FT8 Skimmer base on kiwirecoder

lazywalker

The idea was learned from wsprdaemon by @rrobinett, and a post back to 2018 by @k1ra http://forum.kiwisdr.com/discussion/1377/suggestion-ft8-decoder.

WSPR is good, yet FT8 mode is more and more popular nowdays, more signals, easer to pickup(compared to wspr, some people use maga watt to transmit FT8).

This can be useful in determining propagation conditions or in adjusting antenna. A wide band antenna kiwisdr located in urban area in my case, with 10 bands requests at the same time, can be easily reach over 10,000 spots within 24 hours.


It is also interesting to see how long it takes to spot 100 different DXCC/countries. (A well placed station with a decent antenna can do this within a week of monitoring, but the best systems can do it within a single day).

You can check the code from https://github.com/lazywalker/DigiSkimmer

BD7MQB Michael

Attachments:
https://forum.kiwisdr.com/uploads/Uploader/4a/23fe5751ff41f561d9ccd1c0e9f8ff.png
https://forum.kiwisdr.com/uploads/Uploader/d9/0a6cf92af642a334cb586149642f0a.png

Powernumpty

Took a bit of fiddling on a fresh Debian but that's just me.
Once working I checked the PSK reporter map and normally for my call that is just showing 6m spots (my DD5JKF OpenWebRX RTL dongle defaults to that when not in use).
The map looked wrong, lines all over then I realised FT8 reports much quicker than WSPR so those were real spots, already.

Thanks for the code
Stu (M0AQY)

--within the 24hr edit window--

lazywalker

Hi, just keep you posted that i have FT4/WSPR and band hopping feature added few days ago, you can mix those modes within one slot.
also, a Dockerize setup will make it easy to go.

wspr spots will upload to both pskrepoter/wsprnet, so you can track all modes at the same time with pskreporter.info.
for details plz check the README

TNX,
Michael

Powernumpty

I did just run it (old version) for a couple of days, will update it and try again soon.
It took from about midday to 7PM local for it to hit 100 countries, 144 in 24hrs from about 36k FT8 spots.
My Wsprdaemon config does currently merge results from two Kiwi's using antennas with nulls N-S, E-W so I'd after the same if I used for WSPR.

Anyway still somewhat surprised others have not tried publicly the script (sorry if others have, and I just missed it).

Probably as a result of how it reports, didn't see many Kiwi's listed on the pskreporter.info listings. IMHO it would be good to be able to list platform and antenna (I may have missed that capability already).

Stu

lazywalker

Hi Stu,
compared to wsprdaemon, digiskr is quite simple since it can't merge the same bands. the benefit of digiskr might be combination of multi-modes and cross-mode/bands hopping (easy to implement JT65/JT9 thought) .

And yes, the new version can grab the antenna information from KiwiSDR and upload it to pskreport, i'll see if I can add some platform infos

Michael

cathalferris

I am currently using the docker image of DigiSkimmer, on the same pi that's under my recently-delivered RaspberrySDR, working fairly well (111 DXC and 9000 decodes on 20-80 FT8 in 24hr). The SDR is running the 4.08 Flying_Dog branch of code.
It's a nifty bit of software, completely fulfilling my desired niche and use case. It's also pulling the antenna info from the SDR, and that's pretty cool.

I have noticed that the docker image does appear to have an issue, possibly with memory usage/memory leak, as the decodes stop after ~8 hours and the free memory of the Pi consistently linearly drops to zero.
Bog standard Docker installation as far as I am aware, and started with this line:
docker run -d --restart unless-stopped -eTZ=`cat /etc/timezone` -v digiskr:/opt/digiskr --tmpfs=/tmp/digiskr lazywalker/digiskr

To mitigate the issue, I'm restarting the docker container every 6 hours via cronjob, and it appears to be functional, the used memory is freed up and decodes are now continually being sent to pskreporter.

I haven't yet delved into the situation to see if it's my Docker config or if it's the container that is showing the problem, but the continuing container restart works well enough for the purpose.

I must also organise better cooling of that Pi, it's not happy without forced convection at the moment. Work in progress.