Maybe a path for the future of modelers?

[MKB]

Right now it seems that we are stuck with models of existing amps considered the pinnacle of guitar tone. Modelers are starting to give some interesting ways of tweaking these amps, but we are still stuck with specific models. Players beg for more models to be created, but these are still copies of existing amps, with no extensive ways of changing them or creating custom models.

How about this: Instead of modeling specific amp designs, model specific capacitors and resistors, transformers, the whole gamut of parts to make an amp. Then create a interface where a sound designer could build an amp with these virtual components. For those of you familiar with the electronic software Spice, perhaps something along those lines.

A further enhancement would be to allow a sound designer to create his own component types (different capacitor dielectrics, carbon comp vs. metal film resistors, etc). The designer could use his custom or the stock virtual parts to create his own amps, and tweak them to sound the way he wishes. These models could then be sold to the general public.

The great thing about this is a sound designer could then become a virtual Dr Z or Dumble, in that he could design his own virtual amps. This would also allow new amp designs to be created, and we won't be stuck with whatever real amps Line 6 or Fractal has decided to copy. Crazy stuff that hasn't been or rarely been tried before could be put together, like using a EL84 as a preamp tube, or interstage coupling transformers, or an output stage with 16 KT88's per side.

Such a platform would allow sound designers and modelers to advance the amplifier art, giving new and potentially better tones than what is available through discrete component amps.

If this has been mentioned before or has already been attempted, my apologies and/or please disregard the above.

[Lution]

I like this idea!

I think as long as it had a very informative guide to help a user get started. Perhaps an optional Tone Component Chart of some type that would walk a user through the process of creating the amp and which components created particular sounds/tones if they wanted it.

[guitarnet70]

I love that!...but I will need your designers to make some work (even prepared to pay for it, like RW or OH) for me, because like the majority of the guitar players I have no cue how to build an amp....

[Elric]

Great idea but, FWIW, it's not new... I've seen it kicked around many times in different contexts.

Revalver is fairly close to a build your own amp model kit, actually. You cannot skin tweaked models but you can save them and some users have created schematic based modified default models for other users. The amp tweaking capability of that tool is ridiculous. It also has individual preamp tubes and separate power tube amps broken out based on topology the whole nine yards.

That "user model" type of paradigm has been discussed and requested of the developer numerous times on their forum but...

The main problem with that paradigm is that the manufacturer loses control over the model set from a business standpoint. Additional models are a source of revenue and an upgrade incentive for future products it has the potential to affect their revenue stream if they are not careful.

[Gasp100]

Many oeople on here state they are already sick of tweaking modelers like the axefx to get "their tone". I dont know how popular something like this, but seems interesting.
I dont thunk i would have the patience myself.

[MKB]

Many oeople on here state they are already sick of tweaking modelers like the axefx to get "their tone". I dont know how popular something like this, but seems interesting.
I dont thunk i would have the patience myself.
Great point, one problem with modelers now is you are somewhat forced to tweak them to get good results. The presets in the HD500 IMHO are somewhat useless for live or studio use. However Line 6 has given a great workaround with their support of user created patches.

The open architecture would allow sound designers to create amp models that work well and sound good right away. But the main benefit is new amp topologies that may be difficult or impossible with current amp technology could be modeled. I'm not familiar with any modelers except for Line 6, Johnson and Behringer, but all these seem to limit what you can do from a modeling topology standpoint. They all seem to be tied to the past; an open architecture would allow new topologies to be explored. I refuse to believe that the current highly regarded amps are the only way to achieve great tone. Why limit modelers to just those tones? Seems a waste of resources and imagination.

This does present a problem for the modeler manufacturers; it removes the ability for them to create a revenue stream through selling updated amp models. Or at least gives that part of the pie away to others. It will probably take a software manufacturer to design such a product for a standard PC or iPad, and maybe expand that to a floor unit that will support the software. I doubt that will happen, given the possibly similar market with softsynths. You'd think that manufacturers would fall all over themselves to make a "PC and display in a box with a keyboard" to run softsynths, but there are only a few available and they are very expensive.

[Sweetwood]

Too much tweaking. People need to stop geeking out so much and just play. I saw a country guy infuse parts of Eruption into a tune, clean, on a Tele and it melted faces. If you can hear the difference between a silver mica cap and whatever in an amp at stage volume, well, you're probably the .0001% that can talk to squirrels. There are already 30 different speaker and cab models, different mic positions etc. It's way too much already. In this case, you would have one type of resistor that sounds the best and that would be what you use 99.9% of the time...so why even bother.

[lspaulsp]

It's funny I downloaded the 543 amps on the 11R users group. Maybe 20 are useable. The rest sound pretty much the same or Mega Death bad. When someone says look how great this sounds and you get a Triple Rec on 12 you've lost me. This might turn out the same way. It it has definite possibilities.

A Fender Champ with a Marshall power amp might be cool.

[Scott Peterson]

Most of the criticism I've seen thrown at professional level gear is the amount of tweaking aka 'learning curve' it takes to dial it in. I think of all the gear out there that the Axe-FX is closest to embracing the 'let them adjust anything/everything' ethos... and a lot of folks don't like that you can even do so... even if you don't need to adjust any of them in the first place.

Just IMHO. YMMV.

[MusicalCircuitry]

How about this: Instead of modeling specific amp designs, model specific capacitors and resistors, transformers, the whole gamut of parts to make an amp. Then create a interface where a sound designer could build an amp with these virtual components. For those of you familiar with the electronic software Spice, perhaps something along those lines.

There is plenty of software available for electronic circuit simulation - you already mentioned SPICE, but there is also Qucs, gEDA, VHDL, etc. I've used a lot of this software myself, but I can't imagine the average guitarist would want to get into it. It's somewhat strange that there isn't much in the way of user friendly options for do it yourself guitar component modeling, because there is plenty of it for the synth world. Synth guys have some awesome software for this type of thing - Reaktor, MAX/MSP, SynthEdit, SynthMaker, etc.

Great idea but, FWIW, it's not new... I've seen it kicked around many times in different contexts.

Revalver is fairly close to a build your own amp model kit, actually. You cannot skin tweaked models but you can save them and some users have created schematic based modified default models for other users. The amp tweaking capability of that tool is ridiculous. It also has individual preamp tubes and separate power tube amps broken out based on topology the whole nine yards.

Thanks for the heads up on Revalver Elric, I didn't know it had those capabilities. I don't typically enjoy playing through modelling software, but I might have to check out Revalver for the Tweak Module GUI.

[teemuk]

How about this: Instead of modeling specific amp designs, model specific capacitors and resistors, transformers, the whole gamut of parts to make an amp. Then create a interface where a sound designer could build an amp with these virtual components. For those of you familiar with the electronic software Spice, perhaps something along those lines.

People have experimented with that already. However, if you are indeed familiar with SPICE and modified nodal analysis you know it's still computationally too heavy to run complex circuit simulations with latencies allowing almost "real-time" performance.

So far most approaches have fallen into being a) too simplified to be any realistic or b) too slow. Great results have been gotten with simulations of plain Tube Screamer -type stompboxes built out of virtual parts that have simplified the overall circuit quite a lot. Completely MNA -style simulation of even such a simple circuit would have still been far too slow.

There are plenty of whitepapers written about experiments with that kind of stuff. Prominently by authors such as David Yeh, Jyri Pakarinen and Matti Karjalainen. Most can be found with Google search in few minutes.

Anyway, I'm pretty sure we will see more of this in the future when processing power increases and becomes more and more cheaper.

Second problem is simply that a ludicrously small percentage of guitar players are actual circuit designers, or even familiar with electronics and stuff. Someone with background in electronics, specially with such relating to pedals and amps might know how to design a virtual amp circuit - for the 99,999% of guitar players that would be something way over their heads. Building an interface to such modeling system, one that can actually meet the generic markets, would be a quite demanding task, if not even impossible. Not all of guitar players are amp designers and people already complain about too difficult interfaces of modeling amps. Imagine what it becomes when they have to constuct amps component by component, SPICE style. Similarly to SPICE and even reality, some of such circuits would not even work - you actually should know how to design them properly, and on top of that, you should also get them to sound good. So far there are only a handful of people in the whole world who can actually fill that last criteria.

[aleclee]

Anyway, I'm pretty sure we will see more of this in the future when processing power increases and becomes more and more cheaper.I wouldn't be so sure. For the past 5-6 years, mainsteram computer "speed" has increased by increasing how many tasks it can do at one time via multiple CPU cores while actual clock speeds haven't really gotten past the 3GHz mark set several years ago. While there's some benefit to parallelization (e.g. Axe 2) detailed modeling as described in the OP is largely a serial operation that wouldn't benefit as much from the current style of processing improvements.

On the DSP side, the current state of the art seems to be all the market demands, if the TigerSharc's lack of updates is any indication.

[kirs]

I've thought about this sort of thing for a long time. It's not possible with today's processors but in 10 years devices the size of a stompbox (and smaller) will be able to pull it off handily. For reference, rendering a 10ms waveform sample through a simple stompbox circuit simulation takes about 1 second in LTSpice on a quad-core box (to be fair, I don't know if LTSpice takes advantage of multiple cores). To do what the OP wants we would have to hit realtime with plenty of cycles to spare for latency.

I've also thought about the analog equivalent, where you have a circuit full of surface mount components: caps, resistors, variable resistors, diodes, op amps, etc. along a bus and through software you map the bus to various circuits' layouts. With the touch of a button you can go from Tube Screamer to Big Muff, over an all analog signal path. Sort of how the Ernie Ball Game Change routes pickup configurations through software.

[teemuk]

I've also thought about the analog equivalent, where you have a circuit full of surface mount components: caps, resistors, variable resistors, diodes, op amps, etc. along a bus and through software you map the bus to various circuits' layouts.

This type of "analog modeling" was introduced already in the 1970's and companies like Digitech exploit it a lot. That is how eg. their JM series of amps and most of the GSP series units work. If I remember right, only the time-based effects in those were implemented using DSP.

The biggest drawback of that type of approach is that it needs an extensive amount of components and a huge board space: Not only does it require the basic components for the actual analog circuitry, it also needs tons of switch / multiplexer ICs, voltage or digitally -controlled resistors, all kinds of interface circuitry for buttons, rotary encoders, switches, jacks, plus all kinds of indicators and displays, and a computer-driven logic + software to operate all that. Not to mention, all this often even includes a DSP portion just for all time-based effects. The more you wish to add features, the more the thing just grows and grows.

A DSP based solution can emulate all of that with practically a single set of hardware, can be updated and entirely reconfigured by, say, downloading a new firmware from Internet, and so on, so it's a lot more tempting approach.

[banjoze]

Revalver was down this path but being bought by peavey probably hasn't helped it in terms of focusing on a variety of amps.....

[MKB]

Maybe I'm not being clear about one point; obviously the majority of guitarists are incapable of designing an amp. There should be a way where they could simply buy a custom amp model and just tweak the virtual tone and volume knobs, with very little other interaction needed. I haven't seen many current modelers that are so easy to work; certainly the HD isn't that way.

OTOH, those with the design chops and ears to do so now cannot design their own virtual amp, and that is a shame.

If a platform allows tons of flexibility, it is critical that this is presented where the normal guitarists is not required to use more than they have to. Just give them volume and tone knobs, and have the resistor/cap choices buried in menus or accessible via an editor where they don't have to go unless they want to. Or make the in depth stuff part of a SDK that the rank and file can't access unless they pay a lot for the privalege.

[jdolll]

The problem is that the R&D needed to make something like this work and work well would be too intensive for the limited market. While it would be a great idea for the limited market that would be able to understand the product and use it well, it's absolutely a product that would never be able to make a profit on and as much as we love the idea that all people who make guitar products do it because they love the guitar, a business is a business and needs to make money