2003-04-04
OK, it isn't large in terms of the wavelengths it was built to capture, but large enough and composed of enough copper to need a substantial wooden frame. It has 51 turns of heavy (4 mm squared) wire, a side length of 1 metre. i.e. about 200 metres of wire.
Construction was a family afternoon affair, with Mum and Dad contributing their muscles to holding and manipulation of the often large subsections, and most importantly gentle advice throughout the procedure. My girlfriend also did the most labour intensive part of the construction, the rotation of the spider as the 200 metres of wire were wound on, increasing its weight dramatically with each turn.
The hub assembly is hinged, so after removal of the four bolts that stiffen the structure the arms will fold in and leave an object only about 710mm long to carry, along with the folded up coil of wire
The wire is retained at the ends of the spokes by a small block of timber with a smooth curve across its face to prevent cutting of the wire. Another block is screwed on top to form a Y-shaped hook which effectively traps the wire in place and ensures it can't move while the spokes are deployed during bolting down of the spreaders.
The orthogonal spreaders are half-lapped at their cross-over point, and bolt to the side of the spokes opposite the hinges, pulling the joints rigid as they are tightened.
As little metal as possible was used in the construction. Non-ferrous metal (brass) was used in all screws and the hinges, but the four bolts were only available in zinc dipped steel. Stainless might be available in the future, I don't plan to leave the antenna outdoors very much, and the low reluctance of the steel bolts has not proved to be a detectable problem yet.
I have not constructed a preamp for the loop, but a low noise transimpedance amplifier should do the job. My first instinct was to plug an audio amplifier into the unit. I was greased by a loud mains buzz that I couldn't completely null. I wasn't expecting anything better, the office is near/in a shopping centre, located in the middle of a residential area, although the power lines are underground for the surrounding kilometre or so.
The heavy gauge of the winding wire should give an excellent low frequency corner, and enable a modest but stable gain to be built into the preamp, using only a single ultra-low noise opamp. This loop should easily see the Schumann resonances in quiet surroundings. Its response should extend into the LF and lower MF spectrum, probably to a MHz or so, I'll measure its self-resonance frequency some day.
Here is the raw spectra of the loop plugged straight into the microphone input of my computer's sound card, captured at its slowest 4kHz sample rate. As you can see, mains harmonics and my monitor's 85Hz refresh dominate. I am not sure if the wideband transients are local switching garbage or sferics, they sound a lot like sferics on the speaker. The occasional burst of noise on the upperside of all of the mains harmonics is curious, I have no idea what is causing that. The mains frequency itself (50Hz here) has AM modulation sidebands reaching down to near 20Hz. The lower corner frequency for the card input appears to be 18Hz which is much better than I expected, but its noise rises sharply, being only -60dB at 1Hz. I'll have to use a DC coupled ADC for lower, sub 10Hz, frequencies.
I got some really excellent recordings of the off-peak hot water signalling tones at around 1050Hz with this very basic initial lash up. I was looking around the 'net for information on them, but there is virtually no mention of them anywhere? Here is the picture from DC to 24kHz, the limit of the soundcard's sample rates. This is the integrated noise levels in each bin for a few minutes of signal.
The wide peak near 20kHz is of unknown origin, it isn't present with the input shorted, unlike the smaller pip at 23kHz which is some kind of noise or alias from the soundcard. The two close together peaks at around 18kHz appear to be FSK tones. The peak at 16kHz is local noise from a piece of telecom equipment. As you can see the mains garbage doesn't give me a break until beyond 10kHz, leaving little choice but to find a more quiet location.
The refrigeration motors next door in the shopping centre make pretty curves either side of the mains harmonics as the cut in and out. A vacuum cleaner makes several harmonically related squiggles but no visible sidebands on the mains? Car ignition noise from the road below has not been seen yet, it is probably still 20dB into the garbage, but should be resolvable (yeah, great!) with a loop preamp. A digital comb filter might be able to clean up the mains harmonics, but it can't get rid of the sidebands which squiggle all over the place. Maybe some more sophisticated DSP can, looking for correlations aliased around harmonics of a specified fundamental (ie a sideband killer)?
2 comments.