Tone Stack profiling (extended profiling)

    Ok, so in the past I thought it would be great to just be able to take a few profiles, set them at certain control positions and then have the Kemper morph between them as a way to emulate the real amps controls....


    ...I now have a better idea. And I think this one might be more feasible for Christoph and his team to implement - Extended Profiling.


    Here’s the idea, let’s start with the bass tone pot, there could be a phase of profiling where you are promoted to or select “profile bass tone control”, the Kemper can send out just a simpler either white noise or other impulse and you would be required to set the bass tone pot to its minimum, press a button to start, dial the pot up to max then press a button to stop the profiling. The profiler could then determine the frequency mid, the q, the min and max throw and possibly even give a guesstimate of the taper (though perhaps that would require more user input, perhaps letting the unit know the middle 12 o clock position would be enough) given that nothing else was changing.


    Repeat for all the tone controls in the amp, maybe even do the same for the gain pot to just grab the frequency shifts as gain goes up and down. And perhapss even have the option to mark or hide the controls that don’t actually exist in the amp, e.g. not all have presence controls.


    Voila - yes a little more (optional) work, but a much closer profile and set of working controls. I know that I’d be willing to go through this process to build my profiles as accurately as possible, it’s not much more than doing a merged profile really. And I think the info wouldn’t bloat up the profile file much or be very costly performance wise as it’s just setting ranges and curves of controls rather than trying to fully morph between different profiles.

    Quote from Monkey_Man

    This is pretty much exactly what I've proposed quite a few times in the past (I suspect others have too), Per, including prompts from the KPA at each stage of the process, so I agree.


    IOW... +1, bud.

    I've requested this sort of ability in the past, the difference is the approach being suggested this time. Previously I thought making multiple profiles was the way to go but now I think that linear interpolation between profiles would still be great, but for custom controls then it'd increase profile size a lot, be a lot more work for the processor, and be super difficult to actually use to extract control settings from.


    The idea here really is to hybridize component modeling and profiling which has a heap more benefits.

    Yeah, there is that distinction; thanks Per.


    I do get it. It'd be not-too-dissimilar to the approach Don takes when "modelling" wah pedals using white noise.


    Quote from old crow

    Matching tone stacks would be the ultimate feature .
    Imaging only needing one profile for an amp.
    Has Mr. K ever commented on it ?

    Not that I'm aware of, OC, but he may have of course...

    How about an even simpler start... An option that allows the user select the frequency and Q for each EQ control. A little internet research would very likely turn up everything we need to know about the behavior of the EQ section of our favorite amps. That information can become various presets for the block. This, combined with the already available pre/post option, might get us headed in the right direction without having to redesign the profiling engine. This approach could also be applied to the Monitor and Main output modules to further enhance our ability to fine tune our FRFR’s and FOH EQ. If this approach is well covered territory, I’m often late to the game, dinner, doctors appointments, family gatherings, work... But never to the gig!!

    Fats


    Art is not a competition

    Quote from Vaughnfats

    How about an even simpler start... An option that allows the user select the frequency and Q for each EQ control. A little internet research would very likely turn up everything we need to know about the behavior of the EQ section of our favorite amps. That information can become various presets for the block. This, combined with the already available pre/post option, might get us headed in the right direction without having to redesign the profiling engine. This approach could also be applied to the Monitor and Main output modules to further enhance our ability to fine tune our FRFR’s and FOH EQ. If this approach is well covered territory, I’m often late to the game, dinner, doctors appointments, family gatherings, work... But never to the gig!!

    i think the problem(s) with all of this is the fact that even knowing the frequecy centre and the Q factor of ech control would fail to take account of the (often) massive interaction between controls lt alone their effect of gain when placed pre the clipping stages.


    Have a play with the duncan tonestack simulator and you can see this effect graphically. In fact this tool lets you build any amplifier tonestack by substituting compnent values and watching their effect. Basically figuring out the characteristics of any amp is childs play to th guys at Kemper. However, I believe the true genius of the KPA and the reason it has survived for something like 7 years without the need for a procssor upgrade is that the amount of real time processing involved in profiles is tony compared to component level modelling. Therefore trying to create tonestack presets that work like the real amps could be the start of a journey down a very deep rabbit hole that ends up needing processor updates like all the existing modellers. - I could be totally wrong about the processing requirements as Ikm not a programmer or IT pro !!!

    @Wheresthedug


    I totally agree with your point that the interaction in the EQ stack is a formidable challenge to duplicate. After reading your post it occurred to me that another serious flaw in the parametric solution would be that, especially in an amp that places the EQ post gain, the effect of boosting or cutting the frequency range also affects every component in the output section.

    Fats


    Art is not a competition

    Like 1

    I understand the wish that profiling also captures the dynamic characteristics and EQ behavior of an amplifier and you get something like a clone of the amplifier.


    Even if it is legitimate to ask about it in the FR, I am afraid that it overtaxes the possibilities a little bit.


    In my opinion it contradicts the philosophy to replicate a certain state of an amplifier sound and to process it with general (not amp-specific) parameters.
    I would say that the profiling process would have to be redeveloped and it did not seem to be a small change.


    At the BBC Talk a participant asked CK something similar and CK pointed out that it is not intended manage more than one profile at the same time.

    I don't think that this will ever be technically feasible.
    At least not with the precision a normal Profile has. If you want to profile a normal 3 knobs EQ and the gain with the profiler you need to adjust every single parameter individual.
    Now we need the convert the continuous movement of the knob into some kind of digital representation.
    Let's just choose 128 options for it (like Midi)
    That should be enough.
    So we got 128^4 options that the profiler has to profile.
    Those 268.435.456 profiles should now be stored in one profile and change by the move of the knobs on the Kemper?
    That's just too much data.
    It also would take forever to profile such a thing.
    Remember this is just if they want their eq-profiles to use the same method of profiling.


    I actually believe that the way they are profiling right now can't be easily changed to add some EQ-Profiling or Knob-Profiling to it.


    I also don't see massive benefits in this sort of profiling. There are just to many variables in an amp. There are amps with a depth control and extra presence knobs.
    The profiler would need to know all these variables to make the product appeal to the public.


    There are other digital solutions that let you control everything up to the smallest details but in my opinion they don't sound as good as the snapshot of a real amp that the Kemper captures in all of its glory.


    And the tone control that is in the Kemper is also very usable IMO.

    I think you're not following the idea there MonkeyPeanutButter.


    You don't need to profile combinations or even make multiple profiles - I specifically said this wasn't a good idea in my post.


    The idea here is about profiling specific controls for known parameters. You only need to store a frequency, q value, range and possibly taper curve for each control. That's at most 3 controls because gain and presence don't work the same way as an EQ, at tops that comes out to 4 floats and maybe 256 char values though I suspect the taper of a pot can be described in simpler terms as a quadratic, so even less information.


    Gain and Presence would be as simple or complex as you want. I'd imagine that just measuring the frequency shift over gain would be enough and the most complex thing as that means storing at most a matrix of e.g. 256x256 char values, however you could simplify the frequency curve and even linearly interpolate fewer sample points along the control. Presence is about harmonics, I've no idea what info would need to be stored, but I suspect it's static frequency data not dissimilar to the EQ controls.


    The idea is about how to profile a whole amp's tone section without doing many profiles, and without storing a huge amount of data.

    I don't think this would be possible:


    The magic of profiling is that it captures one particular amp setting, cab, mic placement perfectly. To replicate that, without sacrificing the perfection of Profiling that amp modelers fall short of, you'd need to capture a profile with every possible combination of tone knobs.


    Suppose your amp had just three tone knobs and gain. Gain, Bass, Mids, Treble. And suppose you wanted to capture the placement of these four knobs, say in 10 places from one to ten. Starting with, for example G1 B1 M1 T1, then G2 B1 M1 T1, then G3, B1, M1, T1, then continue, all the way up to Gain on 10. After those ten, do the same thing again with bass set to two, then three, then four....Then the mid knobs, treble knobs, then put the Bass back to 1 and do the same with mids on 2, then.....You get the idea.
    Mathematically, I think I'm correct, to do this, you'd need to profile 10x10x10x10=10,000 profiles. If you added a presence knob you're at 100,000. My math may be way off, but at the very least, you'd be looking at capturing hundreds of profiles to capture the amp with all the combinations of knob setting. In any case, you'd be looking at having multiple profiles loaded at once while blending multiple ones at once with some sort of crossfade, which in itself would probably alter the tone and feel.


    A better compromise might work like this: You capture the profile, and once done, if you choose this option, the KPA asks you what position each of your gain, treble, mid, and bass, and presence knobs were at when you captured the profile. So when you call up the profile, the default position of the knobs shows where your real amp was at when you profiled at, instead of all at 12:00. Then the KPA has a handful of EQ types built tin hat function similarly to the EQ stacks of handful of common amp types that you can select. Then perhaps these EQ types could more closely mimic what would happen had you done so to the original knob. My guess though, is that this would still be a compromise and more trouble than it's worth, vs. just making profiles of the handful of ways you actually you actually dial up and use on the amp. Maybe the ability to have two profiles loaded at once and being able to crossfade between them?

    @Grooguit - You're doing the same thing that MonkeyPeanutButter did (re-read my original post and my response there).


    There's no need to make as many profiles as there are combinations of settings. You simply have to make some assumptions - namely that EQ controls are really just that.


    As you change those settings they don't change frequency, they don't change q, they just change the gain for a set of frequencies at a certain point in the circuit within a range of strengths.


    Of course it would help to know if the tone stack is pre or post the preamp but you can set that manually right now in your Kemper. Otherwise with a few notable exceptions they're predictable things, virtually linear, which makes them easy to evaluate. You don't need to sample combinations of them, you only need to determine the frequency, q and throw of each control, provided you don't change other settings as you adjust those it shouldn't be that difficult to determine these values.

    Hmm I skimmed over those; I should read more :) It would be cool if it could be done though. In particular, I think being able to bump up the kemper gain on a profile more before it starts sounding harsh and off would be the most useful aspect. I've personally been away from using real amps for so long that the paradigm of achieving good tone for me has morphed into finding approximate profiles for what's in my head and mildly tweaking them the rest of the way there vs. playing around with a real amp and combos of overdrive pedals and hoping that I can tweak them to match the sound in my head.

    The only problem is that tone controls aren’t linear by a long shot. The frequency center and q of valve amp tone stacks are interactive so changing the treble actually moves the middle to a different frequency as does moving the bass. If you haven’t already done so try playing with some common tone stacks in the Duncan Tone Stack Simulator. Or try building the Mesa Dual Rectifier stack from component values in the program.


    I suppose Kemper could model thos behaviour but there are already modellers out there that do this extremely well. However the downside is processing power.

    Quote from Wheresthedug

    The only problem is that tone controls aren’t linear by a long shot. The frequency center and q of valve amp tone stacks are interactive so changing the treble actually moves the middle to a different frequency as does moving the bass. If you haven’t already done so try playing with some common tone stacks in the Duncan Tone Stack Simulator. Or try building the Mesa Dual Rectifier stack from component values in the program.
    I suppose Kemper could model thos behaviour but there are already modellers out there that do this extremely well. However the downside is processing power.

    Even if that were the case it's still not really a problem, though I find that highly suspicious, I'll try it with my own Mesa some time but as far as I've tried it just operates by having the EQ before the preamp, the frequencies don't appear to change.


    In any case having a curve for frequency center and q for all three controls wouldn't add much either to processing or file size as far as I can see.

    Even the mesa manuals talk about the interactive nature of the tone controls. In fact they even specify that the gain pot also has an effect on the tone getting brighter as you lower gain and darker as you increase it.


    Some people go on about how difficult mesas are to dial because of their interactive tone circuit but the truth is it’s exactly the same cuircuit as all other TMB tones stack designs which go right back to pre Fender. The TMB stack is also often referred to as an FMV tone stack. FMV being Fender Marshall Vox.


    Honestly the best thing to do is down load the free Duncan Tonestack calculator and play with it.


    http://www.duncanamps.com/tsc/

    Okay say it is as simple as just generating one continuous test tone that gets send while the person profiling is turning the knobs.
    How should the profiler tell the person or vice versa on what position the knob is or has to be?
    Maybe you can get close to it with having a few stable (0 5 10 although you would probably need more) positions and then you would create (English is not my first language so please excuse this unscientific term) a mathematical function based on the specific frequency response of these points.
    Of course the method gets more precise for every point that you add to the input.
    If tone stacks are this easy (referencing how they influence the tone) then it would maybe work out.
    My point was that this would need to be a very new thing that Kemper would have to implement.
    I'm sorry for missing your point @Per
    I'm more of a passive reader here and really enjoy your posts.