[sdiy] Fast envelope follower circuit needed..

[rsdio at sounds.wa.com]

On Mar 29, 2013, at 14:32, Scott Gravenhorst wrote:
> POSSIBLE NAIVE POSTER WARNING...  And even if you did and used full  
> wave rectification, the peak
> detection part of the algorithm which would supply the filter with  
> its input data removes the
> spikes (and their associated transients) anyway - so aliasing due  
> to transients becomes moot.  The
> signal you filter is then a band limited signal with very small  
> potential for alias artifacts due
> to very low frequency harmonic components.


The peak detection part of the algorithm actually creates aliasing  
rather than removing it. That's because peak detection allows the  
output to suddenly change if the input is greater than the hold  
value, and any time a digital algorithm allows a sudden discontinuity  
in its output you have the potential for non-band-limited square wave  
type aliasing. You are correct that peak detection will remove any  
downward-pointing spikes, but the tradeoff is that peak detection can  
introduce upward-pointing spikes. I suppose you could argue that the  
peak detector will never make a sudden jump unless the input signal  
already has a sudden jump, but the nonlinear nature of abs() destroys  
the continuity of the original signal.

If aliasing were a problem, then I don't think it's correct to assume  
that peak detection would avoid it. However, my previous message  
tried to make the argument that aliasing is not a problem for  
envelope control, and thus it doesn't really matter.

Brian Willoughby
Sound Consulting