# DIY Random Number Generator : making a series of 0s and 1s out of a random electrical signal

Discussion in 'The Projects Forum' started by Refrigirators, Mar 13, 2016.

1. ### Refrigirators Thread Starter New Member

Mar 13, 2016
2
0
Hi,
First I'd like to apologize in advance if my communication skills are imperfect because English isn't my native language. If you don't understand what I'm saying, feel free to tell me and I will try to clarify as much as possible.

My project for my studies consists in creating a series of random numbers out of a random electrical signal. The first step is to make an actual random electrical signal (electrical noise) that can't be predicted. Next, I have to somehow transform it into a series of random numbers. That's about it.

STEP 1 : Creating a random electrical signal
What I did is I created an electric breakdown between two conductors :
What I've learnt (https://en.wikipedia.org/wiki/Breakdown_voltage, https://en.wikipedia.org/wiki/Shot_noise) is that the electrons during breakdown are actually randomly going from one conductor to the other through the air space between them. I placed them about 1 inch apart.
This is the output I've manage to get on the oscilloscope : (I don't have access to the circuit right now so I'm posting what I remember)

DESCRIPTION : As you can see, there is a series of impulses, separated by the same time step, all with amplitudes varying between 0 and a maximum value. IMPORTANT NOTE : in fact this graph is ALWAYS changing : every fraction of a second, the amplitudes would be different (in a random way) from the previous amplitudes, changing all the time. In other words, what I've drawn on the image you can see is what could be seen during a fraction of a second. During the next fraction, the graph would be different, having still impulses at the same absciss but the amplitudes would constantly be changing.

So yeah, this signal seems pretty random. Which is exactly what I need.

STEP 2 : Using this random signal to make a series of random numbers
This is where it gets tricky for me. I'm a beginner in electronics, so I don't really understand how to do this part.
My teacher gave me a "SHIFT REGISTER" and even after reading the shift register wikipedia page, I still don't quite understand what it is supposed to do or even with the datasheet, I'm so confused because there are several inputs and outputs and combinatoric knowledge is required (it seems).

I found this useful article explaining exactly this, but it seems unnecessarily complicated if you look at the circuit schematics. http://makezine.com/projects/really-really-random-number-generator/
Can you please explain, with simple words, what should be done to transform this specific signal into a series of digits that I can somehow import on a computer ? (The digits can be 0s and 1s for example, and I can code them into decimal.)

2. ### dannyf Well-Known Member

Sep 13, 2015
2,196
417
The site you linked to goes into pretty details to show you how it works. If that still doesn't work for you, it is quite hard to make it any easier.

In general, it is either the magnitude or phase information that you use to randomize

If there is a mcu involved, it is much easier hardware wise.

3. ### Alec_t AAC Fanatic!

Sep 17, 2013
5,973
1,135
A high voltage arc device, such as you show, isn't necessary. You could use the reverse breakdown of a semiconductor junction instead.

4. ### Bernard AAC Fanatic!

Aug 7, 2008
4,240
414
This random pulse generator has been working for many months & I have not yet found a pattern. In the past using similar circuits was surprised to see fixed patterns develop, to stop this learning process I added a small variable signal to pin 5 of 555 derived from a flickering
candle LED. Call it Firefly in a Bottle.

5. ### Wendy Moderator

Mar 24, 2008
20,772
2,540
Zeners make decent white noise from what I've heard. I am interested in such a project myself.

True white noise is a true random signal. The trick is proving the random, by doing a bell curve of the generated numbers over time.

When I did S/N on radios we had to use a white noise source on the RF input, notch specific frequencies, then see how much the amp under test filled in the notches.

6. ### #12 Expert

Nov 30, 2010
16,703
7,338
I accidentally met that when designing a bias supply. A little capacitance fixed my problem.
It was (2) 1N4004 diodes in series, forward biased with just a few ma to get a fairly stable -1.2V supply.