# Dumb question on noise source

Started by January 11, 2018
```My test gear (along with most of my brain) have been packed away, so I can
only ask this as a question rather than lash it up and know.

I have a +/- 5v source and a 6.8v zener diode as a noise source. I'm
expecting about 0.3mV noise. My question is: will I get more noise by
biasing with a resistor (about 3.3k) or biasing with a constant current
source. TIA

--
I look forward to the day when a chicken can cross the road without having
its motives questioned.
```
```On Thursday, January 11, 2018 at 11:47:47 AM UTC-8, David Eather wrote:

> I have a +/- 5v source and a 6.8v zener diode as a noise source. I'm
> expecting about 0.3mV noise. My question is: will I get more noise by
> biasing with a resistor (about 3.3k) or biasing with a constant current
> source. TIA

The zener won't care, but a 'constant current source' might be
bandwidth-limited with an internal op amp.   That would rather
complicate the noise spectrum; resistor (noninductive, metal film) would
be preferable.
```
```David Eather wrote...
>
> I have a +/- 5v source and a 6.8v zener diode as a noise source. I'm
> expecting about 0.3mV noise. My question is: will I get more noise by
> biasing with a resistor (about 3.3k) or biasing with a constant current
> source. TIA

"zener" diodes are actually field-effect diodes below about
5 volts, and true avalanche diodes, with short microplasma
breakdowns, above 7 volts.  In the transition region, zener
diodes have low Rac, low dV/dI, low noise and other "good"
properties.  So 8 volts or higher is a better voltage for
high noise generation.  The noise is related to current,
whatever the source.  You have to empirically determine
the best parameters, and the diodes are all different.

--
Thanks,
- Win
```
```On 01/11/2018 02:47 PM, David Eather wrote:
> My test gear (along with most of my brain) have been packed away, so I
> can only ask this as a question rather than lash it up and know.
>
> I have a +/- 5v source and a 6.8v zener diode as a noise source. I'm
> expecting about 0.3mV noise. My question is: will I get more noise by
> biasing with a resistor (about 3.3k) or biasing with a constant current
> source. TIA
>

The current source will need to have very large gain to be effective,
because if the zener is operating in the avalanche region (where you
want it to be to get the most shot noise) it's operating in the steepest
part of its I/V curve. A simple one or two transistor current source
might not be that much better than a resistor, maybe worse depending on
bandwidth.

Reverse-biased BJTs apparently make better noise sources at low voltages
than regular Zeners; well that's what everybody else seems to use, anyway.
```
```Am 11.01.2018 um 22:24 schrieb bitrex:

> Reverse-biased BJTs apparently make better noise sources at low voltages
> than regular Zeners; well that's what everybody else seems to use, anyway.

Probably you can expect less shunt capacitance for a BFR93 BE junction
than for the average Z-Diode. That type actually works quite good to
2 GHz, at least for me.

Gerhard
```
```Am 11.01.2018 um 22:16 schrieb Winfield Hill:
> David Eather wrote...
>>
>> I have a +/- 5v source and a 6.8v zener diode as a noise source. I'm
>> expecting about 0.3mV noise. My question is: will I get more noise by
>> biasing with a resistor (about 3.3k) or biasing with a constant current
>> source. TIA
>
>   "zener" diodes are actually field-effect diodes below about
>   5 volts, and true avalanche diodes, with short microplasma
>   breakdowns, above 7 volts.  In the transition region, zener
>   diodes have low Rac, low dV/dI, low noise and other "good"
>   properties.  So 8 volts or higher is a better voltage for
>   high noise generation.  The noise is related to current,
>   whatever the source.  You have to empirically determine
>   the best parameters, and the diodes are all different.

Where the TC goes through 0 at 6V or 7V , both effects are
"equally strong". You can see the onset of avalanching already
at low voltages. See for the BZX84 family, how the voltage noise
gets more from 2V7, 3V3, 4V7, 6V8. That grows faster than the
Zener voltage itself. All diodes were powered with 1 or 2K from
a 14V NiMH battery. The battery adds nothing, as has been
tested by using a 47 Ohm resistor instead of the Zener. It shows
just the expected thermal noise. 0 dB is 1nV/sqrt Hz.

<
https://www.flickr.com/photos/137684711@N07/24411798996/in/album-72157662535945536/

>

@David: A 3K3 resistor should do no harm. The noise depends on the
current, more current is usually quieter with wider bandwidth. I have
not done enough tests to be statistically convincing.

Gerhard
```
```On Fri, 12 Jan 2018 05:47:32 +1000, "David Eather" <eather@tpg.com.au>
wrote:

>My test gear (along with most of my brain) have been packed away, so I can
>only ask this as a question rather than lash it up and know.
>
>I have a +/- 5v source and a 6.8v zener diode as a noise source. I'm
>expecting about 0.3mV noise. My question is: will I get more noise by
>biasing with a resistor (about 3.3k) or biasing with a constant current
>source. TIA

Shouldn't matter. The zener dynamic impedance will be a lot less than
3.3K.

You might expect something like 300 nV RMS per root Hz of measurement
bandwidth.

--

John Larkin         Highland Technology, Inc
picosecond timing   precision measurement

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

```
```Winfield Hill wrote:
> David Eather wrote...
>>
>> I have a +/- 5v source and a 6.8v zener diode as a noise source. I'm
>> expecting about 0.3mV noise. My question is: will I get more noise by
>> biasing with a resistor (about 3.3k) or biasing with a constant current
>> source. TIA
>
>   "zener" diodes are actually field-effect diodes below about
>   5 volts, and true avalanche diodes, with short microplasma
>   breakdowns, above 7 volts.  In the transition region, zener
>   diodes have low Rac, low dV/dI, low noise and other "good"
>   properties.  So 8 volts or higher is a better voltage for
>   high noise generation.  The noise is related to current,
>   whatever the source.  You have to empirically determine
>   the best parameters, and the diodes are all different.
>
>
Correct; even the same1N between various manufacturers, or the same
1N and manufacturer but different packages and/or power rating.
Multiply those variations for a given voltage by the number of
voltages you wish to characterize, and take a loan on the house!

```
```On Fri, 12 Jan 2018 06:40:48 +1000, whit3rd <whit3rd@gmail.com> wrote:

> On Thursday, January 11, 2018 at 11:47:47 AM UTC-8, David Eather wrote:
>
>> I have a +/- 5v source and a 6.8v zener diode as a noise source. I'm
>> expecting about 0.3mV noise. My question is: will I get more noise by
>> biasing with a resistor (about 3.3k) or biasing with a constant current
>> source. TIA
>
> bandwidth-limited with an internal op amp.   That would rather
> complicate the noise spectrum; resistor (noninductive, metal film) would
> be preferable.

ok - and it simplifies things. A cc source might better deal with noise
and fluctuatons on the power supply?

--
I look forward to the day when a chicken can cross the road without having
its motives questioned.
```
```On Fri, 12 Jan 2018 07:16:19 +1000, Winfield Hill
<hill@rowland.harvard.edu> wrote:

> David Eather wrote...
>>
>> I have a +/- 5v source and a 6.8v zener diode as a noise source. I'm
>> expecting about 0.3mV noise. My question is: will I get more noise by
>> biasing with a resistor (about 3.3k) or biasing with a constant current
>> source. TIA
>
>  "zener" diodes are actually field-effect diodes below about
>  5 volts, and true avalanche diodes, with short microplasma
>  breakdowns, above 7 volts.  In the transition region, zener
>  diodes have low Rac, low dV/dI, low noise and other "good"
>  properties.  So 8 volts or higher is a better voltage for
>  high noise generation.  The noise is related to current,
>  whatever the source.  You have to empirically determine
>  the best parameters, and the diodes are all different.
>
>

more than 8 volts and  up the current. gotcha.  I was planing on a vca810
or for a low speed version a lm13700 as a variable gain stage. The worst
possible outcome would be trusting the end user to make adjustments.

--
I look forward to the day when a chicken can cross the road without having
its motives questioned.
```