Parametric EQ distortion/overdrive question

[KarenColumbo]

Me again.

I've been thinking for this for quite some time. The idea is to "extract" a defined frequency band from the input signal by way of a parametric equalizer (http://sound.whsites.net/articles/state-variable.htm#s30 Figure 5).

In my ideal world, my stompbox would leave the input signal MINUS the "extracted" frequency band at this stage.

The "extracted" signal then is sent through some overdrive/distortion circuitry.

The thus processed signal should be "inserted" into the input signal (which was left WITHOUT the extracted band), so the signal becomes "whole" again.

The idea is to NOT just split the input signal and then mix it with the processed signal at the output, but to rip it apart, process one "half" of it, and then put it together again. Very metaphorically spoken, I'm afraid.

Is something like this possible? Would I have to build 2 EQ circuits in parallel, one for emphasizing the frequency band, one for simultaneously attenuating the same band?

[GibsonGM]

Sure it's possible!  A neat idea, too.     I don't know how something like that would sound, if it would be pleasing or odd, funky or quacky, silly or badass...it would take a bunch of tweaking to find the right frequencies to mess with and bandwidth to make it sound good, I'm pretty sure. 

I think a bunch of experimentation is in order.  Mainly to see what is possible with garden-variety components and techniques.  Finding out if a 1st order filter is fine, or if you need to go further....what kind of Q do you need? Questions like that. 

It seems to me that you could pull the band out and work with it without needing to build parallel EQs or anything...keeping it simple, unless the results tell you that you need to go more complex.  This feels a little bit like some kind of modified crossover, and then a mixer.   It's not clear to me how 'sharp' you'd need to be with the cutoff freq., and so on.   Pulling JUST a slim freq. range would probably sound un-musical; not slim enough, and you're just mixing a fuzz and clean!

Interesting idea, I look forward to see what the more 'guru' folks think! 

[Isak]

your idea is something i want to do for a while, well, not exactly the same but sort of the same.
i want to take off the low frequencies at the input and gain it back at the output.
the idea was born from my need for a good overdrive to use with old analog synths.
the problem is the very thick fat bass those synths have and it sounds very muddy, choked at the output.
hope youll have good progress and please do share.
the overdrive i'm working with is the old MXR dis+ clone from tonepad http://www.tonepad.com/getFile.asp?id=115
its funny you post this.

i think i can help with the mix pot part
http://electricdruid.net/wp-content/uploads/2015/06/VCBigMuffTonecontrol.jpg
just take out the LP and the HP components your wet input should be from R10, the dry input should be from R9, replace R4 with 100k trimmer, at the cv input use 10k pot.
works amazing!

cheers

[KarenColumbo]

I would (if I could) get a step fruther and make the whole shenanigan velocity sensitive Like in my first thread in this excellent board Like: the harder you dive into the axe, the higher the center frequency goes up. Well ... still routing step 1, the parametric eq circuit in expresspcb

[thermionix]

Phase would be a concern.

[KarenColumbo]

I see - well, I'll have a go at it and dodge the bullets that come flying at me in the progress

[Mark Hammer]

Me again.

I've been thinking for this for quite some time. The idea is to "extract" a defined frequency band from the input signal by way of a parametric equalizer (http://sound.whsites.net/articles/state-variable.htm#s30 Figure 5).

In my ideal world, my stompbox would leave the input signal MINUS the "extracted" frequency band at this stage.

The "extracted" signal then is sent through some overdrive/distortion circuitry.

The thus processed signal should be "inserted" into the input signal (which was left WITHOUT the extracted band), so the signal becomes "whole" again.

The idea is to NOT just split the input signal and then mix it with the processed signal at the output, but to rip it apart, process one "half" of it, and then put it together again. Very metaphorically spoken, I'm afraid.

Is something like this possible? Would I have to build 2 EQ circuits in parallel, one for emphasizing the frequency band, one for simultaneously attenuating the same band?

What you describe is more or less how some "exciter" circuits work.  For instance, this classic circuit that appeared in Electronic Musician many years back splits off the upper mids and highs, distorts them, and then mixes them back in with the clean signal.

I guess one of the things to remember is that distorting a band - of any width - implies adding harmonic content.  So, if you used a theoretical "brick wall" filter that restricted content to only what is between 400 and 800hz, the output of a distorted version of that would include a lot of content well above 800hz.  Maybe that's what you were aiming for, and maybe not.

Since the intended distortion would also require applying gain to the selected portion, you would need to figure out how much to attenuate back, in order to restore tonal balance when you re-insert.  I suppose, though , that simple back and forth comparison between processed and processed would permit finding a suitable setting.

[robthequiet]

Jack Orman Has an article on frequency clipping, might be worth a read.

[KarenColumbo]

I wanted to start with one of Rod Elliott's EQ circuits - mainly because I remember it from quite some years ago and because it's a f &/((/%%&/ biatch to bring to PCB stage Please see this schem.



Then it occurred to me that this one is the first I've ever seen with a Polcap/resistor combination right at the input (and the output and most, if not all, of the ground connections). Since I wanted to split the signal at the input (and buff it up with a inverted opamp buffer to have it buffed up again at the mix at the output) I wonder now what these polcaps do at those places.

Last, but not least, a huge THANKS again to you for all those circuits and knowledge you keep bringing to this thread - they are all read, filed and printed out and appreciated very warmly!

[KarenColumbo]

Edit: Well, that sure didn't work out so well In the meantime I'm just breadboarding a simpler para EQ circuit suggested @ Musikding.de. Got to get this one right as a center of all.
Problem - as you all well know, I assume - is that there's lots and lots of some devices and a painful lack of those you need right this moment. So every few hours I either order some of those or take the sub and buy them. Like some 4001ers diodes. And so on ...
Don't give up on me 

[Groovenut]

The Klon Centaur does this, but not exactly by using a parametric eq. The schematics are out there, so they may provide some inspiration. Using a series of filters, the midrange band is passed through the clipping section then reintegrated with the input signal. The clean signal is not quite as aggressively carved as the mid band and I think this is why the blended signal works as well as it does. The clipping of the signal is going to add harmonic overtones which will change it's frequency content. This should be adjusted for in the clean signal to be remixed so as to achieve a nice tonal balance in the final signals sound.

I have seen software versions of this, where each band of a parametric's output was clipped individually (or not at all) and then remixed. I would think that the phase issue might be circumvented this way as the phase alignment on the output of the parametric will be closer than if you mixed the clean signal with the filtered signal.

Just my two cents

[KarenColumbo]

Yeah - this is quite along the line of my thoughts. It's taking time and maybe it's fruit- and senseless, but I'm trying to construct a sort of "mirrored" paraEQ - with stereo-pots, on half of the gain/cut-pot wired the other way, so I can attenuate in the original signal while I emphasize the same band in the FX-"channel".

In my head this will not eliminate the generated overtones but it will "even out" the mixed signal somewhat - and those overtones in most cases are intended, at lest when I overdrive or distort the signal. May be too much when inserting a phaser/flanger, but that's another story right now.

I d/led "Active Filter Cook Book" to get some bearings - so eventually I will get the hang of phase control - but that's a long way off - too little knowledge and time.

But.

Maybe this just works and sounds good.

[KarenColumbo]

First step is the hardest. But my artistical vein is throbbing with anticipation. Heh. This is gonna be a long way to professional PCBs. But what the heck