Measuring the White Noise Produced by a Resistor

[jonlg_uk]

Hi all, I want to measure the white noise produced by a resistor in my circuit. In order to do this I did the following:

1. Ground the excitation voltage of the circuit and measure the white noise voltage produced by connecting the output of the circuit to a 8.5DVM

2. Subtract the Avg DC offset of from each white noise voltage measurement taken. This will leave me with the AC content of the measurements.

3. Perform FFT on the AC content.

4. Divide the Spectrum magnitude produced by the FFT by 2^0.5, in order to obtain the spectrum magnitude in terms of Vrms/root Hz

6.Replace the resistor in question with a short circuit (copper wire)

7. Repeat the measurement using the 8.5 DVM, measuring the output of the circuit for white noise voltage for when the resistor is replaced short circuit.

8. Subtract the Avg DC offset of from each white noise voltage measurement taken. This will leave me with the AC content of the measurements.

9. Perform FFT on the AC content.

10.Divide the Spectrum magnitude produced by the FFT by 2^0.5, in order to obtain the spectrum magnitude in terms of Vrms/root Hz

11. From the 2 FFT plots obtained, choose a frequency and read off the magnitude.

12. Use the following equation to calculate the Noise Voltage Rms of the resistor:

Ediff_rms= [(Eres_in_circuit)^2 - (Eres_replaced_with_short)^2]^0.5 * freq


Can anyone verify if this method of measurement is correct?

I thank you all in advanced.

 

[Floid]

Generally I think you just calculate this (http://en.wikipedia.org/wiki/Johnson–Nyquist_noise).


However, if you wanted to measure it, a few things to keep in mind.

1.) You are going to need an anechoic chamber. Noise from other sources will swamp out the noise you are trying to measure.

2.) You need a very accurate way of measuring. You will be measuring in the nanovolts. Not sure the best way to do this, might need an amplification circuit before your measurement device (this will add its own noise though).


Another thing to keep in mind is that the wire itself will have its own Nyquist noise. So what you will end up finding is the difference between the noise of the wire and the noise of the resistor.

 

[yungman]

Any active electronics produce more noise than resistor, this means all the equipments you using create more noise than the resistor you try to measure. Why don't you trust the thermal noise equation?

 

[jim hardy]

since it's thermal noise

perhaps you'd want readings with the resistor at several different temperatures, from as cold as you can get it to as hot,

and plot results against temperature ??

i don't know what is a 8.5 dvm.

i'd have used a high gain AC amplifier positioned outside the temperature chamber.
that way its contribution should be constant.

your fft sounds like a good approach.
and the experiment sounds educational.

how'd your results compare with theory?

 

[f95toli]

Measuring white (Johnson) noise is not very difficult, I do it quite often since it is a nice way to calibrate equipment (to make sure there is no unwanted excess noise).
Note that since the magnitude of the noise dependens on the resistance, you can always just use a big resistor (fairly obvious, but still.,).
Hence, some of the comments above are not correct; even a good standard op-amp will have an input noise level much lower than the thermal noise of a reasonably sized resistor; meaning measuring it is not very difficult.

(also, an anechoic chamber is chamber without reflections, for electrical low noise measurements all you need is a screened room; neither is needed for this measurement)

I would recommend the wiki on Johsson noise