questions on zener diode selection (for noise generation project)

[kubuni]

I found the following link for generating a random signal from a reverse bais zener diode.
http://www.ciphersbyritter.com/NOISE/NOISRC.HTM

How can  I calculate or get the min current going through a zener diode for an apply voltage (e.g 9V)?

Also, would there happen to be any zener diodes with a smaller max reverse voltage bais?

[Benta]

IIRC, 6.2 V Zeners are preferred for noise generation, as there exists a 'sweet spot" here between Zener and avalanche operation, which provides the most linear noise spectrum.

Your link is a bit strange, as an LM336 is used as "zener", which it is not (it's a bandgap reference). The LM336 is not known for being noisy, exactly the opposite, in fact. The circuit would probably operate just as well measuring the noise from a resistor...

[kubuni]

Would there happen to be a better example of a noisy zener circuit I could use?

[Benta]

Lots of them around the web, search for "zener noise source".

Here's an example:
http://www.radiohobbyist.org/blog/?p=1884

[Zero999]

I was under the impression that higher voltage zener diodes are generally more noisy than lower voltage ones. I know that 6.2V is best for temperature drift but wasn't aware that it gave a better noise spectrum too.

There is a sweet spot, as far as reverse current vs noise is concerned. At very low currents, there will hardly be any nose and a much higher currents there's less noise too. Perhaps someone else here will know more about this?

[MagicSmoker]

The best zener for generating noise? Probably a reverse biased emitter-base junction of an NPN BJT. The breakdown voltage for this junction is in the range of 5-7V and the highest noise output is usually around 10uA. That is to say, noise energy content tends to decline on either side of the ideal current, though the value for "ideal" can be quite broad.

[VE2UM]

...Reverse-biased BJT B-E  junctions make nice noise generators for experimenting but a B-E junction is NOT designed to be reverse-biased onto avalanche conduction and is absolutely NOT guaranteed to work on a long term use.

https://electronics.stackexchange.com/questions/289058/are-reverse-biased-transistors-stable

On their side, Zeners / Avalanche diodes are designed for long term reverse conduction.

On a chinese noise source i bought on Fleabay, the diode is a 24volt device  biased at 3.3mA (3k3 resistor fed at 35V from a bost DC-DC converter).

[David Hess]

I agree with evb149.  If you want a low voltage noise diode, then use a 3 volt Vbe RF transistor.  But a common 2N3904 or similar small signal transistor is very good because of its low base-emitter capacitance.

VE2UM, base-emitter breakdown damages the junction but the same applies to diodes.  The damage is only significant for operation as a transistor.  Reliably is not affected as long as the transistor is not driven beyond its capability.

I was under the impression that higher voltage zener diodes are generally more noisy than lower voltage ones.

High voltage zener diodes are actually avalanche diodes.  Low voltage zener diodes use the zener effect.  They balance out at about 6.2 volts where their temperature coefficients cancel.

I know that 6.2V is best for temperature drift but wasn't aware that it gave a better noise spectrum too.

I do not know that it does give a better noise spectrum.

There is a sweet spot, as far as reverse current vs noise is concerned. At very low currents, there will hardly be any nose and a much higher currents there's less noise too. Perhaps someone else here will know more about this?

Zener impedance is lower at higher currents where the voltage noise is less.  But the lower impedance drives the junction capacitance better yielding better high frequency response.  In practice it is easy enough to vary the drive current for maximum output noise.