850 W4POX profile is very similar, although its saved bandwidth is 9,925 Hz and some filtering differs.

What the Build Accomplished

The end result is less about having an RTL-SDR attached to a computer and more about making the receivers into network infrastructure.

The SDRs no longer need to occupy USB ports on the main desktop. The N100 can sit in the radio/server environment and run continuously, while another computer provides the SDR++ interface. Either physical radio can be targeted independently, their identities survive rebooting, and their configuration is kept separate.

It also gives the system a straightforward upgrade path. Replacing 2METER with an Airspy R2 later does not require rebuilding the entire architecture; it only changes the receiver behind that particular SDR service.

Future Plans

The existing multi-receiver SDR++ server design should make these upgrades relatively straightforward, allowing new receivers to be integrated without rebuilding the entire system.

Future plans include replacing one of the current RTL-SDR receivers with an Airspy R2, primarily for VHF/UHF monitoring. I also plan to add an Airspy HF+ Discovery to expand the server’s capabilities into HF reception.