Analyses - Spectral Analysis

Navigation: Analyses Commands > Spectral Analysis

Spectral Analysis

Page Contents

Description

Navigation

Controls

Application Notes

•Non-uniform wavelength distributions

Related Topics

Description

The Spectral Analysis feature creates a Spectrum from an existing rayset by binning the ray powers according to wavelength. Additionally, this analysis can apply the spatial (or directional) extents of an analysis surface and/or directional analysis entity as well as their additional ray selection filters when binning the rays. When the analysis has been performed, a new Spectrum node is added to the Spectra folder of the object tree.

Navigation

This command can be accessed by selecting Analyses > Spectral Analysis from the menu.

Controls

Control	Inputs / Description	Defaults
Filter Rays By
Analysis Plane	List of available analysis surface's in the Analysis Surface(s) folder of the object tree. The spatial extents of the analysis surface and the accompanying ray selection criteria will be applied to the rays during the analysis.	None
Directional Analysis Entity	List of available directional analysis entities in the Analysis Surface(s) folder of the object tree. The directional extents of the DAE and the accompanying ray selection criteria will be applied to the rays during the analysis.	None

Maximum wavelength sample count	The spectral analysis feature has two calculation modes; discrete and fixed. In the discrete calculation mode, there is one bin for each wavelength included in the calculation. When the number of wavelengths counted exceeds the "Maximum wavelength sample count" value, the fixed mode is used. In this method, the wavelength range of the rays included in the result is divided equally into the number of bins given by the maximum wavelength sample count.	200
Rescale peak spectral value to	Rescales the resulting spectrum to have its maximum value corresponding to the designated value.	1.0
Extend wavelength limits by 1/2 cell	When rays are generated, their wavelengths are weighted such that the power in each wavelength represents some portion of a spectrum. Depending on how the ray wavelength weights were generated, there can be an ambiguity in the interpretation of the extreme minimum and maximum wavelength values. Consider the case below where a wavelength list has been weighted according to a spectrum (green). In this case, each wavelength represents an area underneath the spectrum having equal width. The same source whose wavelengths are represented by a spectrum, as opposed to a discrete wavelength list, will only represent the bandwidth between the extreme minimum and maximum wavelengths. The bandwidths represented by the two weighting methods are therefore different by 1/2 of a cell on either end of the spectrum. Choosing the "extend" option on the spectral analysis accounts for this 1/2 cell difference in the bandwidth and may need to be used when the wavelengths are not represented directly by a spectrum.	Unchecked

Add spectrum to tree	If checked, a chart window displaying the result of the analysis is opened and a new spectrum node containing the result is added to the tree. If unchecked, the result is displayed in the chart window but no spectrum node is added to the tree.	Checked

OK	Perform analysis and close dialog box.
Cancel	Discard analysis and close dialog box.
Help	Access this Help page.

Application Notes

Non-uniform Wavelength Distributions

Consider the following scenario; three LED's (R, G and B) are combined to provide white light illumination. This is modeled in FRED by creating three sources, each with the appropriate R, G, B spectrum and power weighting. The sources are set to generate wavelengths randomly according to a spectrum, using the individual R, G, B spectra shown on the top of the graphic below. If we create these sources (resulting in several millions of rays) and perform a spectral analysis on the resulting rayset, we get the spectrum shown on the bottom of the graphic below. This combined spectrum contains several millions of unique wavelengths by virtue of having used the "Randomly according to spectrum" wavelength generation method. Consequently, the spectral analysis mode would use the fixed bin calculation mode because we have exceeded the maximum wavelength count of 200.

Now lets change the wavelength generation method on the three sources from "Random according to a spectrum" to "Evenly-spaced, weighted according to spectrum" and use 12 wavelengths. Each source now uses 12 discrete wavelengths evenly spaced along its spectrum. When we re-trace and re-calculate the spectral analysis, we get the following combined spectrum for the output.

How did changing the wavelength generation mechanism on our sources lead to the spectrum shown above? The maximum wavelength sample count value in our Spectral Analysis is set to 200 samples and our sources only generated a total of 36 unique wavelengths (12 wavelengths each). So, our spectral analysis is binning the rays using the discrete sampling mode. In the discrete binning mode, the distance between adjacent samples defines the bin size for a given sample point. Although each source sampled its own spectrum using evenly spaced samples, the resulting composite spectrum shows sample points which are very non-uniform, as shown below.

As a result of this non-uniformity, the bin sizes used by the spectral analysis do not match the bin sizes which were used at the source when the wavelengths were weighted. This inconsistency causes the spectrum formed by the spectral analysis to either over-shoot or under-shoot the expected spectral value, leading to a very ragged appearing spectrum. It is important to note that on average, the spectrum will still be accurate and the area underneath the spectrum will be correct.