ELF Generator

[Jeg]

Hi to all
I am interesting in building a sinus generator in a bandwidth of 0-50 Hz with a step of about 0,01 Hz, and controlling it by a simple potentiometer.
The output voltage doesn't have to be over 2-3 Volt. Does anyone have any idea?
Thanks in advance

 

[Russlk]

Zero Hertz is very difficult to achieve! My Wavetek generator goes down to .01 Hz, so that circuit would be suitable and the control is a pot. You won't get frequency steps with pot control, you would need a frequency synthesizer for that. With pot control, the max range is about 10:1 so you would have .01 - .1 Hz, .1 - 1 Hz, 1 - 10 Hz, 10 - 50 Hz. ranges. You would not be able to reliably set the frequency to 50.01 Hz using pot control. What do you want to achieve?

 

[Jeg]

Thanks for the reply Russlk,

I think 1-10 Hz is fine! Can u help me on this? At least on the basics. I am experimenting on Schumman resonances which they are from 7,83 Hz the fundamental, up to 45 Hz harmonics. But a band from 1-10 Hz it works fine for me as it includes the band of Alpha waves, the one that interests me more!

 

[Russlk]

National Semiconductor Application notes have a lot of circuits, Here are some that will apply to your application:
AN-20, Fig. 24
AN-29, Fig. 18
An-31, Fig. tl/h/7057-29, this is the circuit used in the Hewlett-Packard model 200CD audo generator.
AN-31, Fig. tl/h/7057-34, this one shows the equation, you should use a dual pot. I recommend this one.
An-32, Fig tl/h/6791-21
Go to **broken link removed** and use the search.

 

[Sceadwian]

I would say just use a PC sound card, they could easily obtain significantly better resolution that .1hz at those frequencies. The only problem is most sound cards use a capacitor on the output to provide DC issolation. It'd be worth looking into finding a sound card that has a DC output or one that could be easily modified to have a DC output. A 44100 sampling rate would give you .01 hertz resolution at 44.1khz At 10 hertz it'd be closer to .001hz resolution. I don't know how much aliasing would occur though, sound cards aren't typically used to create wave forms that low.

 

[justDIY]

oh, ac, never mind

 

[Jeg]

Thanks a lot my friends for your replies!

I'll notify you of my final choice

 

[Jeg]

Russlk i simulated the circuit (AN-31, Fig. tl/h/7057-34), and after a very hard try, i managed to take an output of 30 Hz minimum. Its very difficult for going lower, and i am trying now to simulate the circuit (AN-31 section 2 Low frequency sinus generator with quadrature output). The problem is that i don't know how much voltage i have to supply and where to give it! Is there a specific way of calculating it?
A second question is about the output impedance of both circuits. How can we find it out? (I know this is a whole lesson, but few words would be very helpful)

 

[audioguru]

The output impedance of an opamp is its open loop output impedance divided by its internal gain, times its set gain: 75 ohms/200,000 x 3= 0.001 ohms.

The above is at DC. At 1kHz the internal gain is reduced by the frequency compensation. A TL071 opamp will have an internal gain of 31,000 at 1kHz. So in the oscillator circuit its output impedance is 0.007 ohms.

 

[Russlk]

The problem with simulating oscillator circuits is that the simulator does not have the noise that makes the physical circuit work, so I usually add a small voltage or current input to get it started. The frequency does not have to be the same as the oscillation frequency, I use something a little higher. You could use a damped transient input to get a clean output, eventually. The input can be at any opamp input node.

 

[Sceadwian]

I'm not sure of other simulators although I'm sure most have the capability, but LTSpice has the ability to take .wav files as an input to a voltage or current source. Using a custom data set and modifying the .wav file header information to reflect an altered sampling rate you can input (based on the 4 byte sample rate size of a .wav file) digital samples at a rate of about 4.2ghz's. I've never tested it to that extreme though, 190khz is the max I've tried to use, the simulation time and memory requirements at that sampling rate are enormous. I think I will test it though and see if it can get at least a few mhz in. Just thinking about how useful that would be, using a raw data set with a simple hand hacked .wav header from a DSO..

Edit: Apparently more than just the sampling rate needs to be modified, there are chunk sizes and a few other bits that need to be adjusted. I'll play around with it this weekend, out of my own curiosity I would like to see if you could compress a 10 second 44.1 whitenoise wav file down to a 4,410,000 100u second wav =) I know the .wav file header supports up to 4ghz but I'm not sure what LTspice's ability to handle .wav files is.

 

[Hero999]

Doesn't the sample rate need to be double the minimum frequency of interest?

So the maximum frequency a sound card sampling at 44.1kHz can generate is 22.05kHz.

My first question of course is why do you want to do this?

Frequencies below 20Hz are useless for audio.

Thre are plenty of ways of doing this.
https://www.electro-tech-online.com/custompdfs/2007/01/AN-263.pdf

Personally I would use a wein bridge oscillator with a wafer switch to select different capacitors and a dual gang pot to tune it.

 

[justDIY]

My first question of course is why do you want to do this?

Frequencies below 20Hz are useless for audio.

he's trying to build a brain-wave modifying device

my question is, what kind of transducer will you use to recreate these waves once you generate them? radiating 0.1 to 10 hz as RF will need one heck of an antenna, radiating as pressure waves will require a linear actuator driving some sort of membrane. I suppose radiating it as light waves would be the easiest - then you're talking led blinker

 

[Russlk]

An electromagnet would be the obvious way, but you would have to be careful about the intensity.

 

[Jeg]

Thank you all once again for your replies!

The main reason i would like to build this kind of gen. is for experimenting around the relation between Schumman resonances and brain's waves.
For the beginning i'd like to build a simple gen, and then i was thinking about an audio modulation in those frequencies. But i think the idea of justDiy with the use of leds, is simpler (equals best)!

Russlk said: "You could use a damped transient input to get a clean output, eventually...."

I am sorry Russlk but i didn't understand what do you mean by that. I supplied it with +/-15 dc at pins 7 & 4 but nothing. I also tried to give +15V at pin 7 and ground at pin 4 and reverse, but nothing again.

**broken link removed**

 

[Roff]

I don't know if this has been cited yet. National app note 263 has more sine wave generation techniques than you can shake a stick at.
I particularly like Fig. 10a. I think you could substitute any matched pair of NPNs for the LM394. You could either use a transistor array like CA3046, or you could match a couple of general purpose NPNs (e.g. 2N3904) by connecting base to collector and then checking forward voltage between collector and emitter with an ohmmeter or the diode function on a multimeter.

 

[Hero999]

Does it have to be a sine wave?

Why not a square wave?

It's unlikely that the harmonics will make any difference.

I think this is rubbish anyway, but that's just my opionion.

 

[justDIY]

a square wave would be a lot easier to deal with ... make a 1 to 10 hz wave with a 555, and then use a 4017 as a divide by ten

 

[Sceadwian]

I believe the rise and fall times of the waves are key to brainwave and Schuman resonance frequencies, IE true RF sine waves. Square waves on the same frequencies aren't even registered on most instruments that are able to receive them because of the rise and fall times. The rise and fall time of a 555 is what, 10u seconds? That's 100khz. Even if the ultimate frequency the 555 is producing is only 10hz, the transitions occur at 100khz, no reception occurs.

 

[Roff]

I believe the rise and fall times of the waves are key to brainwave and Schuman resonance frequencies, IE true RF sine waves. Square waves on the same frequencies aren't even registered on most instruments that are able to receive them because of the rise and fall times. The rise and fall time of a 555 is what, 10u seconds? That's 100khz. Even if the ultimate frequency the 555 is producing is only 10hz, the transitions occur at 100khz, no reception occurs.

I don't understand what you are trying to say. The fundamental frequency of a 10Hz square wave is, not surprisingly, 10Hz. The amplitude of the Nth harmonic is 1/N times the amplitude of the fundamental. The nearest harmonic to 100kHz will be less than 0.0001 times the amplitude of the fundamental.
I only know what I read in the Wikipedia entry on Schumann resonances, so forgive me if I'm missing your point completely.