Envelope Follower
gstopp at fibermux.com
gstopp at fibermux.com
Wed Feb 28 18:40:10 CET 1996
Oh yeah I fergot to mention the multiplier type of envelope follower -
basically this is made by running the audio signal into both inputs of
a ring modulator, and smoothing it with an integrator like Don says.
This acts as a full-wave rectifier with a logarithmic transfer
function. Probably you could get the same results with an amplitude
recovery rectifier followed by a log generator (right?).
I remember seeing an untested design proposal way back in an old
Electronotes - it consisted of two multipliers, with the audio signal
running into a wideband 90-degree phase difference network, with the
SIN output going to both inputs of one multiplier, and the COS output
going into both inputs of the other multiplier, and the outputs of
each multiplier averaged together. The proposal was that the valleys
on the SIN multiplier's output would be filled in by the peaks of the
COS multiplier's output, and vice versa. This way there would be no
smoothing filter required, or so it seems. Would this work?
Re: the heating resistor method - talk about response time! :-)
- Gene
gstopp at fibermux.com
______________________________ Reply Separator _________________________________
Subject: Re: Envelope Follower
Author: don at till.com at ccrelayout
Date: 2/19/96 10:18 AM
The two most common methods of envelope detection are:
Cheezy: Rectify the input voltage and average.
Proper: Square the input voltage (ie., multiply the signal by itself),
average, then (optionally) take the square root of that.
The difference shows up in how low and high input signal voltages are
relatively weighted in contibuting to the average.
(There's even a mechanical variation of the "proper" approach: Use the
input signal to drive a resistor and measure how much it heats up.)
I can't find my catalog now to double check, but it appears that Serge
is saying they don't use the "cheezy" apprach.
-- Don
More information about the Synth-diy
mailing list