Goal: Trying to measure low-frequency noise such as 1/f noise spectrum from a material

What other people did: apply a DC current source --> take the voltage across the material --> amplify it with a preamp --> into a FFT spectrum analyzer

In my case, the problem is the noise is relatively small compared to the DC component, and the amplifier can easily get overloaded by the DC offset before amplifying the noise to a reasonable level. But if I use a DC block, the low cut-off frequency is still not low enough. For 1/f noise, I would like to measure down to 1 Hz. Can anyone suggest a way of doing it? Thanks.

  • \$\begingroup\$ What about a differential off the same signal with a capacitor to ground on one of the inputs? \$\endgroup\$ – Ignacio Vazquez-Abrams Sep 8 '13 at 5:06
  • \$\begingroup\$ Just thinking out load, but I wonder if you might be able to use a chopper/mixer to move the signal to another frequency; assuming you can keep the phase noise of whatever you drive it with sufficiently low. \$\endgroup\$ – Chris Stratton Sep 8 '13 at 6:29
  • \$\begingroup\$ @ChrisStratton: Regardless of the merits of his present answers, Tony Stewart is supposed to be obeying a ban from EE.SE that doesn't expire until 2014-10-2, as a result of past behavior. Instead, he keeps creating these "sock puppets", which we delete as fast as we find them. \$\endgroup\$ – Dave Tweed Sep 8 '13 at 11:45
  • \$\begingroup\$ And now you are removing my technical posts as well... Apparently because I had the honesty to say who had first stated what I was thinking might be an answer. \$\endgroup\$ – Chris Stratton Sep 8 '13 at 13:51

You say apply DC current to "a material" and this suggests to me the possibility of using more than one independent block of material...

The essence of your problem seems to be how to apply a high DC gain amplifier without the (naturally, DC) bias overloading the amplifier - and one approach suggested in comments is to apply a DC bias : the objection to that is that the DC bias source is also a 1/f noise source, and so provisions must be made to compensate for that somehow.

Another classic approach is a "bridge" configuration in which you acknowledge the DC bias is another noise source - but make it of the same material with an identical current source (or a current mirror) - and amplify the difference between the two DC voltages. Then compensation for two (presumably) independent noise sources is purely mathematical : adding 2 uncorrelated random (both assumptions!) noise sources adds 3dB to the noise level, i.e. doubling the noise power, or raising the noise voltage by sqrt(2).

There is a variety of amplifier called a "chopper amplifier" which converts these two DC inputs to AC and amplifies the resulting AC signal, thereby eliminating any 1/F noise contribution of its own. Probably useful here...

  • \$\begingroup\$ The bridge idea seems interesting. Another idea I have is to make a high-pass filter with very low low cut-off frequency. If I have a big capacitor (e.g. 100 uF) and some resistor, even 0.01 Hz is realizable. What do you think about it? \$\endgroup\$ – shva Sep 9 '13 at 3:55

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.