Arapuca WSPR

A headless WSPR and FST4W receiver daemon for Linux. It was developed to run as an unattended receiver without a graphical interface on a Raspberry Pi. The program receives 48 kHz 16-bit audio samples via IP multicast, decodes weak signal spots using wsprd and jt9, and uploads the results to WSPRNet.

The name Arapuca comes from a traditional Brazilian bird trap — here repurposed as a trap for weak radio signals from around the world. No birds were harmed — we do not endorse capturing birds! Only signals.

Features

• Dual-mode decoding: WSPR and FST4W
• Receives audio via IP multicast (UDP)
• Time-aligned 2-minute recording windows
• Accepts 48 kHz or 12 kHz input sample rates
• Automatic 4:1 downsampling when using 48 kHz input
• Spot upload to WSPRNet with retry logic
• Sun elevation and DX distance logging
• Configurable log directory

Dependencies

• Qt 5 (Core, Network)
• wsprd — WSPR decoder (from WSJT-X)
• jt9 — FST4W decoder (from WSJT-X)

Both wsprd and jt9 must be installed and available in PATH.

Building

qmake
make

Usage

rxwspr -c <callsign> -l <locator> -f <freq> -g <group> -p <port> -i <iface> [options]

Required options

Option	Description
-c, --callsign	Station callsign
-l, --locator	Maidenhead grid locator
-f, --freq	RX frequency in Hz
-g, --group	Multicast group address
-p, --port	Multicast port (1025–65535)
-i, --iface	Multicast network interface name

Optional

Option	Description	Default
-s, --samplerate	Input sample rate: 48000 or 12000	12000
--deep	Enable wsprd deep search mode	Off
--no-upload	Disable spot upload to WSPRNet	Off
-d, --logdir	Log file directory	Home directory
-q, --quiet	Suppress stdout output	Off

Example

rxwspr -c PY2SDR -l GG66 -f 14095600 -g 239.0.0.11 -p 15004 -i dummy0

This starts the receiver on the 20m WSPR frequency, listening for multicast audio on group 239.0.0.11, port 15004, interface dummy0.

Log output

Each 2-minute cycle produces a separator line followed by any decoded spots:

260310 2004 -----------------------------------------  19.6  39.1 dB
260310 2004 -10  0.5     28.126007  `  TI4JWC EK70 30  50.7  5363 km  @  0.247
260310 2004 -26  0.2     28.126022  0  K3NUQ  FM09 23  31.2  7673 km  !
260310 2004 -13  0.7     28.126057  0  WW0WWV DN70 30  42.3  9104 km  !
260310 2004 -25  0.8     28.126065  1  0H5ADZ EA27 47                 !
260310 2004 -23  0.2     28.126094  0  KN6QZY CM87 23  47.6 10163 km  !
260310 2004 -24  0.4     28.126108  0  AC9YY  EN51 23  35.2  8298 km  !
260310 2004 -20  0.5     28.126120  0  KJ5SZ  EM32 23  44.1  7732 km  !
260310 2004 -22  0.4     28.126161  0  N5NBD  EM10 30  47.7  7823 km  !

The separator line shows the date, time (UTC), local sun elevation in degrees (19.6) and received audio power level (39.1 dB).

The spot lines contain the following fields: date, time (UTC), signal strength (dB), time offset (s), TX frequency (MHz), drift (Hz/min), TX callsign, TX grid locator, TX power (dBm), sun elevation at TX location (degrees), distance (km), and mode (! = WSPR, @ = FST4W). FST4W spots include an additional spectral spread field at the end. Balloon telemetry callsigns (starting with 0, 1, or Q) are shown without sun elevation and distance.

Log files are written to $HOME/rxwspr_<freq>.log by default.

How it works

1. Audio recording starts synchronized to the top minute at every 2 minutes
2. Audio samples arrive via multicast UDP and are accumulated for 115 seconds
3. If the input sample rate is 48 kHz, the samples are downsampled to 12 kHz; at 12 kHz they are used directly
4. The samples are written to a WAV file
5. wsprd and jt9 are launched in parallel to decode WSPR and FST4W spots
6. Decoded spots are parsed, logged, and queued for upload to WSPRNet
7. A 3-second upload timer sends spots to WSPRNet with up to 10 retries

Audio level

The received audio power level displayed in the separator line is computed as the mean power in decibels:

$$P_{dB} = 10 \cdot \log_{10} \left( \frac{1}{N} \sum_{i=1}^{N} s_i^2 \right)$$

where $s_i$ are the 16-bit PCM samples and $N$ is the total number of samples in the recording. This provides a quick indication of the input signal level for each 2-minute cycle.

License

Copyright (C) 2020 by Edson Pereira, PY2SDR

This is free software, released under the GNU General Public License v2.