(-A) refuses to acknowledge an FSK signal with a weird encoding?

[Chris West]

I've got what looks (in a waterfall) like a low noise, pretty clear-cut FSK signal: https://quad.pe/e/w8t776CJFT.jpg

`rtl_433 -S unknown` doesn't seem to capture it (hard to say as it's not exactly quiet in here).

`rtl_433 -r f.cu8 -A` on a manual capture/trim prints no matches.

I can't get `-X` to print anything, e.g. `-r rtl_sdr-raw.cu8 -s2048000 -X n=cc,m=FSK_PCM,s=80,l=300,r=800`.

Is it dependent upon there being an underlying encoding like Manchester, which is apparently missing here? After manually decoding it in e.g. pulseview, the bits don't seem to make any sense, so it's possible it's some weirder encoding (run-length?). The clock sync doesn't seem great, but it's less than like 5us of error per bit, on 100us bits, which should be recoverable.

Any hints? Or am I doing something else comically wrong?

The data is coming from a https://www.renew-energy.co.uk/ "CHARGE+" current clamp, using this daughterboard I've failed to identify: https://quad.pe/e/lb8O7CFBmK.jpg .  I'm not expecting a full decode at this point, I was hoping to capture some bit streams and look for variations across transmissions. I just want the bits, dammit. :)

Cheers,

Chris.

--

Raw from gqrx:

https://b.goeswhere.com/gqrx_20239999_162849_433000000_3200000_fc.raw

Raw from `rtl_sdr -g 0`, trimmed with inspectrum:

https://b.goeswhere.com/cc-2023-09-15T0940.cf32 (cf32, 433195000Hz, 3.2Msps)

Manually lowpassed and decimated to just the signal:

https://b.goeswhere.com/cc-2023-09-15T0940-lp16.sr200000.cf32

[Christian Z.]

Some notes: the .raw can be renamed to .cf32

Readable metadata about the sample rate needs to be e.g. _3200k_ for it to work automatically in rtl_433 or pdv, Similarly _433M_ for the center freq.

rtl_433 needs a quiet lead-in and lead-out in the samples files to frame the signal.

There are two signals in the recording:

The first signal is clipping, much to loud, and produces a visible alias, the second one seems to be just the alias. I guess the sender is hopping and this is actually off-band.

Your processed signals are centered on DC (433M here) which is not optimal.

I'd try to record again and adjust antenna, sender distance, and gain such that you don't get clipping and no alias from the off-band signal.

rtl_433 works best with 250k to 1024k sample rate so you may need to decimate afterwards.

If you have Sigrok Pulseview then you can inspect the internal demod state with e.g.

rtl_433 -W myfile_433M_250k.sr myfile_433M_250k.cf32

[Chris West]

Still no joy here.

I've tried significantly quieting the signal, by putting the transmitter behind some walls, and forcing the amplifier really low on the radio (just a (nooelec) rtl). I've tried decimating it down to the right sample rate range. I've tried lots of quiet on the front.

It's almost eliminated the pre-lowpass image, so it's probably about right: https://quad.pe/e/X8I0ldapVn.jpg

Raw capture (low gain, so quiet, if you see nothing on your waterfall): https://b.goeswhere.com/g16-leadin.sr2500k.fc32

Decimated /4 to 625ksps: https://b.goeswhere.com/g16-leadin.sr625000.cf32

The sigrok view, on the latter, still reports the same thing, it can easily turn it into a square wave (although can't decode it, apparently, as it's not centred around the average anymore). Also amusingly sigrok can see a square wave in the AM signal, probably because I did a poor job of bandpassing it. The "FSK" line reads flat zero (or null?). I don't know how to interpret that? https://quad.pe/e/2eFGsDnpCj.png

[Benjamin Larsson]

On 16/09/2023 16:14, Chris West wrote:
> Still no joy here.
>
> I've tried significantly quieting the signal, by putting the transmitter
> behind some walls, and forcing the amplifier really low on the radio
> (just a (nooelec) rtl). I've tried decimating it down to the right
> sample rate range. I've tried lots of quiet on the front.
>
> It's almost eliminated the pre-lowpass image, so it's probably about
> right: https://quad.pe/e/X8I0ldapVn.jpg
>
> Raw capture (low gain, so quiet, if you see nothing on your waterfall):
> https://b.goeswhere.com/g16-leadin.sr2500k.fc32
> Decimated /4 to 625ksps: https://b.goeswhere.com/g16-leadin.sr625000.cf32
>
> The sigrok view, on the latter, still reports the same thing, it can
> easily turn it into a square wave (although can't decode it, apparently,
> as it's not centred around the average anymore). Also amusingly sigrok
> can see a square wave in the AM signal, probably because I did a poor
> job of bandpassing it. The "FSK" line reads flat zero (or null?). I
> don't know how to interpret that? https://quad.pe/e/2eFGsDnpCj.png
>
>

Can you post the actual command lines you are using? You post so much
data that it is hard to sort out the details.

MvH
Benjamin Larsson