CSV output prints only headers

[HaxtonFale]

I'm trying to capture the signals from my weather sensor outside so I tried running the following invocation:

$ rtl_433 -v -F csv:weather/sdr.csv -R 19

The problem is that it produced the following file:

$ cat weather/sdr.csv
time,msg,codes,model,id,channel,battery_ok,temperature_C,humidity
time,msg,codes,model,id,channel,battery_ok,temperature_C,humidity
time,msg,codes,model,id,channel,battery_ok,temperature_C,humidity
time,msg,codes,model,id,channel,battery_ok,temperature_C,humidity
time,msg,codes,model,id,channel,battery_ok,temperature_C,humidity
time,msg,codes,model,id,channel,battery_ok,temperature_C,humidity
time,msg,codes,model,id,channel,battery_ok,temperature_C,humidity
time,msg,codes,model,id,channel,battery_ok,temperature_C,humidity
time,msg,codes,model,id,channel,battery_ok,temperature_C,humidity
time,msg,codes,model,id,channel,battery_ok,temperature_C,humidity

This feels less than ideal.

I'm using RTL-SDR V4 with Debian 12 (bookworm), rtl_433 version 22.11 (2022-11-19) from the repo.

[gdt]

You didn't say what version of rtl_433. If not the most recent release or git master, please upgrade.

You didn't say what happens if you don't give any arguments (beyond -v). It would help to sort out a general problem from a specific decoder.

[HaxtonFale]

I mentioned the version at the bottom, but it is indeed out of date (22.11). I downloaded the most recent release binary and tried using it.

If I don't specify any parameters, it seems to... not pick anything up, for some reason. It's very close to a digital clock that reads the same sensor, so I'd expect it to be fine even if I don't use the largest antenna I can for the radio. Are there any libraries I might want to upgrade as well?

[zuckschwerdt]

Each new start will produce a header and append that. The csv file looks like expected.
Try to run with -Y autolevel -M level -M noise to get an idea what the reception levels are.

[AkaBoing]

hi HaxtonFale,
on bookworm try gqrx (gui version) -(should be in repo)
check if your RTL-SDR V4 was found and any signal at all at 433.9M

Next try -with -R2 and -R19, > rtl_433 -f 432.85M -s1024k -R2 -R19 -M level
without -F !
-R2 Rubicson and -R19 nexus use the same decoder, so one or both should find the meter.
and give an output.
the -F json variant gives me:
{"time":"2025-03-17 10:53:17","model":"Rubicson-Temperature","id":134,"channel":1,"battery_ok":1,"temperature_C":14.6,"mic":"CRC","mod":"ASK","freq":432.89728,"rssi":-0.544018,"snr":34.61048,"noise":-35.1545}

now test with -F csv

have fun

[HaxtonFale]

Each new start will produce a header and append that. The csv file looks like expected. Try to run with -Y autolevel -M level -M noise to get an idea what the reception levels are.

Seems like the program settles on noise being at ca. -31 dB if I use all the options you listed. Other than that, if I ask it to be a bit extra verbose (-vv) it (maybe) reads some data but most of them fail:

[markisol_decode] bits_per_row[0] = 1
[risco_agility_decode] Preamble not found

And so on.

hi HaxtonFale, on bookworm try gqrx (gui version) -(should be in repo) check if your RTL-SDR V4 was found and any signal at all at 433.9M

Next try -with -R2 and -R19, > rtl_433 -f 432.85M -s1024k -R2 -R19 -M level without -F ! -R2 Rubicson and -R19 nexus use the same decoder, so one or both should find the meter. and give an output. the -F json variant gives me: {"time":"2025-03-17 10:53:17","model":"Rubicson-Temperature","id":134,"channel":1,"battery_ok":1,"temperature_C":14.6,"mic":"CRC","mod":"ASK","freq":432.89728,"rssi":-0.544018,"snr":34.61048,"noise":-35.1545}

now test with -F csv

have fun

I'll check out gqrx shortly. As for the rest of your reply: see above for picking up any signals; I was able to see the signals on my Windows PC via SDR# (and decode with the rtl_433 plugin that's how I identified the protocol as Nexus) so I'm pretty confident the hardware can see it, and I just need to figure out how to configure it properly or work around obsolete dependencies (like I had to fix in URH). I ran rtl_433 with the parameters you provided, with timeout after 2 minutes (I know my sensor sends every ~70 seconds), and nothing arrived.

[HaxtonFale]

Running gqrx shows me a sea of noise.

[AkaBoing]

sorry, i didnt know , u already checked it with sdr# on Windows.
the sea of noise (red noise) is normal, but there should be some signal from time to time, even some heavy pilot, jammers or TPMS signals.
for the depencies:
check the "BUILDING.md"

on Debian:
sudo apt-get install libtool libusb-1.0-0-dev librtlsdr-dev rtl-sdr build-essential cmake pkg-config
is all u need.

CMake:
git clone https://github.com/merbanan/rtl_433.git

cd rtl_433/
cmake -B build
cmake --build build --target install

or, if your CMake is V 3.13 or older:
( i startet this way years ago with DEB 10 )
cd rtl_433
mkdir build (if not present)
cd build
make
sudo make install.
done, now run rtl_433.

start rtl_433 -f433.85M -s1024k -M level -g +20 -S unknown.
-g +20 gives a fix gain of +24dB (see the start messages of rtl_433)

view the cus on https://triq.org/iqs/ to get an overview.

[markisol_decode] bits_per_row[0] = 1
[risco_agility_decode] Preamble not found
or
Possible Oregon Scientific v2.1 message, but sync nibble wasn't found. Raw
codes : {24}555555
inFactory, nor-tec, FreeTec NC-3982-913 temperature humidity sensor
codes : {40}f780641362

all these "errors" are normal -vv debug outputs, u will get a lot of them, no problem, but in between
there should be a message like this:

time : 2025-03-13 22:41:08
model : Oregon-v1
SID : 3
Channel : 2
Battery : 1
Temperature: 21.7 C
Integrity : CHECKSUM

or

time : 2025-03-17 23:41:11
model : Rubicson-Temperature
Channel : 1
Battery : 1
Temperature: 6.1 C
Integrity : CRC

im shure, the cracks here will help u more..

GN8

[HaxtonFale]

Yeah, the constant noise was weird because there was nothing that I have seen in SDR#, such as various regular signals or even the sensor talking. That's why I'm thinking that maybe some radio-specific dependency on my system is obsolete, since I had a similar problem with Universal Radio Hacker on Windows because it had an obsolete rtlsdr.dll that couldn't understand the V4 radio. I'll try building as soon as I'm able.

[AkaBoing]

R19_nexus_433.92M_250k.cu8.zip

i made a very noisy "cu" of my -R19 Nexus

R19_nexus_433.92M_250k.cu8.zip)
open the cu8 with rtl_433 R19_nexus_433.92M_250k.cu8
it should show
-R2 Rubicson
or
-R19 nexus-TH

even a solight uses the same decoder for it

if no success, try a live version of Debian, bullseye, and install rtl_433 with the commands above.
tip: use another USB-port , and check if a DECT phone is in range, it can disturb your sdr dongle

have fun and success.

[HaxtonFale]

I was able to read your input file successfully, both on the old and current version. I built from master and it seems to have the exact same issue: not picking up any signals at all.

The version of librtlsdr-dev I have in the repo is 0.6, while the latest is 0.9. I suspect that might be the main culprit.

[AkaBoing]

the v0.6 indeed is the repo version and "old", there are a lot of newer versions, each for a special tuner,
like SharpRTL, R820T, R820T2,E4000, need to check whats up to day and working with your dongle
my version is v2.1.0 (e4000)

Christian? any idea?

edit:
just checked 3 other linux machines, they all have v 0.6.0.4
and all of them work so the 0.6 lib is ok, but i have a e4000 tuner dongle.
mb your sdr-dongle is the problem
https://osmocom.org/projects/rtl-sdr/wiki for details

follow this threath: newest version is 2.02
https://github.com/librtlsdr/librtlsdr.git
clone it and we will see..

dpkg -l | grep librtlsdr
ii librtlsdr-dev:amd64 0.6.0-4 amd64 Software defined radio receiver for Realtek RTL2832U (development)
ii librtlsdr0:amd64 0.6.0-4 amd64 Software defined radio receiver for Realtek RTL2832U (library)

[zuckschwerdt]

The Osmocom rtlsdr version after 0.6 is 2.0.0. The librtlsdr version has a 0.9, but a 2.1.0 seems current. And there are more libs still.

[AkaBoing]

thx Christian, as i checked, the lib 0.6 works for me (with e4000), the new v2.x also, i dont really know whats up with his dongle (seems to work on windows, but not with deb12.

[AkaBoing]

HaxtonFale:
i really suggest, you take a Debian 10 or 12 live version, (Mint or MX does it also) , its no witchcraft..
then sudo apt-get install libtool libusb-1.0-0-dev librtlsdr-dev rtl-sdr build-essential cmake pkg-config
then clone:
git clone https://github.com/merbanan/rtl_433.git

cd rtl_433/
cmake -B build
sudo cmake --build build --target install

note any error message!

then check your dongle, in special the antenna part (mb. the sma is broken)
then take a short wire of 167mm as antenna..
and start rtl_433 -M level
good luck.

[HaxtonFale]

I know it's no witchcraft, but I don't really have many computers lying spare to test things out lol. I'm working on it right now, though one of the machines is currently indisposed and needs to be urgently fixed.

[AkaBoing]

the more computer, the more trouble, i feel with you.
but im shure you come to an result.
stay tuned and dont hesitate to ask for help.
this is a friendly community with a lot of know how around rtl-sdr. so take your time.
meanwhile: whats the model number of the digital clock you mentioned, and-
if possible the model number of your sensor, mostly on backside or in the battery housing ?
my 433Mhz clock is a infaktory zx7382.
have fun

[HaxtonFale]

I ended up needing to reinstall Debian onto the machine and now it is finally back up and running (mostly). It still does not pick anything up... I'll be trying with a live Debian on another machine.

[AkaBoing]

if Debian (or any DEB based live version) is running install gqrx (if not done already) sudo apt-get install gqrx-sdr
check if your sdr dongle was found and check with 433M and 866M with a short antenna.
for me it shows a lot of meters, clearly 6>12db over noise every 6>30 sek
good luck.
btw: on my HP notebook i have to use a 1mtr USB cable, otherwise to much garbage noise too.

and 915Mhz with a lot of LoRa devices:

[HaxtonFale]

Yup, on multiple machines I see nothing but noise with gqrx. I know the radio works, because as I said before I'd used it on a Windows machine with SDR# and have picked up various signals.

[AkaBoing]

Dont know the max possible gain of your V4 SDR, but try -g +42
so rtl_433 -g +42 -M level should show a result.
If Not: try gqrx with 100Mhz or so (a FM Radio) should show Up.
really No clue whats Up with your dongle.. sorry.

[HaxtonFale]

But the thing is that the dongle seems to be working fine? Like, I know it works with other software.

[AkaBoing]

Christian: any other rtl-sdr tool to check whats going on?.
A simple way: take your device meter or any other 433 meter close to the antenna.. 0.5 mtr

[zuckschwerdt]

Sounds like the gain setup in the rtlsdr lib is not working ("works on Windows but not on Debian").
Not sure, is this the Blog V4 without updated lib rtlsdr problem? I.e. no V4 support? You did try multiple (current) versions of rtlsdr libs, right?

[HaxtonFale]

I did also try downloading a compiled version, and building a newer version, of rtl_433, to no avail.

[zuckschwerdt]

For the Osmocom rtlsdr lib you need at least 2.0.0 (the next version after 0.6.0), as this change is needed osmocom/rtl-sdr@1261fbb

[HaxtonFale]

Okay, that did it finally! I built rtl-sdr from the repo you linked, then rtl_433 from master, and it's finally reading the signals. Thank you for the help!

[HaxtonFale]

[zuckschwerdt]

Perfect. We should note somewhere that Debian release are likely always outdated. Not much a problem for rtl_433 as we have binary builds, but the rtlsdr lib is a trap.

[AkaBoing]

Note: the libs v2.0.0 and the newest 2.02 need systemd, no chance with SYSV (MX or antix) - i tested it.
therefore the libs 2.x.x are NOT in the default MX and antix repos. (still 0.6.0-4)

[HaxtonFale]

I have systemd so that shouldn't be an issue. I'll try getting (building if need be) a recent version.