Elements of the spectrum window
Spectrum window displays spectrum and peak evaluation information graphically.
This section describes static display elements as well as dynamic display elements, which also provide controls for quick fit manipulation.
Spectrum window is basically an X-Y chart, which depicts spectrum details on a counts versus energy scale.
Horizontal axis of spectrum chart
Unit of horizontal axis of the spectrum chart may be either channel or keV, while its scale is always linear. Toggling between these units may be accomplished in several ways:
Clicking on the “keV” or “ch” label on the navigating toolbar.
Selecting the proper unit from the main toolbar's “X unit” drop-down list.
Selecting View / Energy scale for positions menu.
Pressing Ctrl-N keys simultaneously.
Vertical axis of spectrum chart
The unit of the vertical axis of the spectrum chart is always count, but its scale may be either linear or logarithmic. Toggling between these scaling modes may be initiated in several ways:
Selecting the proper scaling from the main toolbar's “Y scale” drop-down list.
Selecting View / Logarithmic scale for counts menu.
Pressing Ctrl-L keys simultaneously.
Other axis settings (grids, tick marks etc.) temporarily may be set by selecting Edit / Edit fit chart menu, and selecting Chart / Axis tabs, then Left Axis item from the list. This may be helpful e.g. when special screen capture is needed.
At most three kinds of counts may be displayed in the spectrum window:
Counts to be fitted as black squares. These are either uncorrected counts (in case of regular gamma spectra), or loss-corrected counts (in case of some Loss Free Counting method), created by some kind of compensation electronics.
Uncorrected variances as gray diamonds, as found in the original spectrum file. These values hold information about uncertainties for each channel of the measured counts. For regular gamma spectra, the values are almost identical with the counts to be fitted (only a small correction is applied for very low counts). In case of Loss Free Counting methods, these are from the Dual LFC spectra's uncorrected half (e.g. for Canbera's spectra), or from a special compensated spectrum half (e.g. “ERR” spectrum half for Ortec's ERR-ZDT spectra).
Count variances as gray squares. These values are computed by HyperLab, possibly by applying Westphal-correction on original variance values. This is required in some cases because the loss-corrected counts have greater statistical variance than the non-corrected ones. As HyperLab fits the corrected counts by its peak functions, proper uncertainties are required during the fit.
For everyday purposes, it is necessary only to display the counts to be fitted, and the other counts are just for checking the details when you are interested in finer details. To switch off displaying the count variances, select View / Chart Details menu, then remove check marks beside Count variances and Uncorrected count variances menu items.
Please note the red triangles in the spectrum window: these are the suspected peaks, whose position is determined by the help of the current FWHM calibration before the deconvolution.
The peak search is performed by a zero-area square wave convolution. The final spectrum deconvolution will use these suspected peaks as an initial estimation of peak positions, when the fit model is determined.
The list of suspected peaks is always re-generated when you change the FWHM calibration. The suspected peaks are replaced by the fitted peaks after a spectrum deconvolution.
Sometimes the suspected peaks remain in the spectrum window, even after a whole-spectrum automatic fit. This is because sometimes the peak search algorithm finds spurious peaks, which are not replaced with real fitted peaks by the peak deconvolution algorithm.
Left and right fitting threshold markers
When a spectrum is loaded into the spectrum peak evaluator, a left and a right channel threshold value is detected automatically.
The limits are designated by a framed blue 'X' pointing rightwards and leftwards.
The peak search algorithm will locate suspected peaks only between these outer values. In most cases, this automatic threshold determination works well, but a manual adjustment may be necessary in some very special cases.
If adjustment is needed, just click on the blue sign and drag it to its new position.
Plot of deconvoluted regions
After spectrum deconvolution is performed, fitted regions will be displayed.
The sum of background functions (2nd order polynomial background, step, tail) is drawn by green, while the sum of peak functions (Gaussian, left and right skew) and the background will be plotted by red color. The peak functions of the individual peaks are plotted by blue lines, in order to depict finer details within closely joined multiplets. Peak labels are displayed under the peaks with their serial numbers, while region labels contain basic region information (region bounds and the reduced Chi-squared) at the beginning of the regions. These displayed items may also be configured through View / Chart details menu.
If you want to quickly adjust the plot to a specific fitted region, just left-click over it, and the spectrum chart will be immediately updated, zooming to the clicked region.
Left and right region bound marker arrows
At the beginning and the end of the plot of regions two blue arrows are displayed, which point towards the center of the region. If a region bound adjustment is needed, simply click and drag the proper arrow to its new place. After releasing the mouse button, an automatic fit takes place with the new bounds, and the new fit will be displayed.
Context-sensitive menu for fitted regions
When you right-click over a fitted region by the mouse, a context menu appears.
Its Region info menu item provides you a quick access to the details of the region fit.
The other menu items allow a one-click fitting model manipulation, as the background and the peak skew fitting function components may be eliminated with them. This tool may lead to a trial-and-error refinement method, but it can be surprisingly effective on typical misfit situations.
If the region fit becomes significantly worse after dropping a component, just press Undo button on the main toolbar, and the previous fitting model will be immediately restored.