Use a DVM to directly digitize low-frequency noise (Part 2)

-December 05, 2012

A Note About Measurement Verification
This step is critical as it is so easy to miss a math step or get a bandwidth correction factor wrong, etc. I recently read an article from some time ago where the authors stated that they used two random noise sources added together to get double the noise power. This would be true if the noise sources were correlated, but uncorrelated noise doesn't add directly. That is, two uncorrelated noise sources added together should give a total noise 1/Sqrt(2) less than one noise source alone. So their calibration math should show a 1/Sqrt(2) or nearly 30% error.

This is why I always try to look for 1/Sqrt(2), or Sqrt(2) times, or 10 times, or - if I really blow it - 1000 times errors in my calculations as these are clues that something is amiss and a digit or some correction factor has been missed somewhere.

A Fresh Look Paid Dividends
By taking a fresh look at the Manual for the 34401 and using it to its fullest extent I was able to really simplify the whole test hardware immensely. No longer did I need analog peak detectors in hardware – the direct digitized readings give me the peak-to-peak values easily later on in post processing. Additionally, since I am now storing the raw data instead of just the peak-to-peak values, I can calculate the “really, really true” RMS value and histogram for every 10-second interval also.

Nor did I need two setups to measure the Time Domain Noise and the FFT Analyzer for the Frequency Domain Noise. I was able to use one low-cost multimeter for both setups and I have the advantage of having one data set on the PC that I can store and manipulate as needed later on.

This all saves considerable time and needless re-running of tests with different test bench setups.

The last advantage is the nearly limitless statistics that I can run: Just turn on the hardware, fire up a Data Analysis Program, and then store and process data until your heart's content. Days and days of data may be captured with really very little incremental cost. The only real limiting factor is the life of the batteries that power the Preamp and the circuit that I am measuring.

[1] Steve Hageman, “The practicing instrumentation engineer's guide to the DFT - Part 1, Part 2 and Part 3, EDN Magazine, June 19, June 26 and July 10, 2012

[2] Art Kay “Analysis and measurement of intrinsic noise in Op Amp Circuits – Part III: Resistor Noise and Sample Calculations”

Loading comments...

Write a Comment

To comment please Log In