KiwiSDR SNR Monitoring Utilities

Based on work by Olaf - LA3RK

Install Dependencies

This gets it going on a Raspbian system...

• sudo apt-get install python3-rrdtool rrdtool python3-numpy python3-matplotlib
• sudo pip3 install suntime

Gathering data

$ python3 snrtorrd.py -s your.kiwisdr.hostname -p 8073

The default frequency span is 0 - 30 MHz. The number of FFT bins is fixed at 1024.

Optional arguments include:

Options:
  -h, --help            show this help message and exit
  -s SERVER, --server=SERVER
                        server name
  -p PORT, --port=PORT  port number
  -a PASSWORD, --password=PASSWORD
                        server password
  -l LENGTH, --length=LENGTH
                        how many samples to draw from the server (Default: 100)
  -t STEP, --timestep=STEP
                        Expected timestep between samples
  --timeout=TIMEOUT     Connection Timeout
  -z ZOOM, --zoom=ZOOM  zoom factor
  -o START, --offset=START
                        start frequency in kHz
  -v VERBOSITY, --verbose=VERBOSITY
                        whether to print progress and debug info
  --spectra=SPECTRA     Spectra Output File (csv). 

For example to login to a private server, and save out raw spectra data, you would use:

$ python3 snrtorrd.py -s your.kiwisdr.hostname -p 8073 -a yourpassword --spectra myserver_spectra.csv

RRD Plotting

$ python3 rrdtograph.py -s your.kiwisdr.hostname --title "My KiwiSDR SNR"

Note that the hostname must match the hostname used above, as this script uses filenames based on the hostname. This script will generate plots with filenames of the form:

your.kiwisdr.hostname_0_30000-Daily.png
your.kiwisdr.hostname_0_30000-Weekly.png
your.kiwisdr.hostname_0_30000-Monthly.png

Spectrograph Plotting

If you have been saving spectra data with the --spectra option above, you can plot out a neat spectrograph showing HF band activity over the last few days. This utility can also plot an estimate of total RX power into the KiwiSDR

$ python3 kiwi_spectrum_plot.py --hours 72 --spectrograph my_spectro.png --title "My KiwiSDR"  --rxpower my_rxpower.png myserver_spectra.csv

Arguments are:

positional arguments:
  filename              KiwiSDR Spectrum Log File (CSV)

optional arguments:
  -h, --help            show this help message and exit
  --hours HOURS         How many hours to plot. (default: 72)
  --cmap_min CMAP_MIN   Colormap Minimum Value (dBm) (default: -110)
  --cmap_max CMAP_MAX   Colormap Maximum Value (dBm) (default: -30)
  --clip                Clip file to the hour limit specified with --hours
                        (default: False)
  --spectrograph SPECTROGRAPH
                        Save Spectrograph to this file. (default: None)
  --rxpower RXPOWER     Save RX Power Plot to this file. (default: None)
  --title TITLE         KiwiSDR Name (for plot titles) (default: KiwiSDR)
  -v, --verbose         Verbose output (set logging level to DEBUG) (default:
                        False)

To avoid the spectra CSV file getting really big, you can use the --clip option, which will remove any entries in the spectra file older than the specified number of hours.

Note that the Spectrograph plot is optimized for 72 hours of span with 10 minute time steps, and will look a bit odd until this much data has been gathered.

Putting it all together

You will probably be wanting to run the above scripts automatically. This can be done using a cronjob.

For example, for a single KiwiSDR, the following could be put into a bash script (e.g. run_snr.sh):

#!/usr/bin/env bash

HOSTNAME=my.kiwisdr.server.org
PORT=8073

cd /home/user/kiwisnr-rrd/

# Capture Data
python3 snrtorrd.py -s $HOSTNAME -p $PORT --spectra=kiwisdr_spectra.csv
# Produce RRD Plots
python3 rrdtograph.py -s $HOSTNAME --title "My KiwiSDR"
# Produce Spectrograph Plot
python3 kiwi_spectrum_plot.py --hours 72 --spectrograph my_spectro.png --title "My KiwiSDR" kiwisdr_spectra.csv
# Produce RX Power Plot (Note - uses different log file)
python3 kiwi_spectrum_plot.py --hours 72 --rxpower my_power.png --title "My KiwiSDR" kiwisdr_spectra_peak.csv

# Copy plots to a remote server for display
scp *.png myserver:~/some/path/

This script could then be run in a cronjob with an entry like:

# m h  dom mon dow   command
*/10 * * * * /home/user/run_snr.sh

Notes on KiwiSDR Overload Detection & Monitoring

• snrtorrd.py includes an empirical 'calibration' value of -13 dB: https://github.com/darksidelemm/kiwisnr-rrd/blob/main/snrtorrd.py#L191

• This is the default s-meter and waterfall calibration set in the KiwiSDR admin interface.

• This value is about 3dB off for a single 29 kHz waterfall bin. (should be -16 for a single bin to be accurate, at least at max RBW).

• For the S-meter output, looking at a CW carrier, the s-meter cal is about 5dB out, so a calibration value of -18 dB is better suited for accurate readings in the s-meter display. This will be applied to all the AREG KiwiSDRs, so at least we know the level entering the KiwiSDR. Eventually we will include a calibration out to the antenna feedpoint.

• These calibrations do not affect the data gathered by snrtorrd.py - the waterfall data supplied via the kiwiclient system does not have any corrections applied.

• An empirical conversion of the total power sum (with a single CW signal injected) is: (-13 dB) + (-3 dB) = total of -16 dB.

• The ADC overload point appears to be approx -17 dBm at the input of the KiwiSDR.

• snrtorrd.py and the kiwi_spectrum_plot.py scripts have been updated to reflect these findings.

• To be able to accurately measure the peak power into the receiver, snrtorrd has been updated to save peak waterfall data along with the average data which is used to produce the spectrograph. Peak data is saved to a filename appended with _peak - refer examples above.

Docker

docker build -t kiwisdr-rrd:latest .
docker run --rm -v $PWD/output:/output -v $PWD/data:/data  kiwisdr-rrd:latest host.name 8073 "pretty name" password