Graw DFM-17 sonde signal right next to 403.20 spur

[Mark Conner]

I'm using a Nooelec SDR with radiosonde_auto_rx to receive Graw DFM-17 sondes from the local NWS office (OAX, Valley NE).  Their Graw system was just commissioned and they selected 403.21 MHz as their primary frequency because it gave them the best performance.  This appears to be close enough to the 14th harmonic of the 28.8 MHz clock frequency used in most (all?) RTL-SDRs to prevent good decoding as the signal gets weaker - that harmonic is at 403.20 MHz.

I have two SDRs, a SmartTee and a SmartTeeXTR.  Both have this harmonic.

The image in this email shows a waterfall display captured while the DFM-17 was in flight (but quite a ways away).  The software will decode the sonde when it is nearby, but appears to lose it at about 50 km.  The same setup will pick up other sondes in this band out to nearly 300 km.

Is there anything I can do to my radiosonde_auto_rx setup, rtl_sdr configuration, or purchase a different SDR with a different clock freq that would mitigate this harmonic?

73 de Mark N9XTN

[Mark Jessop]

Ahh, the classic RTLSDR spur... They really did pick a bad frequency for their launches then...

10 kHz off from the spur is a big problem, as the passband of the demodulators are about 30 kHz wide, so what will happen is is that when the sonde signal is weaker than the spur, the FSK demod will lock on to the spur and not decode the sonde. 

The demod passband is set this wide intentionally, as it allows for the tracking of drifty sondes, and the DFM09 sondes in particular are known to drift a fair amount.

I'd suggest giving the station a call and seeing if there's anything they can do to move the frequency a little. Even going up to 50 kHz would probably help...

73

Mark VK5QI

--
You received this message because you are subscribed to the Google Groups "radiosonde_auto_rx" group.
To unsubscribe from this group and stop receiving emails from it, send an email to radiosonde_auto...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/radiosonde_auto_rx/CA%2BTd0WTX0-rUp93%3DWBxZVddge12azYcJDskGNxvcdoj83Ggpsw%40mail.gmail.com.

[Mark Jessop]

One trick that you could test:

Set the value on this line to 0.075 : https://github.com/projecthorus/radiosonde_auto_rx/blob/master/auto_rx/autorx/decode.py#L783

This will have the effect of raising the bottom end of the frequency estimation window of the FSK demodulator to not include the spur. 

However, if the sonde does drift throughout the flight (and maybe the DFM17's aren't as bad for this?) and it drifts down in frequency, decode performance will be impacted.

73

Mark VK5QI

[Mark Conner]

I don't think the DFM-17s drift that much.  I've been watching one in flight for a couple hours and it hasn't gone very far in frequency at all.

I now have a copy of the edited Python code in a file on the Pi.  Forgive my Docker ignorance, but how do I get it inside the container so it's run?

73 de Mark N9XTN

[Mark Jessop]

Ahh, running inside the container its a lot harder to make modifications.

You could try getting a shell inside the container: https://github.com/projecthorus/radiosonde_auto_rx/wiki/Docker#opening-a-shell-within-an-existing-running-container

and then editing the file, which should be located somewhere in /opt/auto_rx/autorx/

Cheers,

Mark

[Mark Conner]

About the time you sent the email, I found out how to copy a file to a location inside the container.  Do I need to recompile, or just restart the container?  I just restarted it, but the sonde has gone over the horizon now.  Next one should launch in about 40 minutes.

- Mark

[Mark Conner]

I tried the modified .py code and the software would not decode the signal after launch.  I copied the original code back, restarted the container, then got decodes out to about 8-10 km.  

It appears the RTL-SDR Blog V3 SDRs do not have this issue.  One station in the KC area receives just fine on 403.21 MHz with one, and another ham in upstate New York also had sondes on 403.21 after initial NWS commissioning.  However, that NWS site was willing to adjust the frequency away from 403.20 so more receivers could pick them up.

I am in the process of getting one of the V3 SDRs to try next.

73 de Mark N9XTN

[Mark Jessop]

The v3 SDRs have the spur there too. It appears to be endemic to all RTLSDR models.

Good to hear that the NWS are receptive to moving frequency though!

Sent by means most mobile.