For my physics lab I have to come up with a way to measure Boltzmann constant using thermal noise.

The circuit that I 'm going to use is something like the one in the figure below: enter image description here

Than I'm going to attach an arduino UNO board to the output and measure the \$\Delta V\$, using a sampling rate of \$50\$KHz, and by computing the st.deviation, get a value for \$k_b\$.

I've some question about this setup:

  1. Is the LF 355 a good op-amp candidate for my experiment? Considering that the thermal noise input will be very weak and thus very close to the GND value is it better to use a RRIO op-amp?
  2. Is the sampling rate of Arduino enough to see a consistent variation of the \$\Delta V\$? Does the fact that the board can measure only positive difference of potential will alter the calculation of the st.deviation?

You'll need more gain than that to see 5 uVrms -- the arduino probably has a resolution of 10 bits (?), equivalent to about 5 mV. To see the signal, you'll need about 50 mV, so a gain of 10,000 is needed. The opamp's DC offset might be an issue, so just use a low value of DC gain, and bypass for AC -- that also provides your LPF.

Now check if the opamp has enough gain*BW to do this at 50 kHz.

If this actually works, you can then arrange to short your 100k R and do a 2nd measurement -- this would measure the opamp's noise, and you can subtract the measurements (actually sqrt(difference in squares)) to get the R's noise.


You need to avoid splitting the signal in half - just use 2x 200k resistors; one to ground and one to Vcc - now your signal will be central but, you'll also need to do the same with the 100 ohm resistors.

However, the op-amp's 20nV per root Hz internal noise might cause a problem. Across a 50 kHz bandwidth, this is a net noise of \$\sqrt{50,000}\times 20nV\$ = 4.47 uV.

A 100k resistor at ambient produces a noise of 9 uV across a 50 kHz bandwidth so you have problems using this op-amp because it's own voltage noise source is half that of the 100k resistor. The internal noises of the bias current of the op-amp chosen are very small and won't significantly affect things.

However, if you look at figure 32 in the data sheet for the TI LF355 you'll see a graph of noise voltage: -

enter image description here

This is indicating that there is considerable noise at the low-end of the spectrum (as per most op-amps) and this could produce a few more micro volts of noise. To avoid this I would high pass filter the output so the LF stuff was pretty much avoided. You should be able to do this in software.

  • \$\begingroup\$ Thank you! I forgot to write about passing the data through an high-pass filter so I'll take your observation as further confirm. I didn't get the part where I have to attach the 220k resistor to the Vcc (positive supply voltage of the op-amp). How I'm going to supply the op amp then? (this can be a newbie question but we didn't studied op-amp yet) \$\endgroup\$ – damides Feb 6 '16 at 15:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.