Here's a comparison between the capabilities of SDR# vs the HDSDR coupled with RDS Spy.
As you can see, SDR# is a little bit weaker than RDS Spy, and doesn't quite manage to decode the PS in the RDS. Other errors occur too.
This is using an IQ file recorded on my mobile SDR set-up. I was not in a good reception spot at the time, hence the lack of anything else surrounding Free FM.
CLICK HERE to download either the IQ file or a single frequency IF file to try yourself in whichever software you choose.
Software that will play the above linked file:-
SDR#
HDSDR
For RDS decoding in HDSDR...
RDS Spy
Use this to pipe the audio from HDSDR to RDS Spy...
http://vbaudio.jcedeveloppement.com/Download_CABLE/HiFiCableAsioBridgeSetup_v1007.zip
Link above taken from http://vb-audio.pagesperso-orange.fr/Cable/index.htm
Remember you need to set up the virtual audio cable to send out and 'record' 192000Hz (192KHz) sample rate. HDSDR needs to be set up to send out via the virtual cable, likewise RDS Spy needs to be set up to accept input from the virtual cable.
10 comments:
There is some interesting imbalance in the constellation of the RDS signal. I wonder whether this is the reason SDR# is struggling :)
Hi Martin. What software do you use to view the constellation? The signal recorded isn't the best as I was mobile at the time. Although I found a spot to settle whilst recording, I was still setting up the aerial for part of the recording.
I should have introduced myself, I'm the developer of SDR Touch. Although the app doesn't have this option yet, I have a workbench for testing the inner workings of the algorithms which include plotting the RDS constellation.
Using your IQ file, this is how Free Radio looks like:
http://martinmarinov.info/temp/freeradio.jpg
While on the same file, you have BBC Bristol which has a normal constellation
http://martinmarinov.info/temp/bbcbristol.jpg
In fact I've never seen an RDS constellation as broken as Free radio. The curious thing is that SDR Touch algorithm managed to decode the RDS with 0% error rate.
This means that whatever RDS encoder is being used, the oscillator that generates the BPSK has a component that is out of phase/amplitude. The actual bit stream is perfectly valid RDS signal with a PI of 0xF000 :)
Ah, hello Martin. Thanks for the images. Have you seen this thread on the SDR# Yahoo group - https://uk.groups.yahoo.com/neo/groups/SDRSharp/conversations/topics/17691?reverse=1 ?
There are files there that contain a strong local station on 103.4MHz which although it's RDS decodes in RDS Spy and SDR Console, it will not decode in SDR#
It might be useful for you, and I'd love to know if there is a similar problem with the constellation.
I think SDR# just has trouble with the weaker signals. There are some other stations I get where the RDS decodes instantly on other software, but barely a character comes through in SDR#.
Here's a proper link to that yahoo page - https://uk.groups.yahoo.com/neo/groups/SDRSharp/conversations/topics/17691?reverse=1
I'm lost in that thread. Can you give me a link to the baseband IQ recording? (I can't work with IF recordings with my setup).
I've spend the last couple of months optimising the RDS reception capability of SDR Touch as much as possible so I was on the look for IQ recordings to test them with challenging receptions.
As for why SDR# is having issues, it is my guess but if an RDS decoder follows the RDS protocol it will rely on the RDS signal being phase locked to the stereo pilot tone. In practice, some radio stations don't have their BPSK coder locked to the stereo pilot. So if SDR# is relying on phase locked RDS, it will not decode RDS from some stations.
If you want to see the RDS constellation yourself, you can use gnuradio and the gr-rds plugin.
As for Free FM, when giving it a good thought, it might mean that the RDS coder outputs a non-sinusoidal signal. It could as well be a triangular or rectangular oscillator, rather than a nice and smooth sine. This could explain the skewed constellation plot.
Which proves that whoever is broadcasting the radio has been very inventive :) I wouldn't be surprised if the RDS signal of the radio is being generated by some kind of micro controller/raspberry PI that doesn't have a lowpass filter on its DAC output.
SDR# IQ Files as requested :)
There are two with the station on 103.4 and another with a weaker station on 98MHz which is one of those that never decodes in SDR# but always instantly pops up in SDR Spy.
https://www.dropbox.com/sh/tqkabqh4161pr4y/AAAQgiDjcigX_zMbygHmp8Vxa?dl=0
Post a Comment