RTL-SDR suddenly not being recognized

SatNOGS Software
1

Hi together,

I’m using the SatNogs client in a Docker container on a Raspberry Pi 5 and an RTL-SDR (Nooelec v5) stick to run a ground (station).

Recently, I’ve been having the problem of my RTL-SDR stick suddenly not being recognized during operation (RuntimeError: No RTL-SDR devices found!). Even setting the hardware serial number in SatNogs Client manually does not lead to any improvement.

Unfortunately, the software then no longer starts recording automatically and no longer recognizes the stick. The stick is displayed (rtl_test and GQRX), but only when I restart the SatNogs client manually does everything work as expected again. Unfortunately, the bug comes back over time…

Has anyone had any experience with this?

2

Good day,

Lets try and see if we can find the cause of this.

But we need more information on your setup and as much details as possible.

Full hardware setup, log files, package information, lsusb output, etc.

1 Like
3

Sounds like rtl-sdr overheating and misbehaving, I have seen them disappear before. Running a high sample rate makes them hotter, proper cooling helps.
samp-rate-rx: "2.048e6" is not high by any means, but I’d try 1.024e6 to test this.

1 Like
4

Hi PE0SAT,

Thanks for the quick reply. I have attached the output of lsusb, rtl_test and docker-compose.

I am using this GitHub-Repo

5

Hi SA2KNG,

Thank you for the quick reply. I’ll test with a lower sampling rate. Heat is indeed a problem there. However, the last failure was in the middle of the night, when it should actually be cool.

The receiver and the antenna are in the centre of a city on the 6th floor - there may be strong sporadic radio signals in the surrounding area causing interference. However, this should not cause the receiver to crash.

Does the SatNogs client not have an automatic restart mechanism after such an error occurs?

6

Any QRM usually doesn’t affect the rtl chip, only the tuner front end.
If it disappears on the USB lsusb then there’s either a cable/connector issue or thermal in my experience.
If it is still recognized as a device, then the rtl driver from lsf can reset it, but that doesn’t work if the above issues is the main problem.

7

Yes, thank you already. Let’s see how it works with a lower sampling rate. I would have thought about the cooling next… That’s not so easy in this room :slight_smile:

I cannot yet say whether the device is no longer displayed on lsusb in the meantime. In any case, I don’t have to reconnect or restart anything. I just have to restart the SatNogs client manually. The device is always displayed when I check it.

8

I don’t know how those nooelec sticks look on the inside, the rtl-sdr.com v3 and v4 have a pretty decent alu case with a thermal pad on one side of the board, not perfect for a hot cabinet thou.
A fan will definitely help, does not need to be strong.

If a client restart is sufficient, then it might have a stuck flowgraph running:

$ ps ax|grep satnogs_
 990588 ?        Sl     0:37 /usr/bin/python3 /usr/bin/satnogs_cw_decoder.py --soapy-rx-device=driver=rtlsdr,serial=1864 --samp-rate-rx=2.048e6 --rx-freq=437425000 --file-path=/tmp/.satnogs/data/receiving_satnogs_10151671_2024-08-31T17-15-20.out --waterfall-file-path=/tmp/.satnogs/data/receiving_waterfall_10151671_2024-08-31T17-15-20.dat --decoded-data-file-path=/tmp/.satnogs/data/data_10151671 --ppm=-1 --rigctl-host=rigctld --rigctl-port=4532 --gain-mode=Overall --gain=3.7 --antenna=RX --enable-iq-dump=0 --udp-dump-host=0.0.0.0 --udp-dump-port=57356

Checking for the device:

$ lsusb |grep -i rtl
Bus 002 Device 024: ID 0bda:2838 Realtek Semiconductor Corp. RTL2838 DVB-T
1 Like
9

The reduced sampling rate has already led to an improvement.

It’s probably really the heat. I’ll test a small fan next week.

I’ll be in touch again.

1 Like
10

Hi @DD7MB_Max,

One other thing that I found troublesome when I first started playing with RPi’s is their power supply and the temperature. If they go into under voltage they become unreliable and erratic and unless you have a monitor attached you get no warning. Similarly if they get overheated they will often reduce the clocking speed and that confuses things. Dropped samples etc.

I found using a POE+ hat with a fan solved both those problems with RPi 3B+ and RPi 4’s but I have not played with RPi 5’s seriously yet.

There are some RPi commands that will allow you to look at voltage, temperature & clock speed of the CPU. Worth considering.

Regards,
John - VK4JBE

1 Like
11

Hi @vk4jbe ,

That’s right. Thanks for the reminder. I had problems with that at first, too. In this case, however, I now also strongly assume overheating.

12

Hi together,

@SA2KNG was right with his guess.

I am now cooling the stick with a small fan. So far, there has not been any further failure, even after increasing the sampling rate.

Thank you all for your contributions.

3 Likes