I put this together in response to a post by @jbaptiste64. He asked if I could modify the 16 band equalizer into something resembling the Serge Resonant Equalizer. The Serge design is a 10 band equalizer where the band spacing is a seventh apart rather than the more traditional octave or half octave spacing. Additionally, in his design the band becomes “resonant” when the gain is set to less than 25% or more than 75%. I’m not really sure exactly what this means in this context. Typically filter resonance is characterized by a response peak at the center or cutoff frequency. Since equalizers by design provide a gain peak at the center of the band, they are all resonant. With a peaking eq style of filter, the bandwidth and Q (which is one measure of resonance)are inversely related so, in the absence of any detailed information, I chose to assume that in this case, resonance means a narrowing of the bandwidth of the band. As the band gain increases past 75% or below 25%, the bandwidth will reduce smoothly to half it’s normal value. Without feedback, this occurs at the +/- 12dB point. Maximum gain or cut is +/- 24dB. The filter is divided into two cascades, each of which has a separate “comb” output as well as their combined output. I also incorporated a feedback circuit and output level control. Because the filter is capable of significant gain, particularly when the feedback is set at high levels, caution is advised. I believe that I have kept the maximum feedback low enough to prevent instability, but it’s hard to be certain. I’ve already damaged a speaker in my Audulus career, so I’ve learned to be careful.
Update
I’ve updated the version to 1.2. I added a tanh() saturator to the outputs to provide limiting and allow for a higher level of feedback. You can now crank it up and get distortion rather than instability. The limiter is in all three outputs and is before the feedback loop, but is before the final gain control so the output gain doesn’t affect the feedback. It’s is now possible to drive the filter into self-oscillation with high enough feedback.
1.3 - Added input gain control and phase inversion for feedback
I/O
Input
Signal Range
Notes
-1 to 1 audio
Output
Signal Range
Notes
-1 to 1 audio
large signal gain is possible - use caution
-1 to 1 audio
“comb” output from each filter cascade
Controls
Control
Function
Notes
band gain
sets gain from -24 to 24 dB (without feedback) at the indicated center frequency
feedback level
can significantly increase maximum gain and decrease minimum gain for each band.
replaced the tanh() saturator with an asymmetric version closer to a JFET or triode. Probably not a good idea to remove the DC Blocker as the asymmetric distortion introduces a DC offset to the output. The drive to the saturator can be adjusted if desired.
Thank you very much for taking the time to make this module. I get very interesting results especially using the Audio Input module with surrounding noises. If I wanted this module it was to reach the sounds in this video of Todd Barton. Only I can not even copy the position of the knobs, I guess the Erb Verb is a central point to get this sound and even with audulus reverb it does not works.
I just started using Audulus, I would just use this module with a reverb but I can not get out of it with the feedback thoroughly, I have no sound. I enclose a photo for you to see my patch. If you can tell me what’s wrong and if you have some tips for getting the sound of the videos I would appreciate it. thank you in advance
It’s nothing you’re doing wrong. I did notice that my frequency bands are reversed from the RandomSource Eurorack module. The main issue lies in the nature of the feedback loop. I have mine clamped down pretty well to prevent run-away oscillation. Without some form of limiting, a digital feedback circuit that has more than a unity gain will rapidly output the highest level that the hardware is capable of. As I mentioned above, I’ve already blown one speaker doing something similar, so I’m a bit gun-shy. Analog circuits are inherently more self limiting so self oscillation is more controllable. I may be able to increase the levels so that the filter self-oscillates more readily. You will also need some kind of input. The analog unit always has some noise to amplify, but the digital unit is noise free. If I can get it to oscillate more dependably, you will need to feed itsome noise or other signal to kick things off. Also the feedback phase control is backwards. I forgot that the tanh() function inverts the signal. I’ll post a corrected version.
Try version 1.5. I think you’ll find it oscillates fine on it’s own. You may not be able to exactly duplicate the sounds from the Serge unit because there is a necessary one sample delay in the feedback loop which affects the feedback and the tanh() saturator doesn’t distort in the same way as the electronics in the analog unit. I may try a different saturation approach if I get a chance.
Here is an improv that I did using the equalizer to add feedback to my Buchaudulus (which I’ll share when I can get it into less messy shape if folks are interested):
Thx for the interest. I am close to having a draft version that might be usable by someone who isn’t me. Here is a “feedback bongo” loop made with Buchaudulus and the resonant equalizer and a goofy spatial modulator made from a couple of StSchoens uVerbs
Thanks, I’m glad you’re enjoying it. Like many of the things I’ve posted, the original idea came from another Audulus user @jbaptiste64. He asked if I could build something close to the Serge unit. I think this comes close, although self oscillation in Audulus is somewhat tricky. You need to have some type of limiting in the feedback loop to prevent run-away oscillation. I has fun creating some weird sounds when I first put it together. I’ve enjoyed your patches. You should post the Buchaudulus, I’m sure it would be very well received.
I’ll definitely post Bucha… I want to add some comments into the patch for those that want to dig around and hack it some more and put together some notes on how to use it. The user interface is sketchy and things aren’t as well-labeled as they could be (I am very inefficient rearranging the u.i. and labels). I am to have it shareable in the next week or so now that I feel like the architecture is stable.
It is very much a frankenstein’s monster. Most of the important chunks were created by other people (you, @robertsyrett , @RudigerMeyer) and wired together.