[sdiy] TR-808 Kick Drum modelling explanation

[Richie Burnett]

> Thanks for this - fascinating stuff!
> In section 9, talking about the output stages, they say
>
> "There are non-negligible loading effects, but they mostly
> change the position of very low (sub-audible) poles.
> ....... these discrepancies are minimal, inaudible...."
>
> Do you agree?

Yes, totally.  There's a slight interaction between the Tone and Level 
controls because there's no buffer in between, but you would never notice it 
unless you looked for it on a simulator.  Basically, when you turn the tone 
control down to get a more mellow tone the overall volume also drops by 
0.6dB!  That's it.  Barely audible, in fact something that sounds more 
mellow naturally sounds quieter to humans anyway.  (I actually modelled this 
just because it was trivial to do so, but the final user will surely just 
adjust tone and volume controls by ear in a mix.)

There's also some high-pass action due to capacitors C47 and C49 (needed to 
stop DC causing "pot cracking") but these act at around the 3Hz mark. 
(That's what the sub-audible poles stuff is talking about.)  This is far 
below where most of the kick's energy is, and obviously well below the 
limits of human hearing and the best bass-bins.  On it's way to the speakers 
the 808 sounds will pass through many more AC coupling capacitors (and 
perhaps the odd "LF rumble filter") largely un-noticed ;-))

> And in the next paragraph, they mention a DSP model by Cycling 74 that 
> models changing the input pulse width. What effect does this have in  your 
> model?

Exactly what Colin said.

It has a big effect on the sounds for which the raw trigger pulse is used to 
drive the ringing T filter directly. The 808BD and 808SD are good examples 
of this, as is the 606BD.  It has much less effect in cases where an 
analogue drum sound generator just triggers off one of the edges (rising or 
falling) of the trigger pulse.  So for instance trigger pulse width doesn't 
effect hand-clap sounds much in the 808 and 909.

As a quick rule of thumb if you look at the spectrum of an particular 
analogue drum sound in an audio editor and it shows a whole series of deep 
notches in the spectrum, then this sound is made by filtering the raw 
trigger pulse in some way, and is likely to be effected considerably if you 
change the trigger pulse width.  The series of deep notches come from the 
rectangular nature of the trigger pulse, and no amount of filtering 
afterwards can hide their give-away signature.  Example below is for a 
TR-808 snare with the snappy turned right down:

http://www.richieburnett.co.uk/temp/808sd-spec.gif

-Richie,