Noise spectrum of voltage reference diodes

[jfphp]

A very interesting thesis

https://core.ac.uk/download/pdf/4393842.pdf

[MK]

I dont know if it is still there, but there was a cryptography website that went into using zener diodes as a noise source, for his aims he tended to run them at a lower current, and had some circuits on there too.

[Tomorokoshi]

Nice thesis. 33 years later the HP 3562A is still a very capable machine.

[David Hess]

My noise measurements using an RMS voltmeter, sampling and then calculating the standard deviation, and the tangential method of noise measurement with an analog oscilloscope, all regularly agree within 5%.  I trust the tangential method to verify that other measurement methods are working correctly.

Many years ago when I designed and fabricated a very low noise low drift low frequency amplifier, I used low value resistors as a test source to verify the noise performance.  By sampling and calculating the standard deviation, I was able to determine the resistance value of a 50 ohm resistor within 1 ohm from just its noise.

[jfphp]

Can you explain in detail your 2 mesurements ways ?
Was your "very low noise low drift low frequency amplifier" commercialy available ?

[David Hess]

Can you explain in detail your 2 mesurements ways ?

I listed 3 measurement methods.  All of them rely on filtering before measurement to control bandwidth for a spot noise measurement.  If I had a good FFT analysis method, then I would use it also to directly produce noise density measurements and I think it could be the most accurate because it supports long integration times.

1. Filter and then measure using RMS voltmeter.
2. Filter and then measure using the standard deviation calculation on a digital storage oscilloscope.
3. Filter and then do tangential measurement using a 2 channel analog oscilloscope - this video describes how to do this.  Note that this method can also be used with sampling oscilloscopes and to make RMS jitter measurements.  Most modern DSOs cannot support his method because of poor index grading of their display.

Was your "very low noise low drift low frequency amplifier" commercialy available ?

This was a project for internal use where I worked for use as the best possible strain gauge amplifier.  It was an extension of the common two path amplifier design to operate with differential inputs and outputs.  It combines a chopper stabilized amplifier for flat flicker noise and zero drift, with a low noise amplifier for low wideband noise down to below 10 Hz, in a differential configuration.  Even lower noise was possible with multiple amplifiers in parallel at the expense of input current noise, or with a discrete wideband stage.

[jfphp]

Thank you very much.