Here is the schematic for noise (Quantum Random Number Generators – Scientific Figure on ResearchGate. Available from: https://www.researchgate.net/figure/Conceptual-representation-of-a-typical-noise-based-random-number-generator-The-voltage_fig6_301899096 [accessed 14 Oct 2024])
Made a random nibbler board based on this circuit :
Here is the board sent to Aisler :
Here assembled :
This is the special hex decoder chip we ordered from ebay :
So far I’ve got something around 200mV of noise. I’ve tried to place that smack in the middle of the 2.5V CMOS level but it remains stable and doesn’t flick back and forth.
I’ve learned that there are challenges with hardware noise, they are sensitive to power supply ripples and to neighbouring sources of electromagnetic radiation.
*****
From pp 558 and starting at 974 AoE
Digital noise with XNORs and big shift registers (64 bits and 8 bits) several bit positions of which that are fed back (called “taps” ) from to the inputs, then low pass filtered and amplified. A register with n bits goes through 2^n -1 states before repeating. It must be shifted higher than the frequency you sample the noise from.
https://en.wikipedia.org/wiki/Linear-feedback_shift_register
The feedback polynomial is determined by the taps chosen (the output is pin 16) :
(1 is because x^0 = 1)
How cool would it be to let people pick the seed, change the taps and the total number of bits operating in the register, and the clocking and sampling frequencies, then to represent the noise on a 2D black and white plane !
******
Other ideas (https://www.mdpi.com/2410-387X/7/2/26) seem to involve multiple ring oscillators :
Here is another article (https://www.mdpi.com/1099-4300/23/9/1168) trying to describe the random using this technique :
********
Vintage hex decoder doesn’t work…
This works though and lets people reconfigure their circuit :
The 8 bit shift register could be replaced with a 16 or more bit register.