Datastick Sharpener™ Peak Clarification Technology
A New Approach to FFT Spectrum Accuracy

This is a summary of the illustrated, seven-page technical note. For a FREE copy of the full seven-page note, please fill out the form below.

We are used to getting repeatable high-resolution results from instrumentation, but we sometimes take the results for granted while forgetting that some of the underlying signal processing functions in the instrumentation have well-documented intrinsic hidden errors.

A significant source of error in an FFT comes from the fact that the FFT “bins” are fixed, evenly spaced frequency spans. An individual FFT bin’s value is influenced by the relative position of a frequency in a bin. If a specific frequency is centered in a bin, it will display a different amplitude than it would display if it is close to the border of two adjacent bins. This creates the potential for significant peaks to be missed because they display at reduced levels when they are not centered in a bin.

One of the major uses of windowing functions (such as Hanning or Flattop) is to compensate for amplitude error due to a measurement’s component frequencies not being centered in the FFT bins.

Despite these peak amplitude errors, an FFT with Rectangular windowing (which is really no windowing at all) naturally reflects the amplitude of centered frequencies more accurately than any windowing method. An FFT made using a Flattop window shows the truest amplitude levels for peaks, but reduces the accuracy of the resolution, distribution, and overall level of frequencies. In terms of frequency distribution accuracy, Flattop is the worst of all windowing functions. Since it is generally accepted that you can’t have your cake and eat it too, windows that strike a compromise between amplitude accuracy and frequency accuracy are most often used (primarily the Hanning or Hamming windows). These compromises give fair amplitude accuracy (at least better than rectangle), and fair frequency accuracy and distribution (better than flattop). Other windows favor one parameter at the expense of the other. But only Rectangle and Flattop can give you reliable results in one regard — amplitude of centered frequencies — but they still give variable results in adjacent frequencies. (The “short-sheet” effect; you can cover your neck or your toes, but you can't cover both)

Since we know the mathematics of the FFT and we know that the bandwidth characteristics of an FFT cause errors in distribution and strength of a signal’s energy between bins, we at Datastick have engineered a tool that provides an alternative to the above compromises. This new tool is not a window function, but a proprietary algorithm that enhances the resolution of both amplitude and frequency in FFT representations.

This new tool can examine all the data used to produce an FFT graph and reliably predict peak level errors and adjust them. We call this technology “Sharpener.” It results in peak level accuracy similar to that of Flattop windowing while maintaining the proper resolution and accuracy, energy distribution, and location offered by rectangular windowing on centered frequencies.

Sharpener is a feature of DAART™ — Datastick Advanced analysis and Reporting Toolkit.

For a FREE copy of the illustrated seven-page technlcal note, Datastick Sharpener: A New Approach to FFT Spectrum Accuracy, please fill out the form below and click the Submit button.