Notch Filter Issues

The natural ELF/VLF radio signals I want to receive are essentially those of the audio frequency range (perhaps going a little lower, so call it 1-20kHz). If you act as an antenna yourself and plug yourself into an audio amplifier by putting your thumb on the input lead, the signal you will hear above anything else is mains hum (50Hz Europe & Australia, 60Hz US). The natural signals are way weaker than this, so if I want to use the receiver even remotely near power lines, I need to get rid of that hum.

While it would be straightforward to cut 50Hz using a digital filter, the level is so relatively high that it will swamp the desired signal going through an A/D filter leaving little useful resolution. There’s also a good chance of it saturating any analog pre-amplification.

It’s worth noting here that the mains hum tends to be pretty dirty, with lots of harmonics (2x50Hz, 3x50Hz etc.). According to Radio Nature the big ones are the odd harmonics. But for now I’ll just try cutting the fundamental, see if that’s enough.

A related issue is that the natural radio signals are of such a low amplitude that noise generated by the receiver circuit components may also be an issue.

So I’m thinking, whether I use a coil (for the magnetic component of the natural signals) or a free antenna (electrical component), I should first have a little fairly wideband but low noise pre-amp, maybe x10 or x100 to minimise noise addition later. Have this followed by a notch filter and then further amplification to get the signal up to line levels.

The standard simple filter is the twin-t notch, more or less a low pass and a high pass filter bang up against each other. But the notch with this isn’t very sharp. This can be improved by throwing in a couple of op-amps, as in this circuit :

So I tried this on the breadboard, only with approximate values (33k rather than 31.8, 5% capacitors).

This didn’t appear to be working, with white noise as input, the scope showing freq domain trace (top, marker line is 50Hz) :

So then I just tried the passive twin-t, and got a similar result 😦 I guess the noise input/freq domain on scope is a losing combination.

The BitScope has a basic signal genenerator built in, so I tried that on the passive circuit:

Here there is a distinct difference between the input (green) and output (yellow) at 50Hz compared to other peaks.

Reassembling, I tried the active circuit with the sig gen :

Boo, no significant difference. But I strongly suspect this is down to measurement fail again. The sig gen is very limited, only offers 50Hz, 100Hz etc, and even if the components in the filter were perfect, the notch would be at 48Hz. Allowing for tolerances, it could be way out. But assuming the active circuit is working, there could be a sharp notch at say 45Hz, relatively flat at 50Hz.

So next step, I reckon I’ll knock together an analog sig gen, hopefully get a clearer view.

Getting Started

I have been thinking about and casually researching for this project a few months now. But today I started fiddling with some hardware, so it seemed a good point to start a blog on it.

I’m currently in St.Arnaud near Melbourne, a long way from my base in Italy. A couple of months ago I ordered a bunch of electronic components to get started playing with the analog parts of the system, starting with an ELF/VLF receiver. Of those the only thing that arrived before I set off over here was a Bitscope gadget to allow a computer to be used as an oscilloscope and/or data capture. Coincidentally that shipped from Australia, arrived in good time, unlike the stuff from Farnell in Europe.

I didn’t bring any other kit with me, but my darling Raven took me to a component shop the other day (Jaycar) where I picked up some bits & pieces.

I had planned to try some of the VLF receiver circuits in Radio Nature, which seems like the bible on this stuff. The ones I was looking at have a 2N3819 FET at the front end (the circuits can be found around vlf.it, the site of the book’s author). Unfortunately Jaycar didn’t have these, but gave me what they said was an equivalent. It wasn’t – just a regular bipolar. But I’d also bought a few TL084 quad jfet op-amps, so I should be able to get something together that’s functionally equivalent, if perhaps a bit noisier.

Here’s my prototyping setup:

Bottom right is a bit of scrap wood with hot-glued to it: 2xPP3 batteries; a speaker; breadboard; Bitscope.

The Bitscope’s hooked up to an old laptop of Raven, running Ubuntu. My first try with this, doesn’t seem like the signal’s getting through and their site just happens to be down today. Grr!

Still got things to play with though. I want to get the sound of VLF coming out of the speaker, and I want it to be portable to get away from power lines at first. So I got an LM386 250mW amp chip, as used in some Radio Nature circuits. Haven’t played with one of these before – I’d have remembered, it’s atrociously unstable. The basic (inverting) configuration from the data sheet oscillates happily. But I was able to get something more stable by tweaking the non-inverting config (in data sheet as “AM Radio Amplifier”).

Hopping ahead a little on the breadboard I’ve also got a white noise generator. I’m going to make a 50 Hz active (bootstrapped) notch filter to rid the signal of the worst of mains hum, hopefully allowing the receiver to be used in (or near) the house. I should be able to tweak the notch when the Bitscope’s working, with noise as input.

PS. Not long after posting this I remembered archive.org and found the BitScope troubleshooting tips. My problem was just permissions on /dev/tt1USB (?). So I had a crack at the notch filter. No joy with that yet, but my problem could well be inexperience with the ‘scope, there are a lot of controls.