[linux-audio-dev] Re: Cheby amp code

Steve Harris S.W.Harris at ecs.soton.ac.uk
Sun Nov 3 11:43:00 UTC 2002

On Sat, Nov 02, 2002 at 08:12:10 +0100, Tim Goetze wrote:
> >In the instantaite block fixes it up more-or-less. Maybe even adds a bit
> >of compression (it boosts the gain to make it roughly 1:1 too).
> have you modified the lut in the meantime? i don't seem to be getting
> the right results with the log scaling you use.

Did you try adding the 0th harmionic to the front of the table, and
dropping the last harmonic? That made it sound pretty good for low notes.
> ah, i'm afraid that's only half the truth. try the link i posted
> to the list; somebody has been doing this before us and published. 
> i still haven't checked for frequency dependency of valve distortion,
> but it seems they have, and found it to exist. 

Figures, on both counts.
> additionally, it seems that to tackle intermodulation distortion it 
> is best to split the incoming signal to multiple bands (those great
> theoreticians have been using FIRs in matlab) and shape each band
> separately. they do in fact use two blended shapers per band.

OK, why do they use two shapers? Or is it one cheby and one non polynomial?
> btw, the 'lower frequency' phenomenon occurs with the real amp, too,
> only it's not as pronounced.

OK, I didn't know that.
> drat. i fear that we'll need to cool this box into superconductance
> to get the effect done *really* right.

Yeah, I'm thinking we're not tackling it right, if behringer can knock out
a virtual amp harware box for E150 it can't the that cycle hungry.

- Steve

More information about the Linux-audio-dev mailing list