optimizing guitar signal through stomp boxes

["Rainer Thelonius Balthasar Straschill" <rs@moinlabs.de>]

I'd like to add some kind of signal theory for what it's worth...

For the order of stompbox etc. discussion, it's healthy to look at the
concept of LTI devices. LTI stands for "linear, time invariant", and means
(simplified) the following.

Linearity: If you double your input signal, your output signal will double.
If you take a mixer in front of your stompbox, use it to mix two inputs and
then send it through the stompbox, its output is identical to the case 
where
you send each input through a separate stompboxe identical to the one above
and then sum the signal after the stompboxes.

time invariance: If you send a certain input, you get a certain output. If
you send the same input, only a certain time amount later, you get the same
output as before, only the time amount later.

Of course, it only makes sense to look at these values in a "normal" range.
That is, if you apply an input of 1V Vpp and
get a certain output, applying 10000V won't get you 10000 times your former
output but most probably a little cloud of smoke...

For LTI devices, the following rules apply (again simplified):

        * the device is completely described by its impulse response - what
it outputs when you put an extremely high,                extremely short
spike on the input.

and more importantly:

        * as the inpulse response can be transformed to a complex frequency
chart (note that "complex" means complex                  numbers, i.e.
amplitude and phase for each frequency), the complex frequency response 
also
defines the
          behaviour of the device.

And as a consequence:

        * the mathematical operation to describe the interaction of several
chained devices in the frequency domain
          is scalar multiplication. As it is commutative, i.e. a*b=b*a, it
doesn't matter which order the devices
          are in. (just a reminder - only true for LTI devices!)


Now let's take a look which typical effects are linear and/or time
invariant:

delay: linear, time invariant

EQs, filters: linear, time invariant
        (this is only true as long as the filter doesn't distort/reach
saturation, i.e. operate in its linear range. This
         is mostly the case for typical mastering EQs, but not for 
screaming
Moog filters)

distortion (and thus amp modelers): time invariant

dynamic effects (compressor, expander): time invariant
        (but wait! you'll say - there are time constants involved! Yes, but
these describe a behaviour in reaction to an     input signal, not a
"freerunning" time constant)

modulated delay: linear
        this includes all of your chorus, flanger and phaser effects

reverb: nonlinear (most of them), time invariant (most of them).


Now putting all theory aside - I firmly believe that there is no best way 
to
chain a certain set of effects, or that there are effects orders that don't
work (unless you're limited in your style to, say, playing rhythm guitar in
a traditional blues band). Sure there are orders that give you more 
creative
options and some which will turn one short note into a big atonal wash of
noise (which may or may not be what you want), but apart from that - be
creative!

Some crazy ideas to try:

delay after a gated reverb.

superweird distortion (think Zvex Machine) after somewhat longer delay with
medium to high feedback setting and playing short notes or sequences.

eq before and after distortion.

cascaded, non-synced phasers.


Have fun!

        Rainer