2
\$\begingroup\$

As a learning exercise I'd like to try creating a random number generator using reverse-biased BJTs. As I understand it, if the emitter is saturated in reverse bias, occasionally electrons will tunnel through the band gap to the base pin.

  1. Can this be done with common BJTs or are special parts required?
  2. What sort of amplification will be required to actually detect this tunneling? I.e. at what voltage will the signal be on the base pin when an electron tunnels through?
\$\endgroup\$
1
\$\begingroup\$

It's easily done. Reverse-biasing the emitter-base junction of an NPN transistor is well-known as a cheap and fairly good noise generator. See http://holdenc.altervista.org/avalanche/ as an example, but if you Google reverse bias noise generator you'll get lots of hits.

And the problem is not so much amplifying the random electrons as it is reducing the number to a level where counting single electrons is possible. Actually, this isn't done. Instead, the standard approach is to AC-couple the noise voltage and look for zero crossings of the detected signal.

\$\endgroup\$
  • \$\begingroup\$ Thanks. "noise generator" was the search term I needed. Can you elaborate on what is involved in ac-coupling the noise voltage? \$\endgroup\$ – nathanvy May 12 '15 at 23:40
  • \$\begingroup\$ You simply put a capacitor in series between the noise source and the load. It's also called a high-pass filter. The result is that the signal average becomes zero. That is, the DC component of the noise is removed, leaving only the AC component. \$\endgroup\$ – WhatRoughBeast May 13 '15 at 0:39
0
\$\begingroup\$

What you're describing is called a Zener Diode. You probably can't get appreciable tunneling current out of a common BJT though, because the doping levels are so low the depletion regions will be large, giving low tunneling current. Whatever tunneling you do get will be swamped out by thermal noise.

People have made random number generators out of Zener diodes. To really get the tunneling current to be significant you need to increase the reverse bias and enter a region called "Zener Breakdown". A standard BJT will enter Avalanche Breakdown (the energy of free electrons becomes high enough to liberate more electrons) before it gets close to Zener Breakdown.

Here is a link to someone who made a random number generator using Avalanche Breakdown: Avalanche Breakdown RNG The design for Zener Breakdown would be similar (but you would need a Zener diode).

Good Luck!

\$\endgroup\$
  • \$\begingroup\$ A bipolar junction transistor is two diodes back to back. (you can't construct a transistor by connecting two real diodes because the reason that a transistor works is because the base region is so thin, the interface between the back to back transistors). A diode will breakdown if reverse biased, a zener diode is the same thing as a diode just that the zener is a) reverse biased and b) designed to breakdown in reverse bias at a much lower break down voltage than a diode. So I think it's not really true to say that "what you're describing is a zener". Zeners are just diodes. \$\endgroup\$ – Dean May 15 '15 at 18:18
  • \$\begingroup\$ I have seen probably 25 years ago a design in an electronics magazine which generated noise using a BJT, using these principles, so it can be done with BJTs. Perhaps a zener might be an easier way to achieve it. \$\endgroup\$ – Dean May 15 '15 at 18:22
  • \$\begingroup\$ @Dean no offense intended but I think you're a bit confused about different diodes. A Zener diode is a diode that undergoes Zener breakdown which is a tunnelling effect. A standard diode is a diode that undergoes avalanche breakdown well before the Zener effect kicks in. When you say your point b) about "designed to breakdown..." the way they do that is to increase the doping level which increases the tunneling current at lower reverse bias --> Zener diode. Which is exactly what I said. You basically described the difference between a Zener and standard, then said they are the same! \$\endgroup\$ – crgrace May 15 '15 at 21:05
  • \$\begingroup\$ @Dean so my point stands. A Zener diode breaks down due to the Zener effect. A standard diode breaks down due to avalanche breakdown. As for the point about noise, it is much simpler to generate noise using avalanche breakdown, but the OP brought up tunnelling so I suggested a Zener diode. \$\endgroup\$ – crgrace May 15 '15 at 21:06
0
\$\begingroup\$

Tunnelling isn't to do with electrons passing across the bandgap. That's electron excitation. Let's take Silicon (because that's one material I can remember the figures for): bandgap is 1.1ev. The bandgap is the difference in energy levels between the bottom of the conduction band and the top of the valence band. So to move the electron from valence band into the conduction band requres the electron receive a minimum extra energy of 1.1 eV. Tunnelling is to do with left-right movement (not vertically) which is based on probability and probability wave functions, which define the position of the electron.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.