1. We're currently conducting systematic tests with the release version of macOS 11 (Big Sur) in order to offer full compatibility of our products as soon as possible. You can find our latest compatibility status - here


    Please note that although announced at the same time as macOS Big Sur, Apple Silicon processors are an independent release with separate compatibility concerns for NI products. Compatibility with macOS Big Sur will not yet ensure compatibility with Apple Silicon processors. At the moment, Native Instruments products are not supported on computers with Apple Silicon processors. Read more - here

    Dismiss Notice

How Hard Would It Be To Model These Distortion Pedals In Reaktor?

Discussion in 'REAKTOR' started by mike_jm, Sep 6, 2017.

  1. mike_jm

    mike_jm NI Product Owner

    Messages:
    281
    Two of my favorite distortion pedals are the ProCo Rat (used by Nirvana, Radiohead, and endless other bands) and the Opamp Big Muff Pi (used by Smashing Pumpkins on Siamese Dream and Mellon Collie & the Infinite Sadness).

    They're both pretty simple pedals. Here are the circuits...

    ProCo Rat:
    [​IMG]
    [​IMG]
    "The design is based around the LM308 single op-amp. The distortion is produced using a variable gain circuit with diodes clipping the waveform. The distortion stage is followed by a tone filter and an output buffer stage that ends up with a tone control."
    https://www.electrosmash.com/proco-rat

    Opamp Big Muff Pi:
    [​IMG]
    [​IMG]
    http://www.pisotones.com/BigMuffPi/psst/BMP_versions.htm

    I've actually built both of these physically with soldering in the past few years from PCBs. But I don't have much education to help me understand how they electrically work enough to translate their circuits into Reaktor design.

    For an experienced Reaktor user would these be difficult to model? I'm willing to pay someone to do it for me as contract job. If it wouldn't be too hard, it's probably worth me just having someone who is experienced do it. If it's very simple though, maybe I could do it myself with a few pointers/tips/directions.

    Thoughts?

    Thanks.
     
    Last edited: Sep 6, 2017
  2. colB

    colB NI Product Owner

    Messages:
    3,399
    Yes, very difficult.
    There's a reason there are very few 'simulations' of distortion units, and it's not because no one's thought of it yet ;)

    You also have to remember that even a real RAT pedal sounds **** if you DI it. It only produces an acceptable sound when it is used to drive a valve input stage, and that in turn only sounds particularly good when it drives the output stage hard which is pushing a transformer into saturation and driving a good speaker hard... (assuming you want some musical distortion rather than a bell like clean tone)

    So even if you were to achieve a good simulation of a RAT, the result would be underwhelming IMO.
    .....

    The schematic you posted for the Muff is the wrong one. The Muff Fuzz is not the same thing as the op amp big muff pi - that is significantly more complex. It's got multiple back to back diode clippers in one of the op-amp feedback paths combined with a filter... lovely ;)

    Have a look at the papers at cytomics website to see just how complex it is to simulate a truly simple circuit, and you might start to understand what you are asking here.
    here's the most basic one, just two very simple filters with no saturation modelled. The first one is passive with just has a single resistor and a single capacitor, and it's already getting a little tricky

    Introducing distortion complicates things significantly. Then on top of that there is the problem of aliasing which is also very difficult to solve efficiently in the context of distortion.

    You would probably be better with a black box solution with some spectrum analysis of the circuit section and using filters and saturators to give a similar overall sound. It will never be a true simulation though...

    I would love to see a good emulation of a fuzzface, which is a very simple circuit indeed, but again, it's a really REALLY difficult thing to do. I'm a guitar player myself and have built lots of fx boxes over the years, I've never yet played a digital simulation of analog guitar gear that convinces me. A good valve amp being driven by a nice (transistor ideally) fuzz or distortion is special. The simpler the circuit, the more important the feedback and interplay between the individual components is, and the more important the parasitic features of the individual components becomes. Very difficult to simulate.
     
  3. mike_jm

    mike_jm NI Product Owner

    Messages:
    281
    As a point of reference, here is a project "Semi Fuzz" by Trevor Gavilan which is an incredibly powerful sounding transistor fuzz:

    https://www.native-instruments.com/en/reaktor-community/reaktor-user-library/entry/show/7409/

    The design is actually very simple though in primary with just a bit of core for modeling the transistors:

    fuzz.PNG
    fuzz transistor.PNG

    Since the clipping in the pedals I posted is happening at Opamps, I'm guessing they are really the only thing of any possible difficult (in place of the transistor here) that needs to be modeled.

    Could this be done with a slew limiter based on published specs and maybe some other built in modules?
     
  4. colB

    colB NI Product Owner

    Messages:
    3,399
    Have you downloaded Robin Davies too-b overdrive? that is about as simple as it gets while still sounding good, and it's not attempting to be a simulation.

    ...The fact that the 'Driver' Block has its core process locked is another clue as to how difficult it is to achieve an interesting and musical sounding distortion - something good in this area is rare and worth protecting.
     
  5. mike_jm

    mike_jm NI Product Owner

    Messages:
    281
    Hey ColB, looked like you replied as I was posting. Just saw your post.

    Thanks for the correction on the Opamp Muff. I will fix that in the OP.

    Actually I already have a very convincing Rat simulation VST. I absolutely love it. More than a real Rat pedal actually because it's cleaner with less hisssssss. You can download it here:
    https://www.tseaudio.com/software/tseR47

    The same guy also made an 808 overdrive VST:
    https://www.tseaudio.com/software/tse808

    As above, I posted a beautiful transistor fuzz someone made in the Reaktor library that is amazingly simple. So clearly there are ways in Reaktor to do elegant distortions that sound very good. I'm hoping I can learn from that how to do an Opamp distortion.

    I also have to say perhaps I'm alone, but I LOVE the sound of solid state distortion on guitars! Tubes are okay. But most of grunge was made with heavily pushed pedal chains. I like the grind and clarity of a solid state distortion pedal. Amp sim following or not. :)

    I will look at the other pedal sim you just suggested also.
     
    Last edited: Sep 6, 2017
  6. mike_jm

    mike_jm NI Product Owner

    Messages:
    281
    That too-B is GORGEOUS as well! Thanks. I will use that a lot. Good find. Running the built in Reaktor Overdrive (which is also nice) into too-B creates very powerful sounds.

    This is the main distortion element of the built in Reaktor Overdrive also for reference, which just seems to be some "Lin Smoothers" I think to smooth out the control knobs, run into a high and low pass filter and then through "tanh Sats":

    od lin smooth.PNG

    OD.PNG

    Not much to it and it sounds very much like a lot of overdrive pedals.

    If the Lin Smoothers are what I think they are, they're not even needed and this would become incredibly simple but effective.
     
    Last edited: Sep 6, 2017
  7. salamanderanagram

    salamanderanagram NI Product Owner

    Messages:
    3,446
    i wouldn't try to model the circuit, i would try to model what the circuit is actually doing. probably somebody has done a pretty decent job of writing this out already.

    here's a source i found in a few minutes:

    www.electrosmash.com/big-muff-pi-analysis
     
  8. mike_jm

    mike_jm NI Product Owner

    Messages:
    281
    Further on the subject of already existing distortion Reaktor units, just found this old gem which actually is so comprehensive and detailed like many commenters are saying on the post there, it beats every amp sim I've tried:
    https://www.native-instruments.com/en/reaktor-community/reaktor-user-library/entry/show/2927/

    What's great is the level of control it gives you. Also great sound design by the maker in terms of presets.

    I think I will try to institute the upsampling/downsampling from "too-b" into this. The maker of "too-b" said it made a big difference in improving his simulation ...

    I think maybe I'll just do that and stop there.
     
    Last edited: Sep 6, 2017
  9. colB

    colB NI Product Owner

    Messages:
    3,399
    That's nice, but it's the wrong circuit. Mike is looking for the op-amp version. There are some significant differences.
    In the end though A black box approach would probably be better - get a unit, and do some tests - frequency responses from different stages of the circuit etc. then use filters and waveshapers to model it.
     
  10. mike_jm

    mike_jm NI Product Owner

    Messages:
    281
    Yeah I'm starting to see that from all the projects so far. No one is doing modelling of the individual circuit components (except that Fuzz I posted where the transistor was modeled). Everyone else is just running distortion with things like tanh or other processes and tweaking the settings to get the rough sound they want...
     
  11. colB

    colB NI Product Owner

    Messages:
    3,399
    That's strange, because I spent some time playing with TT-amp, a few years ago while working on my own experimental amp sim project. And while I agree that it is comprehensive in its features, I couldn't find a single satisfying sound when compared to even a cheap real amplifier. Even my solid state pignose amp with tiny speaker sounds better on pretty much every setting than TT-amp can on any setting I could find (and I'm pretty good at dialling in a sound).

    I suppose everyone has a different Idea of what sounds good, but there are too many shortcomings with digital modelling when trying to simulate something as chaotic and organic as the dynamic nature of guitar distortion stomp boxes and amps.

    I shelved my project because I was never satisfied with the result, you should take a look though, might be of interest even if just from an academic point of view: https://www.native-instruments.com/en/reaktor-community/reaktor-user-library/entry/show/7063/

    I actually need to put a bug fix in the cab sim (I think it never got uploaded), so I'll try to do that soon.
     
  12. mike_jm

    mike_jm NI Product Owner

    Messages:
    281
    I think I can venture a guess as to why. I am going for extreme distortion levels for which the TT amp which is based on Mesa architecture excels at. It can create massive amounts of distortion without the typical garbled "break up" you get in the high frequencies with such levels of distortion. Ie. It still sounds smooth. Think the heaviest metal distortion (though I'm not actually a metal head - it's just something I want to be able to achieve).

    I am guessing you were going for lower/medium/normal levels of distortion. Agreed if so that it does not sound nearly as good on clean sounds. Kind of sounds like a solid state amp on those cases.

    I need cleans and light distortion as well but I have found they are done usably well enough by many sims like amplitube. It's the ultra high gain stuff I've always had trouble finding a great effect for. Most things come out with high gain just sounding garbled.

    Also TT amp does not utilize cab impulses which would be a shortcoming to some but I don't actually like the sound of a guitar cab. A guitar cab in my opinion is just a poorly conceived low pass filter with some low quality distortion, low frequency boom, comb filtering, and reverb which in my opinion only degrades the sound and makes it sound muffled and more distant. I also just dislike convolution on principle and prefer things to be synthesized from the ground up.

    Can sims are "necessary" in my opinion with poorly built guitar amp and sims that create far too much high frequency fizz or chaotic/garbled/unmusical break up in the high frequencies. With a well executed sim/amp in my opinion a simple daw eq lpf and reverb sounds cleaner and preserves the immediacy and coherence of the signal.

    I am less interested in replicating a real guitar amp/cab than I am getting strong and interesting sounds. I have a Marshall stack in my basement if I need it but I don't need that hisssssss and degradation.

    Again I would definitely agree if that's what you meant from first impressions the TT doesn't do a good/rich clean or medium distortion. Probably something to do with the archicture. Mesa design was never renowned for the richness of its cleans or light crunch. That's more Marshall/Vox/etx territory. But whatever makes it duller on cleans I'm guessing is also what makes it rip very hard without falling apart on ultra high gain...

    I will check out your project for sure. Thanks. No worries about the cab sim. I'd probably be turning that off anyway. :)
     
    Last edited: Sep 6, 2017
  13. colB

    colB NI Product Owner

    Messages:
    3,399
    My understanding of high gain amps is that they heavily filter the incoming sound to reduce the harmonic richness, the richness is added back in with distortion, which is filtered again before more distortion at each amplification stage. This allows the harmonic structure to be controlled without becoming messy/muddy. They also have tighter rectification with minimal sag, again to prevent muddiness and mess in the sound.

    I guess you could do something similar with a few simple waveshaper stages and filters.
     
  14. colB

    colB NI Product Owner

    Messages:
    3,399
    That's odd, because the heaviest metal distortion is not something you would use a RAT or a muff for normally.
    The RAT is used more often like tube screamer is used - to boost the level with a bit of filtering in order to drive the front of an amp harder. A huge amount of the sound also depends on the amps cab and speakers.
    Muff is more of a fuzz than a distortion - again not a metal high gain thing really. It has lots of high end fizz and fuzz, and creates a mess out of whatever you throw at it - definitely not a tight metal or mesa boogie kind of thing at all.

    In the digital realm, I suspect, at least some of the garbled break up you get in distortion processes is because of the aliasing they generate. first stage distorts, adds high end harmonics but also adds inharmonic aliasing distortion. Second stage distorts, but in addition to the musical stuff from the first stage, it also amplifies the aliasing artefacts, and creates new ones from the musical stuff and from the first stage aliases... By the time you get to a third stage, any higher frequency content from the input has generate huge levels of horrible aliasing mess that ruins the sound. I suppose this is why TT-amp can do the mesa type of thing the filters most of the high end away... It also explains why a good fuzz like the muff is difficult to do - it excels in exactly the high end distortion that is worst for aliasing.
    There are some techniques that can mitigate this, but they are all challenging.
     
  15. mike_jm

    mike_jm NI Product Owner

    Messages:
    281
    Agreed on everything, colB. The opamp muff I just wanted to run into a jcm800 style sim to see if I could catch that Siamese dream sound. But at this point it's not critical. It's more about having a broad range of usable sounds which I think I now have satisfactorily. Guitar Rig and Amplitube have a fair number of stompbox sims as well. Plus the vsts I already have and using reaktor through reaktor fx to chain and I think I'm taken care of in this regard. More important things to focus on instead.

    Either way this was all a very informative and helpful discussion. Will still check out your sim later.
     
    Last edited: Sep 6, 2017
  16. colB

    colB NI Product Owner

    Messages:
    3,399
    Just had a play about with Efflam Le Bivics 'infinite linear oversampling' concept, and although it is a bit glitchy, it's possible to get it sounding pretty good, I also surprised myself by managing to extent it to tanh, so I might upload something if the experiments end up as something worthwhile.

    The infinite linear oversampling is a relatively cheap way to avoid a lot of the aliasing that otherwise causes problems with distortion in DSP. So it seems to be able to get a little close to fuzz territory... We'll see I suppose.
     
  17. salamanderanagram

    salamanderanagram NI Product Owner

    Messages:
    3,446
    using polyblamp to round corners is my favorite method for anti-aliased distortion.

    i feel like the essential nature of the siamese dream sound is really that there are like dozens of guitar tracks layered on top of each other. i mean, sure, there's distortion and all, but it's really the studio production that made it sound so big, IMO
     
  18. Kostas K

    Kostas K NI Product Owner

    Messages:
    23
    my 2c on this:

    If you want to model a distortion pedal, the simplest approach you can take is implement a waveshaping function, along with a pre/post filter. That's trivial (there are already tons of modules that do this). It's not very faithful though.

    If you want to go into the "proper" circuit modelling route, things can get complex very fast: you essentially have to solve the circuit for each input sample and output the solution. That's quite demanding, because you'll end up with a non-linear differential equation system which can only by solved with iterative methods (making a guess of the solution, feeding that back to the system and using it to do a better guess, etc etc in a loop).
    Now, doing that in Reaktor is a real pain in the a**, because it invloves endless copy & paste in a spaghetti nightmare of wires. I don't know if some methods are even possible.

    I've done some research and implemented a simple diode clipper with the second method. It's here: https://www.native-instruments.com/en/reaktor-community/reaktor-user-library/entry/show/11099/
    You may find it useful. It's the most simple of circuits, but still, making it in Reaktor was quite an uncomfortable experience. I don't think Reaktor is meant to do this stuff (yet).

    If you're in the mood to read, check the academic papers of David Yeh: https://ccrma.stanford.edu/~dtyeh/papers/pubs.html
    He's the defacto source of circuit modelling knowledge.