Spectrum Window Bandwidth Correction (BW Corr)

Controls: Spectrum Dialog >> BW Corr
Macro: SpectBW

You can toggle Bandwidth Correction via the BW Corr button in the Spectrum dialog, or directly via ALT+B even when the dialog is not visible.

When BW Corr is active along with the Window button, it corrects the spectrum to show true energy instead of the default "best peak" amplitudes. To understand why this is needed, consider that an unwindowed spectrum will show erroneous peak heights for any frequency component that doesn't fall exactly on a spectral line; its height is reduced due to leakage into adjacent spectral lines, even though the total energy is still correct if you consider all the leakage components.

Applying a window function causes the leakage energy to be squeezed back toward the true signal frequency, but the window function isn't magic; in effect, the new peak is a little too wide at the top, in exchange for being a lot narrower farther down. The resultant peak is then scaled to give the best peak accuracy; it will still be somewhat low at many input frequencies (unless you use a Flat Top window), but the reduction of broad leakage skirts makes the spectrum much easier to interpret.

However, the extra peak width (more apparent high-level components) gives an apparent boost to the total energy. This is not normally an issue; most often, windowing is used in situations where energy is not being measured, so anything that reduces leakage and improves the peak accuracy and visual appearance is definitely worthwhile.

But if you are measuring energy, this BW Correction option sacrifices peak height accuracy in exchange for true total energy accuracy. You still get the improved visual appearance of reduced leakage skirts, but the peaks will be much lower... and their heights will definitely not correspond to actual signal amplitudes.

You'd normally use BW Correction whenever you use the Sigma mode of the cursor readouts to measure effective signal amplitude, and a windowing function is active.

Another way to look at the effect of windowing is that it widens each spectral analysis band that is normally shown as a single spectral line; since the bands are still only one spectral line apart, each band overlaps its neighbors. That band leakage means that when a signal has components present in adjacent bands, they are "double-counted" and the overall level appears too high. This is a problem with broadband portions of a signal, such as the "noise floor".

In general, you would probably want to just toggle Window off during noise measurements. That will give proper noise level measurements (such as by spectral averaging) and it will also give proper energy measurements via the Sigma option.

However, if there are also strong tone signals present, their leakage skirts could then contaminate the noise floor measurement. In that case, use Window to reduce the leakage skirts and BW Correction to correct the noise floor reading.

Always remember to toggle BW Correction off before making peak readings.

Macro Notes:

SpectBW=1 activates BW Correction, SpectBW=0 turns it off, and SpectBW=x toggles between on and off.

See also Spectrum Control Dialog