Looking at MW Carriers

[ChrisSmolinski]

I've started to work on an app to process SDR I/Q recording files, and generate high resolution waterfalls centered around the carrier frequency of AM channels. Below is one example:

This is a recording of 1230 kHz, one of the low power graveyard channels. Though it may look short, it is actually slightly over four hours long. One of the biggest pains with looking at MW carriers is that you need to use very large FFTs, in this case 8 million points. The sampling rate is 250 kHz, so the bin spacing is 0.03 Hz. That means that if you're watching it in real time, the waterfall crawls along... in this case the update rate would be every 34 seconds. And of course if you make an adjustment to one of the settings, you need to wait a few updates to see the effect, that means minutes. Plus you can only see one frequency at a time. Computers are supposed to make this easier!

My solution is offline processing. Let the SDR recording for hours while you're doing something else, then go crunch the numbers. With a 250 kHz sampling rate, you can see 25 MW channels at a time, in this case I had it centered at 1300 kHz, so I can see from 1180 to 1420 kHz. The I/Q recording files are processed once, the resulting FFT data is written to file. Actually only some of it is, a +/-25 Hz span around each carrier frequency, to reduce the size of the output files. I may make this larger if necessary.

The viewing app lets you select the channel of interest, as well as set the mapping of dB value to the color scale, for the best contrast. It just takes a fraction of a second to re-plot the entire waterfall.

The bottom axis shows the offset from the center frequency in Hz. (note that I don't have my 10 MHz reference connected to the SDR right now, so there is a little drift, once I get that connected again it will go away).

How many carriers can you spot? I can see at least 42. I should note this this recording was made from 1pm to 5:30 pm in the afternoon the other day, so not even at night, and yet there's that many stations present.

It is interesting to observe the wide range of variation in carrier frequency. Some have obvious patterns, others are just a single line.

[ChrisSmolinski]

Here is another graveyard channel, 1340 kHz, recorded at exactly the same time:

[ChrisSmolinski]

One more, 1380 kHz. You can see several carriers with a sawtooth pattern. I am going to try to run a very long recording one of these days.

[Kilokat7]

Great stuff Chris.  I look forward to the results when you have your 10 mhz reference connected, especially on the saw-tooth carriers.

[Kilokat7]

Not near the resolution that you're achieving Chris, which looks great BTW, but here's a close-up view of 1629 kHz spanning several minutes captured here last week.  Note the four dominant carriers that stand out, and the high one might have multiples underneath if I could get the resolution tighter.  There's about 10 low powered stations on 1629 kHz in Australia, most listed at 400w.  You're seeing 4 of them here as they peak at my local sunrise.  'Tis the season for Aussie MW DX!

[redhat]

The sawtooth pattern your observing is most likely due to ambient temperature swings within the transmitter building that house the frequency control gear.  Most AM's use a non-ovenized crystal oscillator for frequency control.  The FCC spec for carrier deviation is +/- 20 Hz of assigned frequency, so the swings can be quite large.  Some IBOC stations have their station synthesizer locked to GPS, so their carrier stability should be many times better, typically better than 1 PPM.  It will be interesting to see this on SW, a new way to separate the men from the boys 

+-RH

[jFarley]

Extremely impressive guys, and great work!

Chris, that is a cool app.  I see the makings of an ultra high resolution backend (not real time, but operating on short IQ recordings) QRSS viewer in there somewhere.  To die for.

kk7, I have been wondering about what the propagation mode is for these long haul TP receptions.  The path is almost entirely dark and over water, and it seems as if the mode could be groundwave only.  It also seems as if the enhancement at sunrise suggests some sort of dumping mechanism as the ionosphere transitions near the terminator.  At the high end of the band could this possibly be multi-hop skywave or even chordal hop skywave?  Would appreciate any take you have on this.

[BDM]

I did the same thing with 1710 some time back. I was amazed at all the carriers observed. I seem to recall my all time high for carriers on 1710 was 24+ noted in the high res waterfall one night

https://www.youtube.com/watch?v=yfFoKvYfqkU

[ChrisSmolinski]

jFarley: That's a pretty nifty idea. Right now the app is set up for assigned MW channels, but it could certainly be modified to allow arbitrary frequencies.

BDM: Here's a look at 1710 from last night. 41 carriers? I lost count. Not sure if these are all "radio stations" or if any are local or semi local QRM. I need to run it again, zoomed in to see the mess in the middle. One of them has a fast wiggle, there is another with a long period wiggle. Kilokat7 and I were musing on IRC this morning whether or not you could "fingerprint" stations looking for patterns like that.

[BDM]

WOW, yep, those squiggly lines are transmitters shifting frequency. Kilo and I spoke about this some time back. Bottom line is we're receiving signals from transmitters though I agree, some are most certainly local "hash". But I believe many are actual transmitters, just so weak, it takes a fairly low noise floor, SDR and good antenna to detect them. One thing I know for sure, many of these "Part 15" players use 1710 often. I have no doubt some of these are simply very weak Part 15 players or transmitters just above Part 15 power levels making it through. I have a couple so called hobby broadcasters in my area, whose signals I can receive for several miles during the day.