3dB / octave filtering for overdrive

[Steben]

Did it ever come to mind that an even gentlier slope than 6DB/oct could help in shaping a mild overdrive tone?
This a generic way of getting a 3dB/oct LOW pass filter to get pink noise out of white noise:



What if one uses these values by reversing the filter?

[ElectricDruid]

Why would you want to reverse the filter? That'd give you a highpass, wouldn't it? Am I misunderstanding you?

I think the idea of using a "pinking filter" (that's what they're often called) for an overdrive is rather good. Yes, a subtle rolloff of high frequencies is often exactly what we want, and a gentler slope is a good way to achieve that. Give it a try.

[Vivek]

All Cabs and Cab Sims have very steep roll off in the highs.

For which application would we want a gentle roll off ?

[Steben]

Why would you want to reverse the filter? That'd give you a highpass, wouldn't it? Am I misunderstanding you?

I think the idea of using a "pinking filter" (that's what they're often called) for an overdrive is rather good. Yes, a subtle rolloff of high frequencies is often exactly what we want, and a gentler slope is a good way to achieve that. Give it a try.

Highpass pre eq, lowpass post

[Steben]

All Cabs and Cab Sims have very steep roll off in the highs.

For which application would we want a gentle roll off ?

Pre eq, gentle softening of lower freqs

[ElectricDruid]

-6dB/Oct *is* already pretty gentle, tbh. If you regard -60dB as "quietened down, but not gone", then it takes you ten octaves to get there - that's basically the entire audio spectrum! So a 6dB HP filter at 20KHz wouldn't *completely* cut a loud signal at 20Hz!!

I should try that, just to make sure, but the theory is sound...

Still, -3dB /oct is another option, and maybe it's good. There's a lot to be said for experimenting with stuff that hasn't been done much before.

[Rob Strand]

What you want is the reciprocal response.

There's also a gain problem, firstly the ideal pink response doesn't have a DC so the gain needs to be defined at some point say 1kHz.  The circuit does have a DC gain because it's not pink below some low frequency.   As shown, the second is the pink noise filter in the circuit has a loss, since it's a divider.   The true reciprocal response would undo that loss with gain.  In practice you would consider the gain of the each section separately.

For a pedal you might also want to limit the region where the pink noise filtering applies since applying it to some parts of the spectrum might not work out so well.  And in this case you might also realize best sound won't be a circuit which is the reciprocal but something like the reciprocal with some tweaks (like at each end of the response).

Here's a general way of making a reciprocal response.

For my circuits below, it should be obvious  H -->  T is unity but also T -->  H is unity.  For a pedal you will probably be using, T --> circuit --> H.  As T has the treble boost.    (You could jig it the other way as H --> circuit --> T but it's probably an unnatural set-up)

For your circuit R1 = 68k,  Z2 = 30k + 100n through to 4n7.

[fix:  in (f) I say Z2 needs to be resistive at DC, that's not actually required.)

FWIW, you will already find pink-noise circuits which use the feedback form and not the divider form.





As another example of reciprocals.     Most people know the bridged-T circuit  can be used for a mid cut.

https://www.muzique.com/lab/notch.htm

If we take the bridge-T circuit and put it in a feedback loop (as shown in the notes), then you will get a mid-boost - the reciprocal of the mid-cut.    If you then move the peak frequency up from mids to treble you get a peaking treble boost.   The idea was used in this old ETI magazine project,