Mixable distortion - this is a little embarrassing

[NPrescott]

A while back I had the brilliant idea to make a pedal that would split your original signal via high pass and low pass filter and route the filtered signals through selectable distortion/gain stages.

Now, I saw the other day that I was not in fact the only one with this idea - which figures, all my best ideas aren't my ideas  .

That wasn't the embarrassing part though, I went about drawing up how I would do such a thing and realized I don't know what distortion sections I would use. If anyone has suggestions I'm pretty open here.

I was wondering, is there any benefit to my using active filters in this instance? Or should I just stick to passive?

Also, I seem to recall Barber Electronics filtering pre-gain stages - specifically bass frequencies. Now I can't remember where I read/heard this, and assuming I'm not making it up - what would be the benefit? I plan on testing out the filtering both before and after distortion, but I'm curious if there is any theory behind this.

I haven't yet draw up the mixer portion

edit: Speaking of embarrassing, ignore my labeling on the schematic

[Mark Hammer]

Been there, done that.  It works, and it sounds interesting and very flexible (hence the name).

Try this out:  http://i414.photobucket.com/albums/pp228/Mark_Hammer/Flexidriveschemv3.png

...or come up with your own variation.  I picked a crossover/freq-split point of around 880hz for the treble/bass sections, but there is no reason why another split-point wouldn't work just as well, if not better.  The other salient feature is that a single pot controls gain of each path in reciprocal fashion (i.e., cranking the gain on one means lowering it on the other).  I suspect the tapering and "gain-tradeoff probably needs some fine-tuning via fixed resistors in parallel with each leg of the drive-balance pot.

But in general....go for it.  I look forward to these sorts of experiments.

[GibsonGM]

These are great ideas, even if others got there earlier.  Mark has done it, so has AMZ, probably PIAA.   Arriving at the same place from other origins means it's worth working on )   I've always wanted to dive into this but got too lazy! 

As for sharp or broad filtering, try them both!   The first will make a rapid transition, the 2nd will provide more of a 'blending', generally speaking.   But you're dealing with harmonics of notes, too, so it's all blurred a little.  Results will depend on how you clip each of the signals....assymetrically, Si vs Ge, etc etc.   
You'll probably need to increase the gain of 1 or the other to get where you want, too, so that might need to be variable until set, then replaced by fixed resistors. 

Get busy, we want to hear how this turns out!  (AUDIO CLIPS!)

[zambo]

I want to do this type of project. I just wanted to simply split the signal full spectrum and run right with tube distortion and left with solid state and recombine into one amp input. Ideas how? I have never made an effective splitter or mixer..

[NPrescott]

I'm no expert but if we're learning this together I can certainly share some of what I've learned. You might take a look at ROG's Splitter-Blend as it will do what you want (I think).
http://runoffgroove.com/splitter-blend.html
I'm not certain on whether you want stereo operation, but I think it should be easily accommodated (well, relatively). Instead of an effects loop you'll just be using the splitter to run one signal to your tube distortion and the other half to the ss portion, blending stays intact.

Side note:
I did some reading today, thinking about filtering prior to the gain stages and as best as I understand you'd want to avoid filtering post gain stages to maintain the overtones/harmonics produced by the distortion itself. This makes intuitive sense to me - but this is all pieced together from various books and I can't find anything comprehensive. Nothing I can understand anyway.

-just thinking out loud here

[goulashnakov]

I've noticed it makes a huge difference the amount of highs you roll off going into a distortion, especially high gain.  After sticking a lo-pass filter after my strat while using the bridge pickup, going into a metal pedal, I found getting rid of some of the highs in the input signal made it a whole lot less slushy, and seems to bring out the harmonics a little better.  Case in point, I figure your lo-pass output would feed well into a (high-gain) distortion.  Perhaps you could let the brighter side of the signal (high-pass output) blend back in with the distorted signal to get a more "transparent" tone, but without sacrificing the icy gain of raging sonic hatred?  Like a tube screamer, but waaay hotter.  I'm curious to put this theory to test, because such I have not yet done myself.  I don't know if hi-gain/metal distortion is really your thing, but if you feel like experimenting with this idea, I'd be curious to know your results.

[GibsonGM]

The Splitter/Blend ought to do pretty well. 
Generally, bass is cut before distortion to eliminate 'mud'.  But like Goul just said above, if you're looking for a specific tone, have bright pups, etc., there's no reason not to mess around and warm things up, then add some sparkle from the clean signal (or a separately-distorted high pass...).  Sky's the limit! 

There can also be phasing effects, depending on what you do to the signal.  Every time you go thru a cap, you get 45 degrees of phase shift.  So when you mix back together, you might hear some unexpected results!

[Mark Hammer]

One of the things that people too often forget is that when diodes are used for clipping you have to be cogniscent of the fact that they have an absolute clipping point.  Combined with the fact that different parts of the spectrum present different signal amplitudes, that means that you would need to adapt the gain of each individual channel/pathway such that whatever portion of the signal you put through it is a high enough amplitude to actually clip to the degree intended/desired.

Let's put it another way.  If the clipping threshold for each channel is  +/-500mv, the average signal amplitude of stuff below 1khz is, say 80mv, and the average ampitude of content that has been high-pass filtered at 1khz is around 10mv, then clearly the gain of the high channel needs to be greater than that of the low channel if both highs and lows are to clip the same amount.

Note, as well, that what counts as "lower-order harmonics" for content at different points in the spectrum varies with where the actual fundamental is.

[amptramp]

I have thought of using a biquad filter to give simultaneous lowpass, midband and highpass filters to drive a colour organ and separate channels of distortion.  If the biquad is variable like a wah pedal, you could have interesting effects with the lowpass plus bandpass as the main signal but separate fuzz/distortion on all three channels.  Lots of interesting possibilities here.

[NPrescott]

Every time you go thru a cap, you get 45 degrees of phase shift.  So when you mix back together, you might hear some unexpected results!

I didn't know that! I'm glad you mentioned it though, as I would otherwise likely spend the weekend scratching my head over whatever phasing I'm likely to run into.

Mark - thanks for all the information, I'll admit I had to read it twice. I'm thinking much of the gain stages will be heavily modifiable (lots of knobs) because of problems with volume/gain. My initial idea was to have the filters sweepable so that you might adjust where the split signal falls (more knobs). If I've read right, this would mean swapping R5 and R6 for pots and carrying on my merry way.

I appreciate everyone's input because most of my reading (Physics of Sound/Musical Acoustics) treats distortion as a bad thing, so reference is slim.

Ron, I was actually drawing up a color organ when I had this idea  .

[GibsonGM]

Actually, I correct myself...every time you go thru an R-C circuit, you get 45 degrees for -3dB of filtering.  The cap phase angle is something like 90 degrees in a purely capacitive circuit, but we have stray resistances and the like.  The short answer there is: worry about it after, you know it's there if you get odd results, lol.  Getting a 180 degree shit is like winning the lottery....not likely.

Mark is hitting 100% on something....make this baby adjustable!  If you 'trim' 1/2 the signal by filtering, of course your signal is now lower in amplitude, so each half of the distortion will need independent gain adjustment to 'recover' the lost amplitude.  Each half will need a different amount of gain recovery. 

So it looks pretty straight-forward:  split the signal into 2 identical parts, trim top from one, bottom from the other, feed to distortion of your choice with variable gain, and tweak the filter-shaping networks of each to taste once it's working.    Control the level of each signal and then mix back together, and control the signal level at the output.  Done.    All that's left is to think about how you want to filter...1st order, 2nd order, overlap the cutoff frequency a little, spread it out...that's where the art is )

As to the books...yeah, we're kind of on our own as relates to distortion, ha ha.  Luckily there are many sites like R.G.'s and Mark's!! And AMZ Labs.     

[zambo]

WOW! A wealth of information! I will look up the rog splitter blender and see if I have the know how to build. I built the big daddy distortion using the LM386 - n4 and it sounds pretty good for a solid state distortion. I want to run the on one side and a 3 tube distortion pedal on the other ( Its name is Voltron I decided). I will post results when I build it. Thanks guys. I appreciate the help. I dont have schematics drawn up but I can explain a pretty ripping tube distortion based and cascaded valvecasters if anyone is in need. Cheers, Zambo

[edvard]

I think the earliest example of this idea was Craig Anderton's Quadrafuzz.
Refer to "35 Do It Yourself Projects for Guitarists".
Excerpt:

What makes the Quadrafuzz different from other distortion devices is that is splits the guitar signal into four separate frequency bands (lo, Mid 1, Mid 2 and Hi), distorts each band individually, and then sums the four bands to create a composite sound.

PAiA still sells a kit, but it's a rack-mount unit with a bunch of bells and whistles:
http://www.paia.com/proddetail.asp?prod=6720K
It could definitely stand a little stripping-down for use in a stomp.