Recorded from a Theremin-in-progress, but the same sound shows up all over the airwaves, LW, SW, professional receivers or homemade. In this case, it shows up when connecting only a foot of wire to a 5MHz oscillator (yes, rather high for a Theremin, this is for S&G's at the moment). Said oscillator seems to be acting in an autodyne capacity.
The waveform seems to be a frequency modulation (advancing at least one cycle) every 8.3ms. Hence the buzzing sound, and hence the FM synth character at different frequencies.
It must be line driven, somewhere, but it's not my power supply, and anyway, that wouldn't explain it being everywhere on the dial. It also seems rather dubious (or impressive?) that it's having such a strong effect on my oscillator, despite using only a one foot antenna.
The ARRL has a collection of common RFI sounds on their web pile: Under household, try the Sony TV: Kinda sounds like a plasma TV. The difference is that your recording has a 2nd signal from your Theremin that is producting a beat note.
Of course, you could identify the source by turning things off in your house until the QRN disappears.
Jeff Liebermann firstname.lastname@example.org
150 Felker St #D http://www.LearnByDestroying.com
From 0 to about 0.7 sec, it sounds like a "carrier", so to speak. From about 0.7 sec to about 1.5 sec, it sounds like there is "data" modulated on it somehow. Then, the theremin starts up.
A quick analysis in Audacity shows some peaks at roughly 17.5 kHz, 14 kHz, 10 kHz, and maybe 6.5 kHz.
Of course, doing an FFT on an mp3 is somewhat dubious. Record to .wav (or other lossless format), grab a copy of Audacity, and see for yourself locally. Not real-time but the price is right.
From about 12.0 sec to the end of recording also seems to be free of theremin sounds. I still think it sounds like "data", but the frequency is different. The spectrum doesn't have peaks, either.
Hit breakers until it stops. If it never stops, try it in the daytime (when streetlights and other lights are off) and and at night. If you have any overhead power lines in your area, wander the immediate neighborhood and look and listen for small arcs at the poles. Is there any (possibly unannounced) broadband over powerline in your area?
I'd hazard a guess that it is a very badly behaved switched mode PSU, speed controller, CFL light or dimmer somewhere. To make such a mess it seems likely that the outward and return paths are not close together so that you have a room bathed in magnetic loop coupled RF noise.
A bit like the method that the T setting on hearing aids uses - might be worth seeing if you can pick it up with a pancake coil and a crystal earpiece then you might be able to metal detect to the source.
Daqarta (sp?) will do it in realtime if you can feed the signal into a soundcard. I have used it for public demos of Sound and Music. ISTR You have 30 days free use before you have to register and it still works as a basic signal source after your evaluation period expires. It isn't outrageously priced so if it does what you want worth buying.
The theremin action is continuous of course, the different periods are just where I put my hand to give you an idea what it sounds like at various places.
Like I said, the waveform looks like frequency modulation -- no matter what the carrier frequency; it seems like constant displacement FSK. I don't get what kind of signal is pulling the oscillator like that, nor how it can do it so strongly from so little antenna. (Is it really that sensitive? Why don't theremins in the MF range do it?)
Beats me. Unfortunately, all noise abatement advice is N/A -- I'm in an apartment, suburban area. The elephant in the room is obviously the computer (for which without, how do I record..), but based on experience, I don't think that's the source anyway.
Since there aren't any liner notes, I can only go by what I hear.
Like Martin Brown said, maybe whatever it is has a huge loop area for some reason.
Jeff Liebermann gave the link to look up BPL deployments.
You can, almost for sure, find out if it's coming from something inside your apartment. (You might not be able to turn off things like a fire alarm system.)
Apartment buildings tend to have main circuit breakers for each apartment at the meter. This may be an unpopular approach, though. :)
Since you said it shows up in commercial receivers too, maybe grab a radio and take a stroll. Or, if your theremin can be made portable, take it along with you.
You could get a piece of plastic 5/32" wide and 5,000 feet long, put some iron oxide on it, and drag it at 1.9 inches per second past a coil with the signal you want to record on it. Naah, it'd never work.
The ARRL page that Jeff mentioned is
, but nothing there really sounds like yours. On the other hand, all of those recordings were made with things that were designed to be radio receivers, so they may sound different than your theremin.
The thing is, *lots* of equipment can generate radio frequencies. At least a handful of people have now listened to your sound sample and nobody has yet said "I know exactly what that is", so it seems likely that you'll have to do more snooping around to find out.
Update: I've put in proper coils (now operating ca. 2.23MHz) and changed around a few things (mixer ports, gain, filtering); I think filtering much over ~5kHz between AF stages has helped (including getting rid of the hiss and whine from the volume section, which uses a slope detector).
It still does it, but only when... get this... it's completely
*un*grounded. I can put ferrite beads on all cables (i.e., power and signal out) and it buzzes; ground it with only one (e.g., clip scope probe to the chassis) and it goes away. This sounds like a grounding problem, but it's build on copper clad, ground plane everywhere, oscillators partitioned. Also, ceramic caps on all outside connections. No opportunity for RF ground loop.
Visual aid: the design is very similar to this one I built a while ago.
Main difference, I split the pitch mixer so there's a separate VGA so I can add effects to the CW pitch signal. (Yes, as a "7 Transistor" project, it's all discrete; this one uses 24 transistors in total, only a little over.)
How are you doing the amplitude modulation? Mine was via a slowish LDR optoisolator to avoid the two RF signals fighting each other.
I also had a bit of fun and games tuning it so that it was linear in distance for logarithmic changes in pitch - the hardest part of all. It drifted like hell depending on room temperature but you could play by ear well enough that people could sort of recognise the tune.
In case it is squegging you might want to try decoupling capacitors at strategic points. Mine proved very problematic when I came to putting it in a nice screened box which was tedious in the extreme until I managed to make it stable again. Something as wild and woolly as a theramin is sat on the edge of wild instability so you have to fight pretty hard to get both the pitch control and the amplitude modulator to behave themselves at the same time.
It could easily be that it needs to be properly earthed to work and protect it form local interference. Mine went through a period where it would only work when earthed too. Eventually it did accept living in a metal box, but it came very close to being put in a shoe box!
Single balanced mixer. Whereas the schematic at the end of my link shows volume (as a control voltage from the slope detector) being fed directly into the pitch mixer (and hence performing gain control), this has separate mixers for each (pitch alone and volume control alone). Pretty basic circuits, seems to work well, and everyone else seems to be doing it (a little searching on VCAs turned up a couple standard circuits following the Moog synthesizer, which is part of the envelope generator).
How was that, pitch into CV into nonlinear function into VCO?
V-to-F's being another synth topic that's notoriously drifty, no surprises there. One thing where monolithic is pretty much required.
Log pitch would be wonderful. These things are so tricky to play because the 'width' of a note varies depending on your distance to the antenna, plus geometry. It's hyperbolic (roughly; there'll be some (x + d)^-3/2 sort of action going on with near fields too), then polynomial or power series (freq goes as sqrt(LC)), then finally log (or exp) for pitch. It's still a one-to-one function, but absorbing all of it analytically (even given a suitable approximation for the antenna field and geometry) into one invertible function is unlikely. An approximation (PWL or whatever) would be possible, but might still require considerable skill to operate because even a 1% accurate correction is off by... about a major sixth? Getting that as clean as a few cents would be miraculous by any means!
Strange, I haven't had much trouble with them (or radio stuff in general, well, a little, but nothing I couldn't iron out).
The present model is happy right now -- good clean tone (while grounded, at least), and playable, though the volume is a bit too sensitive (i.e., you have to move your hand rather far to go from 'on' to 'off'!), and I need to install fine tuning adjustments (the trimmer caps are tunable, but not very user-friendly).
Obviously it's being more sensitive to noise somehow -- but it doesn't make sense that it's more sensitive when the chassis is less grounded. The only obvious 'mode' would be dipole action across the width, but at
2ish MHz and a couple feet across, it's hardly a resonant length. And being a balanced mode, that wouldn't change with grounding (much). So, clearly it's an unbalanced mode. Dunno.