Monitoring a 24 kHz frequency, logging a digital dB reading

[mishima]

Hi, I had a project in mind but I'm not sure its possible, or what I might need.

Current Setup:
A small loop antenna monitors a 24kHz (VLF) radio signal, and passes it to an amplifier (schematic can be found here: https://www.physicsforums.com/threads/schematic-symbol-help-supersid.709128/). This then goes into a sound card with a 96k sample rate, and software picks it up and plots a frequency spectrum and a dB vs time graph.

Change Desired:
I am very much wanting to eliminate the sound card and computer requirement, and move this to a small micro-controller circuit (that stores the information on an SD card).

Known:
-I need an analog to digital converter with a minimum 60k sample rate.
-I don't care so much about the frequency spectrum plot, and want to just concentrate on dB vs time. Not sure if that can eliminate some electronics or not.

As mentioned I'm not really sure, on a block diagram level, what I need to accomplish here. I would be building this myself and am confident in everything except for this initial signal processing.

 

[Tom.G]

https://www.google.com/search?source=hp&q=microprocessor+with+adc

 

[mishima]

Is that really it? So just something like an attiny85 with a 10bit 200kHz sample rate? No other peripherals required?

edit: for example, how would it know how to pick out the 24kHz signal from the full spectrum?

 

[Tom.G]

I would be building this myself and am confident in everything except for this initial signal processing.

The schematic you linked in Post #1 looks reasonable as the initial signal processing.

I am very much wanting to eliminate the sound card and computer requirement, and move this to a small micro-controller circuit (that stores the information on an SD card).

Google search suggested in post #2 and you picked a possible solution in post #3.

-I don't care so much about the frequency spectrum plot, and want to just concentrate on dB vs time.

edit: for example, how would it know how to pick out the 24kHz signal from the full spectrum?

As you stated in Post #1, specifically:

and software picks it up and plots a frequency spectrum and a dB vs time graph.

I would be building this myself and am confident in everything except for this initial signal processing.

Did I misunderstand your willingness to write the software? Or you could do it in hardware (probably easier), resonate the loop antenna to 24kHz (Google 'resonate circuit') and in the amplifier, change C5, 68pF to 33pF (or maybe 47pF).

Is that really it? So just something like an attiny85 with a 10bit 200kHz sample rate? No other peripherals required?

Depends on what you want for output. an LED, a text display, interface to an SD card, graphics display, ability to connect to a larger computer for control or data transfer. That particular microcontroller looks way underpowered for most of that. I would recommend an assembled single board computer that has support for the interfacing you are after. The Arduino and the Raspberry Pi are quite popular. However never having used them, I know nothing about them. Seems worth investigating though.

Good Luck... and have fun!

 

[Baluncore]

You will need a loop antenna tuned to 24kHz with a parallel capacitor.

You will need to find a microcontroller that has A to D converter and the ability to log data to SD memory.

There are many Logarithmic or Limiter Amplifiers with Received Signal Strength Indicator, RSSI, outputs that will do all the RF for you. For low frequencies select one that goes to DC.
Examples from Analog Devices are;
AD8307. DC to 500 MHz, 92 dB Logarithmic Amplifier.
AD8310. Fast, Voltage-Out, DC to 440 MHz, 95 dB Logarithmic Amplifier.

 

[Baluncore]

Consider an Arduino Trinket $8, with a Micro SD Card Breakout Board, $8.
https://learn.adafruit.com/adafruit-micro-sd-breakout-board-card-tutorial/intro
The Adafruit Trinket has a 12 bit AtoD converter.

 

[mishima]

Thanks for the responses, I was definitely misunderstanding the role of the sound card and the role of the software. I was erroneously thinking software was just plotting graphs, and sound-card was doing frequency selection.

I'll tune the antenna and use an RSSI (didn't know those existed).

 

[Svein]

You will need a loop antenna tuned to 24kHz with a parallel capacitor.

A 24kHz tuned loop antenna... the wavelength of 24kHz is 12 500m!

My suggestion: Create a 24kHz reference yourself, multiply the incoming signal by this reference and use a low-pass filter on the output. A received signal of 24kHz (and multiplies thereof) will give a DC output different from 0V. I seem to remember that Motorola (now On semi) had an IC based on that principle, used for detecting the 19kHz sideband of an FM stereo broadcast. Check out http://www.circuitstoday.com/stereo-decoder-circuit.

 

[mishima]

A 24kHz tuned loop antenna... the wavelength of 24kHz is 12 500m!

Good point, this is a small loop antenna that only uses the magnetic part of the spectrum as I understand it. So I can get away with antennas on the 1 or 2 meter scale. Not sure how/if that affects tuning...I would guess it wouldn't?

And good idea, definitely need to make a 24kHz source to test with.

So you are saying the low-pass filter setup has the advantage of not having to tune the antenna? That's interesting, because usually its my students who design/build the antennas and tuning would be different for each one.

 

[Svein]

So you are saying the low-pass filter setup has the advantage of not having to tune the antenna? That's interesting, because usually its my students who design/build the antennas and tuning would be different for each one.

No!
What I was saying was:

1. There is no way you can have a tuned antenna at 24kHz (it would be impossibly long)
2. Trying to construct a tuned circuit for 24kHz is hard - just check out the size of the components you need!
3. There is a way around the dilemma - do it in another way. What I described is a simple version of a phase detector.

Expensive and complicated versions of such detectors exist. They are called lock-in amplifiers (Google it!)

 

[Baluncore]

There is no way you can have a tuned antenna at 24kHz (it would be impossibly long)

The wire in the loop antenna is not λ/2 = 6.25 km long. The wire makes an inductor, not a dipole.

A 1m by 1m square loop can be orientated so it is a directional antenna. It would have a single turn loop inductance of 5.662 uH. If the parallel tuning capacitance is 10 nF, then you will need about 25 turns on the loop to resonate at 24 kHz. Wire length will be only 100 metres.

 

[mishima]

This is very exciting.

In the unchanged setup with the sound card and computer, we've had success detecting solar flares (it monitors the submarine time signals sent from Cutler, Maine bounced off the ionosphere, which change dramatically during flares) with ~1m square loops, untuned, and only 15 m (number of turns varies with design). The signal to noise for certain constructions has occasionally been good enough to detect C-class flares (M and X have never been a problem, regardless of design...we verify with NOAA's solar flux data).

So in other words, we've been way off for years! Perhaps with these modifications (additional length and tuning) we can reach B or A.

it is a directional antenna

That is very true, if we are even a few degrees off alignment, the signal disappears.

 

[Baluncore]

That is very true, if we are even a few degrees off alignment, the signal disappears.

The response of a small loop antenna is a dipole pattern. There will be two deep but narrow nulls when the axis of the loop points towards the source. The receive lobes will be quite wide. If you rotate the loop to place the worst interference source accurately in a deep null, you may then get quite a good signal to noise ratio.

 

[mishima]

About setting the sample rate. The attiny85's (heart of the trinket) listed rate is 200kHz which sounds great for my 24kHz signal at first glance.

But another look through the datasheet...there's also ADC conversion time. For "normal" conversions, its 13 cycles. I see its also possible to set a prescaler value. How do all these things come together to be sure I have at least a 60kHz sample rate?

edit: Like, if I set the prescaler to 2 and was using the default 8MHz clock...the ADC would use 4MHz and my true sample rate would be 4M/13=~308k?

 

[Baluncore]

How do all these things come together to be sure I have at least a 60kHz sample rate?

The RSSI output can be low-pass filtered to change very slowly. You do not need high speed conversion because you do not need to do an FFT of the data to produce a spectrum and estimate of the 24 kHz amplitude.
The tuning is done by the loop antenna. The 24 kHz detection and logarithmic conversion is done by the Log-Limiter-Amp chip. The trinket then converts and records that analogue amplitude in dB, maybe every second or tenth of a second. The trinket then writes that data to the SD card as it is collected. The trinket will probably be in sleep mode for most of the time, waiting for the next conversion time.