BBC World Service DRM @ 7355 kHz, 30 Sep 2012

In the evenings, the noise floor in the lower bands spikes to -100 dBm. I've yet to find a cause, and I can only guess it's due to some electronics in a neighbouring house, because the interference doesn't experience fading and the auto-attenuator frequently kicks in.

Suffice it to say I tend to avoid everything up to the 9 MHz band, except maybe REE DRM on 9630 kHz, but only because it's so strong, and only in the last hour on Sundays because they run a music segment. Solid reception is still not guaranteed due to co-channel and the raised noise floor.

This is all a long-winded way of saying that I had no business scanning around so late at night, but I did, and discovered BBC DRM on 3955 kHz. It was a disaster of course, but I stuck around and until BBC migrated to dual frequency broadcast on 5875 kHz and 7355 kHz. 5875 kHz was too spotty, so I eventually settled on 7355 kHz which, while not great either, was a lot quieter noise (-110 dBm) and interference-wise.

Sick! Journaline and AFS? I've never received these before. Unfortunately it was still not stable enough to populate all of the Journaline hyperlinks.

Interference around 11670 kHz

Note how there are streaks of interference that intrude on the left hand side of the RNZI DRM? Compare that to a similar screenshot taken in July. Interference shifted up in frequency. So this actually happens and I'm not seeing things. Still ridiculous.

Radio Romania International DRM @ 11875 kHz, 30 Sep 2012

Remembered to actually get a reception plot this time. The spike is -10 minutes is the transition from the RRI French program at 11830 kHz as it signed off, to this one.

Although this English program is AAC+ Stereo, most of the volume is in the left ear. They should just leave it Mono.

Fading is more severe, and there's co-channel from EWTN on 11870 kHz.

Radio Romania International DRM @ 11830 kHz, 30 Sep 2012

Caught the tale end of the half-hour French program, which had a snippet of Beethoven's 9th in 20.96 kbps AAC+ Mono. Usually I get the station ID at most, because the interference to the right overlaps. Tonight though, I think it's actually shifted more to the right, and signal strength is quite strong, with peak pilot power at -75 dBm and peak data subcarrier at -80 dBm.

The Disco Palace @ 17875 kHz, 29 Sep 2012

Most weeks I tend to hear the same tracks over and over (like Grease Lightning, In the Navy), but the first song past 20:00 UTC was different.

Lately I've been seeing the signal frequently hit with brief but deep flat fading. When it does occur, it tends to be the dominant fading phenomenon.

In the fade that occurs at -27 s, a 2 second long interleaver isn't sufficient. Not only are most of the subcarriers wiped out over 2 seconds, to recover from it would require SNR to spike immediately after. In two more recent fades, this happens, but the fade itself isn't as long.

Like I mentioned at the beginning, when kind of flat fading happens, it happens frequently, and many times the interleaver isn't long enough to deal with it. Choppy reception ensues.

Voice of Nigeria DRM @ 15120 kHz, 29 Sept 2012

The past couple weekends there was really nothing on the waterfall at 15120 kHz. Poor propagation conditions? Power went out? Who knows.

Took about 15 minutes, but SNR is now 14 dB and fading is currently modest. Last time, there were brief flat fades that the interleaver could sometimes handle, sometimes not. Also note that there is a -127 Hz carrier offset. Doppler is quite high. Only from Africa it seems.

Audio quality may be marginally better, but I'm not a connoisseur of people talking with a bag over their heads.

Bleed-over from REE was light as well, until...

At 19:26 UTC, audio feed temporarily cut out despite solid reception. Early on when decode was spotty there was also no audio feed.

RNZI DRM @ 17675 kHz, 27 Sept 2012

Recently propagation for RNZI weren't so great, with even 11675 kHz not being decodable, but they've gotten better in the last two days.

There's co-channel interference with Radio Azadi, something I don't recall ever seeing before. Net SNR is 15 dB as a result. We'll see what happens in about 20 minutes.

Edit: Radio Azadi went off-air just in time for RNZI to begin to fade. Also, some electric motor kicked in and raised the noise floor 5 dB, but the fading alone was sufficient.

RNZI DRM @ 9890 kHz, 7440 kHz, 16 Sep 2012

Was up way later than I had should have been, but wound up catching the tail end of RNZI on 9890 kHz.

Solid MSC reception. What's also impressive about the RNZI transmitter is that it doesn't have any of the splatter seen everywhere else, except maybe Abuja, but that's only because I haven't seen enough of it (and they were offline this weekend).

7440 kHz was 15 dB weaker, and there was co-channel interference in the upper sideband. No MSC successes.

Signal splatter from EWTN @ 15160 kHz

The Lord commands you to spray energy +/- 25 kHz outside of your designated carrier.

The Disco Palace @ 17875 kHz, 15 Sep 2012

Very little fading at the beginning, but propagation conditions can turn on a dime.

Loss of 10 dB in a matter of seconds, across the board, followed by a further 15 dB selective fading.

WR-G31DDC Excalibur may need to warm up

All those times where I've complained about large frequency offsets and non-stop drifting may have been a result of the Excalibur's oscillator. Most times that I've observed this happening on The Disco Palace is when I just turn the receiver on, and when it's stable and only off about 5 Hz is when the receiver has been on for about an hour, because I've been monitoring Voice of Nigeria DRM before and complaining about the drift there.

As an experiment a few days ago, I left the Excalibur on, but did not activate the DDC pressing the On/Off button in the GUI. The result was immediate frequency offset, so it's not enough to leave the receiver powered on, and there's the possibility that exiting the GUI would cause offsets to come back after a while.

The Disco Palace @ 17875 kHz, 10 Sep 2012

Looks normal once again.

I wish that 20 dB SNR wasn't marginal for 20.96 kbps. These days we can do better than just using a convolutional code plus interleaver. At least append a Reed-Solomon code or something.

Voice of Nigeria DRM @ 15120 kHz, 10 Sept 2012

16-QAM, but Mode C? No one uses Mode C, even if they're worried about reliable reception in a far target area. Mode B can handle 1.1 Hz of Doppler at 64-QAM, see RNZI @ 17675 kHz where I'm not even on beam.

At the time of these screen captures there was some financial news program where they were talking about going cashless in Nigeria. You know what would be easier than going cashless? Not recording your interviews by putting a microphone in front of the speakerphone that you're using to conduct your interview. And how is it that the interviewer is quieter than the interviewee?

Also, your mobile phones are sending tones into speakers and microphones. Turn them off.

The Disco Palace @ 17875 kHz, 9 Sep 2012

Here we go again. But this time, I managed to get SDC constellation, since it seemed to be breaking frequently.

It does appear as if there are 8 groups Real, and 8 Imaginary, so 64-QAM in the SDC, at least for brief periods of time.

But modulation shenanigans went away at 20:05 UTC. Not that that fixes the broken audio stream. Since 20:20 UTC my SNR has been falling, so whatever. I guess when they say they target North America, they really mean the USA.

Voice of Nigeria DRM @ 15120 kHz, 9 Sept 2012

VoN decided to call it a day at 19:37 UTC. Maybe it's because they were too embarrassed about their audio quality and realized they were just waiting money leaving their transmitter on. Someone needs to be fired. It's worse than double sideband buried in the noise.

Nevermind, I spoke too soon. They came back at 19:43 UTC. I've only got -100 dBm subcarriers so I'm not decoding anything. 16-QAM is where it's at.

Also, stop drifting. Don't people pay money for temperature stabilized oscillators any more?

The Disco Palace @ 17875 kHz, 8 Sep 2012

Signal was really strong, SNR was high, but I was kind of concerned by the waterfall. There was a ribbed or chain-mail texture to the modulation, like they were individual subcarriers being overmodulated or modulated with the wrong constellation (QPSK), and that was something I had never seen before. There were also serious decoding issues, with silent audio frames, a snippet of audio, and then a series of undecodable frames.

When the transmission abruptly stopped around 20:01 UTC I thought they might fix it. No such luck.

I tried widening the bandwidth and re-centering, thinking that may have actually helped. But with an SNR of 25 dB, frequency tracking was obviously not the issue.

As you can see from the above, solid audio decoding began around 20:45 UTC, the last 15 minutes of the broadcast. This was marked by a disappearance of that ribbed texture. But it made one final appearance, and audio dropped out in response.

One thing that I should have done was check the MSC constellation to see if it was indeed not using all elements of the 64-QAM alphabet.

Redeeming quality: Stable frequency offset (about 5 Hz if I recall correctly).

Voice of Nigeria DRM @ 15120 kHz, 8 Sept 2012

I was listening to Radio Kuwait on 15540 kHz and looking around the DDC spectrum when I spotted a DRM signal. As you can see, signal power steadily increased, with snippets of audio decodable beginning 18:46 UTC.

At the time, the suppressed carrier was slightly below 15120 kHz, more like 15119.78 kHz. Over the course of the broadcast it would drift higher.

Beginning 19:00 UTC, REE came on, and its upper edge occasionally bled into the lower edge of the DRM signal. The above was my futile attempt to adjust the demodulation window to mitigate some of that bleed, but there's not much to be done about co-channel interference.

The effect of a relatively high power bleed is apparent on the per-subcarrier SNR plot, knocking down the first 10 or so subcarriers. Overall SNR is marginal so this results in lost audio frames.

Towards the last half hour SNR went up to the point where co-channel interference didn't really do a lot of damage.

That said, actual audio quality was trash. It's overmodulated to the point of unintelligibility at times, and is generally unpleasant to listen to. The only decent part was a recorded phone interview, and only the interviewee had any kind of clarity. Step away from the damn microphone or something.

Also, 13.08 kbps on 64-QAM? Just use 14.06 kbps 16-QAM like Vatican Radio did before they stopped broadcasting.

The Disco Palace DRM @ 17875 kHz, 2 Sep 2012

Yesterday was a write off, but today is way better with 26 dB max SNR.

Unfortunately the stream is aesthetically displeasing (read: unlistenable) because audio channels are dropping at the input. They probably run their internet stream into a DRM encoder, and the stream is taking packet loss at the encoder input.

I get only left channel, and then only right channel, and then nothing at all, and then things come back but the audio loops twice before proceeding, all symptoms of the encoder trying to interpolate (or failing to interpolate) for lost input audio frames.

Also, the song title message is stuck on "The Boss" by Diana Ross. Guys and/or gals, this isn't amateur hour with a 100 kW transmitter. Get your act together.

So I want an account at the drmrx forums

Edit: Manual sign-up required.

Except the administrator has disabled registration. What a dick move.