Zener diode noise spectral density

[Thomas Philips]

On looking at the specifications of various zener diodes, I see that
they tend to have a noise spectral density of about 50 nV/swrt(Hz) at
1 mA. I have seen it suggested that the noise DECREASES as the zener
current increases. Four questions:

1. Is this true, and if so why?
2. What is the floor on noise spectral density and what determines it?
Does it lie at the point at which shot noise in the bias current
times zener dynamic impedance = zener noise voltage?
3. Is there a 1/f region, and does the corner frequency rise as the
current increases?
4. What are typical 1/f corner frequencies?


There are clearly implications for low noise power supplies. Thanks in
advance.

Thomas Philips

 

[Winfield Hill]

Thomas Philips wrote...

On looking at the specifications of various zener diodes, I see that
they tend to have a noise spectral density of about 50 nV/swrt(Hz) at
1 mA. I have seen it suggested that the noise DECREASES as the zener
current increases. Four questions:

1. Is this true, and if so why?

If you read my postings and others in the zener-diode oscillation
thread of a few years back here on s.e.d., it will become clear to
you why this is so, and the answers to some of your other questions
may become more clear as well.

Thanks,
- Win

whill_at_picovolt-dot-com

 

[Thomas Philips]

Win,

Thanks for the pointer. I looked up the thread, and found it very
informative on high voltage (i.e >7V) zeners. It is mentioned by more
than one participant that the microplasma mechanism does not seem to
exist for low voltage (i.e. <7V) diodes, and it is this range that my
interest lies. On looking through various datasheets, I find that low
voltage zeners typically have a noise spectral density of about 50
nV/sqrt(Hz). This can clearly be filtered to improve the noise, but
I'm curious as to the improvement that can be made by increasing the
current, and I have found essentially no information on this. Can it
brought down to 10 nV/sqrt Hz? 5? 2? If so, at what current? Does this
increase 1/f noise? If so, what is the functional relationship between
corner frequency and current?

Sincerely

Thomas Philips

 

[John Larkin]

Win,

Thanks for the pointer. I looked up the thread, and found it very
informative on high voltage (i.e >7V) zeners. It is mentioned by more
than one participant that the microplasma mechanism does not seem to
exist for low voltage (i.e. <7V) diodes, and it is this range that my
interest lies. On looking through various datasheets, I find that low
voltage zeners typically have a noise spectral density of about 50
nV/sqrt(Hz). This can clearly be filtered to improve the noise, but
I'm curious as to the improvement that can be made by increasing the
current, and I have found essentially no information on this. Can it
brought down to 10 nV/sqrt Hz? 5? 2? If so, at what current? Does this
increase 1/f noise? If so, what is the functional relationship between
corner frequency and current?

Sincerely

Thomas Philips

If you want a nice 10 nv/rthz noise source, just use a 6K resistor.

John