Are the Amps in Axe-Fx modeled after "Real Amps" or "Schematics"?

D

dumbeat

Inspired
I am curious to know, and the new view option raised my curiosity.

Are the Amps been Modeled after "Real" Amps including idiosyncrasies and slightly "off" specs of components, as they behave in real life, or are the models actually based on the "Schematics", or pure Design, which means all the components are at ideal spec (which they can not be in real life).

As we all experienced, an amp goes through a break in period and the more or less out of spec components are eventually what give the amp its tone we love.

So i am wondering if the models are based on a good sounding version of a given amp or on the "Scientific" Plan... The "Schematic".

The reason im asking is because the fact that previously we had the same controls for all amps makes me think that perhaps its a one Template fits all with different binary tolerances places into the slots based on perfect world components...

Would love to hear from Cliff about this.
 
Last edited:
as far as i know, almost all were modeled from a real, physical amp that was examined.
 
chris said:
as far as i know, almost all were modeled from a real, physical amp that was examined.
Click to expand...

Thats awesome if so. So the tolerances and interaction of components are unique to a specific amp example vs. How the semantic wanted it to be? Is that correct?
 
From what I've read on here, it can be done from a schematic, and it's easier with a schematic, but the schematics aren't usually completely accurate (sometimes deliberately), and the schematics don't provide details of all the specific values needed, it might specify a component but maybe not the type or value.

They can trace the circuit of an amp without the schematic but that takes way longer and is a huge pain. So it's best to have a schematic of the amp, and the amp itself, to cross reference details where needed. But they pretty much have an amp of every model that's been modeled, or they did at one point.
 
shatteredsquare said:
From what I've read on here, it can be done from a schematic, and it's easier with a schematic, but the schematics aren't usually completely accurate (sometimes deliberately), and the schematics don't provide details of all the specific values needed, it might specify a component but maybe not the type or value.

They can trace the circuit of an amp without the schematic but that takes way longer and is a huge pain. So it's best to have a schematic of the amp, and the amp itself, to cross reference details where needed. But they pretty much have an amp of every model that's been modeled, or they did at one point.
Click to expand...

Im not talking about whether its true to the Schematic, lets assume that is possible. Im talking about how a real amp behaves. Its always somewhat out of spec, which really gives it its sound. That's what is called break in. So its a major difference whether its based on unique interaction between out of spec components or a copy of a schematic, Which would render the model "Better than new"... from a technical standpoint...
 
dumbeat said:
Im not talking about whether its true to the Schematic, lets assume that is possible. Im talking about how a real amp behaves. Its always somewhat out of spec, which really gives it its sound. That's what is called break in. So its a major difference whether its based on unique interaction between out of spec components or a copy of a schematic.
Click to expand...

all the components are modeled individualy, different types of the same component too, diodes/tubes/capacitors etc etc. Then those components of the same type and value can be reused in a different virtual circuit, a little virtual component module, plug and play.

The way I understand it is that the values and types of the components are combined in the virtual realm to build the circuit with the right component types and values and right order, and then that virtual circuit is used in the MIMIC process, to do some kind of fine tuning / tone matching something or other, so that the slight variations of the modeled components in the virtual circuit compared to the real amp are accounted for on the final output. Basically witchcraft.

There are simulations that account for component interaction, stuff a straight schematic virtual circuit wouldn't do, see the "cathode follower harmonics" parameter in the manual. Stuff like that.

here's the MIMIC description: https://www.fractalaudio.com/downlo...actal-Audio-Systems-MIMIC-(tm)-Technology.pdf

the majority of sound differences between one amp of one type and another amp of the same type (two JCM800s on the shelf next to each other) will be probably from the tubes more than anything else. Tubes are old technology, it's a light bulb. Slight variations in the tube's age/wear/tolerance from one amp to another make a big difference, probably way more so than the infinitesimal differences between the diodes/caps/resistors on the board of one amp vs the same diodes/caps/resistors on the board of the other amp. That's why power tube mismatch has such a drastic immediate effect, you can just nudge it slightly and it goes from sounding like a 'modeled reference amp' to 'the amp you just bought on craigslist'. It shifts into something immediately real, moving off of zero.

what did i miss?
 
Last edited:
years ago cliff "G3'd" amps, and from what i understand/remember, had to open up all the amps and physically measure some things to get them to the new generation of modeling. it took some time, and we got a bunch of updated amps in groups as he progressed.

real dedication to this project. it's inspiring.
 
shatteredsquare said:
the majority of sound differences between one amp of one type and another amp of the same type (two JCM800s on the shelf next to each other) will be probably from the tubes more than anything else
Click to expand...
Tubes and pots I'd say, those are the components that tipically have the worst tolerances.
Even though old caps and carbon resistors values were not the most precise either
 
DLC86 said:
Tubes and pots I'd say, those are the components that tipically have the worst tolerances.
Even though old caps and carbon resistors values were not the most precise either
Click to expand...

oh yeah i forgot about pot taper, those are legit probably the single largest variable, more than tubes.

FractalAudio said:
A typical pot used in a guitar amp has a tolerance of ±20%. On top of that there is the matching of the taper. The taper can also be off as much as 20% at the midpoint.

So if we take, say, a 100K linear taper pot that's perfect it will have a resistance of 50K at the midpoint. We assume perfect pots in the models. However an actual amp may have a pot that's low by 20% so that would be 80K. If the taper is perfect then it's only 40K at the midpoint (20% error). If the taper is off then it might only be 32K at the midpoint for a total error of 36% (!!!). To get the same response on the model you would need to set that control to 3.6 instead of 5.0. That's an absolute worst case and I've never seen that but I routinely see pots that are 20% off at the midpoint.
Click to expand...
 
So if i gather what i hear from you guys, they are not based on a specific specimen but rather on the design, when the components are "perfect" world spec, and then compared in tone to real amps to make sure the sound is in the ballpark?
In other words, since no two amps sound the same, the Axe model also does not sound the same as a specific amp. Its the Axe FX model.

Would that be correct?
 
dumbeat said:
So if i gather what i hear from you guys, they are not based on a specific specimen but rather on the design, when the components are "perfect" world spec, and then compared in tone to real amps to make sure the sound is in the ballpark?
In other words, since no two amps sound the same, the Axe model also does not sound the same as a specific amp. Its the Axe FX model.

Would that be correct?
Click to expand...
No. They are matched to the specific amps that Fractal possesses.
 
dumbeat said:
So if i gather what i hear from you guys, they are not based on a specific specimen but rather on the design, when the components are "perfect" world spec, and then compared in tone to real amps to make sure the sound is in the ballpark?
In other words, since no two amps sound the same, the Axe model also does not sound the same as a specific amp. Its the Axe FX model.

Would that be correct?
Click to expand...

The MIMIC process uses one specific amp specimen to fine tune the virtual amp, so when it's all said and done, that one amp specimen is the target FAS shoots for, to be exactly like it in every way, as close as they can.

Some components are 'ideal', theoretically/mathematically perfect, and some are modeled with and off of real world variations (the different tube types, Xformer parameters)

so yes, another Mark II C+ you find on eBay might not sound like the Axe-Fx model, but depending on what you know about it (the amp you found), you can tweak the advanced parameters of the Axe-Fx model to get it to imitate whatever the eBay amp is doing that's off center from where the Axe-Fx model is starting. Unless you're talking about a full on heavily modified circuit, as if you got one from voodoo mods, then it would basically a different amp, which you could still get close to though.
 
i've been seeing "amp packs" lately, and i think these are mostly fine-tuned settings in the amp block to make them more like a specific real amp of that model.

there are so many aspects to guitar amps and audio that many people may not realize or be familiar with. for $2500 you aren't just buying an axe-fx. you're getting the HISTORY of amps, cabs, mic preamps, and so much more gear, and also the connectivity and capabilities of making them all work.

if you think you could buy the physical equivalents of these things and just plug them into each other directly, you may be wrong. you'd need additional preamps, buffers, cable, routing, and so much more.

there is a ton to take for granted with this, and a ton to be blissfully ignorant about (not a negative thing). that's why sometimes people explain certain things or say things are the way they are because of these physical requirements with the real gear, that the Axe-Fx models. of course much of the physicality is solved already, but some things just are what they are.
 
check out the Amp models list on the wiki:

http://wiki.fractalaudio.com/axefx2/index.php?title=Amplifier_models_list

in particular Cliff's notes. For example on the AFS he writes:

"Cliff's comments:
  1. " I have an actual Slash signature model and spent a lot of time measuring, listening and comparing." source
  2. "The Brit 800 #34 does a better Slash than the AFS100 IMO. The AFS100 models are based on the actual amp but I think my version of #34 sounds more like AFD. I based the model on information I've collected over the years about the mods that were done." source
  3. "The Brit 800 #34 has a ton of treble boost and will squeal with certain guitars and/or certain IRs and/or certain types of amplification. Nature of the beast. No other amp has anywhere near the amount of treble boost. The AFD100 amp that Marshall produced years later (that was supposed to be a #34 copy) has nowhere near the amount of treble boost and has a huge snubber cap on the PI (which rolls off the treble). My gut tells me that they did this because they feared there would be too many reports of pickup squeal (or the amps would be too unstable at high gain)."

    so yes, at least some models are based on actual, individual amps, in some cases plus some tweaking to sound like what they are supposed to be a copy of.
 
phil92 said:
check out the Amp models list on the wiki:

http://wiki.fractalaudio.com/axefx2/index.php?title=Amplifier_models_list

in particular Cliff's notes. For example on the AFS he writes:

"Cliff's comments:
  1. " I have an actual Slash signature model and spent a lot of time measuring, listening and comparing." source
  2. "The Brit 800 #34 does a better Slash than the AFS100 IMO. The AFS100 models are based on the actual amp but I think my version of #34 sounds more like AFD. I based the model on information I've collected over the years about the mods that were done." source
  3. "The Brit 800 #34 has a ton of treble boost and will squeal with certain guitars and/or certain IRs and/or certain types of amplification. Nature of the beast. No other amp has anywhere near the amount of treble boost. The AFD100 amp that Marshall produced years later (that was supposed to be a #34 copy) has nowhere near the amount of treble boost and has a huge snubber cap on the PI (which rolls off the treble). My gut tells me that they did this because they feared there would be too many reports of pickup squeal (or the amps would be too unstable at high gain)."

    so yes, at least some models are based on actual, individual amps, in some cases plus some tweaking to sound like what they are supposed to be a copy of.
Click to expand...


Ill dive into it at the earliest.
 
shatteredsquare said:
the majority of sound differences between one amp of one type and another amp of the same type (two JCM800s on the shelf next to each other) will be probably from the tubes more than anything else. Tubes are old technology, it's a light bulb. Slight variations in the tube's age/wear/tolerance from one amp to another make a big difference, probably way more so than the infinitesimal differences between the diodes/caps/resistors on the board of one amp vs the same diodes/caps/resistors on the board of the other amp.
Click to expand...


I’ve wondered sometimes if being able to model the ageing of a tube is worthwhile...I don’t know...
 
You must log in or register to reply here.
Back
Top Bottom