Signals from FineOffset WH51 not decoding on Debian VM

[bassschlumpf]

Hello,

I'm trying to receive data from a FineOffset WH51 sensor in a Debian VM running in Proxmox. RTL_433 seemingly can't decode the incoming data, as the window stays empty.

I'm using a generic black DVB-T Stick with a Realtek RTL2832U and an external small antenna.

Testing on Win10

For testing purposes, I tried receiving the signals on a different machine running Win10 (native) and was successful using the following parameters:

rtl_433 -f 868.3M -Y classic

Sample Analysis

I saved some samples of the incoming data on both machines and compared them via triq.org/pdv. The analyzer can guess a modulation for the sample catched in Win10 (right side), but not for the one on the Debian VM (left side).

If I'm using the guessed settings from the win10 sample, some data can be seen, although the package seems to be shorter. If I'm running rtl433 on the debian VM with -f 868.3M -Y classic -R 142 -vvv I get the following output:

Additional tests

• running rtl_433 on Proxmox natively
• running rtl_433 on a different VM (HA OS with rtl_433 addon)
• using different USB ports on front and back of machine (USB 2 and USB 3)
• using an USB extension cable
• moving the antenna as far as possible from the machine

All of these tests didn't work out

Am I looking at a noise issue here or is it something different? Any help is greatly appreciated.

[zuckschwerdt]

Both pdv screenshots don't look too good. The protocol is 58 µs FSK, any pulse that's not an approximate multiple of 58 is bad reception / noise.
The console screenshot shows that the preamble is detected but then the signal breaks off.

When using pdv zoom in and look at the signal. Is it clipping? Is there noise? How stable is the frequency and amplitude?

[bassschlumpf]

Thank you for the tip! Seems like the signal is indeed clipping:

Frequency seems kinda stable, it seems to be overlayed with a signal of a lower frequency, hence the "up-and-down wandering" of the pulses.

So my next steps would be to play around with the signal gain, and if that doesn't work out, a better receiver position or another dongle?

[zuckschwerdt]

That looks like OOK not FSK. Shift the frequency a little (e.g. -f 868.25M), you don't want to be at DC (i.e. both frequencies should be well away from the center).

[bassschlumpf]

So the little shift in frequency and a gain reduction with -g -3 has brought me a steady signal, it looks like this now:

RTL_433 is detecting every pulse now. Thank you very much!

[zuckschwerdt]

The waveform looks perfect as can be. (reduce the FFT window width some steps to also see it in the spectrogram.)