Typically in the block scheme of a spectrum analyzer a video filter, with a bandwidth VBW, is located after the envelope detector and before the screen.
As pointed out in this document, the VBW may (and sometimes should) be less than RBW (Resolution BandWidth, that is the bandwith of the IF filter): the resulting signal will be less affected by noise (page 28). But if the VBW is smaller than RBW, this means that part of the RBW output is "wasted", which is undesirable.
1) Isn't it?
The envelope detector itself is a low-pass filter applied after the RBW and before VBW, so (even without the VBW) the signal displayed on the screen has a lower bandwidth than RBW. But the bandwidth of the envelope detector is nowhere specified. So:
2) How can I become sure that no useful signals are cut off by a VBW < RBW? Should I display the signal with VBW > RBW and then lower VBW till all the harmonics are still preserved and stop when a harmonic is cancelled, in a trial-and-error procedure?
Note: the document was just an example, the only useful I found so far. I am not referring to a particular model, but to the (theoretical) reason why a video filter is used in a spectrum analyzer.