Login / Register 0 items | $0.00 New @ KVR
stratum
KVRAF
 
1676 posts since 29 May, 2012

Postby stratum; Fri Jan 12, 2018 1:43 am Re: Adjusting the knee of a sigmoid limiting function

Weasel words if ever I heard them


In my opinion waveshaping with tanh(x) as guitar distortion tends to sound OK at low gain settings as long as pre-and post-distortion EQ settings are suitable. It gives a somewhat lackluster sound, i.e. while it works, there is nothing impressive about the result. At higher gain settings it leaves a lot more to be desired, none of the liveliness of an actual valve amp is approximated by it. The circuit in the paper is meant to be a common element in solidstate overdrive/distortion pedals though, and I guess these aren't particularly great either.
~stratum~
matt42
KVRian
 
994 posts since 9 Jan, 2006

Postby matt42; Fri Jan 12, 2018 2:03 am Re: Adjusting the knee of a sigmoid limiting function

I guess there's an art to building a diode based pedal. There are some very well regarded ones and others not so much.

I didn't read the paper, so no idea how they made decisions on perceptual differences. In general to make such a claim you need to first build a valid listening test and then for a guitar pedal it's probably best to use experienced guitarists familiar with these kind of effects as the guinea pigs. Hopefully a double blind kind of thing where they get to play through the effects themselves
stratum
KVRAF
 
1676 posts since 29 May, 2012

Postby stratum; Fri Jan 12, 2018 2:13 am Re: Adjusting the knee of a sigmoid limiting function

Returning back to the idea of trying to fix the result with bias adjustment, i.e. tanh(gain*x+dc_bias), that fails because of the amount of gain required, and introducing a dynamic bias shift without actually doing a careful simulation yields doubtful results. Furthermore when there is more than one gain stage as in a typical amp, a conducting valve grid (which resembles a diode) loads the previous gain stage to some extent on one half of the wave (thinking of a pure sine wave as input), so the effect isn't even just 'dynamic bias shift'. Furthermore that means the whole circuit needs to be simulated, not just parts of it.

p.s. I have seen people hand picking germanium transistors by measuring their current gain (beta), other than that a distortion pedal can be easily mass produced and there isn't really an art in replicating one unless it happens to be a fuzz pedal using germanium transistors. Similarly perhaps silicon transistors can also be considered to be unreliable to a lesser extent, but an opamp circuit using diodes as in the tubescreamer ( https://www.muzique.com/tech/scream.htm ) is pretty consistent.
~stratum~
Previous

Moderator: Moderators (Main)

Return to DSP and Plug-in Development