Amp Model Updates

kennyg75

Member
I have a fun discussion question: Is it worth Fractal revisiting older models and re-modeling since we have new DSP available in the Axe-Fx III? Some of those older models are fantastic but could they be improved? UAD for example revisits some of their older models and comes out with a new one that does audibly sounds better. I don't want to assume that updates are needed but wanted to know what you folks think.
 
Then you get folks complaining about how the old model was perfect, the key to their tone for better or worse etc, and then you need to give in and have the old model, and the new version etc and it gets totally confusing which is the “better” one etc.

I think sometimes it’s best to just leave well enough alone, move forward, keep the existing stuff as is, if people like it great, if they prefer a new model, plenty of those too
 
Then you get folks complaining about how the old model was perfect, the key to their tone for better or worse etc, and then you need to give in and have the old model, and the new version etc and it gets totally confusing which is the “better” one etc.

I think sometimes it’s best to just leave well enough alone, move forward, keep the existing stuff as is, if people like it great, if they prefer a new model, plenty of those too

Very true. Maybe a version 2.0 along with 1.0 could interesting too.
 
IIRC the ARES amp modelling was designed from the ground up for the new hardware. FAS has a history of porting new developments to the firmware very quickly, and free of charge which is amazing.
 
More than happy to be corrected, but my understanding is that no [pre-existing] amps and efx have been re-modeled or re-measured but rather the new ARES architecture - hardware and software optimized to be more-better togethter - means that pre-existing modelling and measurements can be more fully used and optimized and therefore sound a bit better .... (?)

Again, this is my [technically] limited understanding based on my readings since the A3 was launched

Ben
 
From what I remember reading, Fractal models amps by recreating the amp's components and schematic in a software environment. Once that has been programmed, remeasuring the amp would be redundant as the schematic doesn't change. Advancements come by optimizing and improving the component algorithms and how everything works together. Of course, I am also happy to be corrected if this is wrong.
 
From what I remember reading, Fractal models amps by recreating the amp's components and schematic in a software environment. Once that has been programmed, remeasuring the amp would be redundant as the schematic doesn't change. Advancements come by optimizing and improving the component algorithms and how everything works together. Of course, I am also happy to be corrected if this is wrong.

I always assumed (?) they had the real amp opened up and measured it component to component (?) ....... if they did it solely from a schematic how would they know if it sounds like the real thing if they dont have the real thing on hand to compare it to (?) ... and if they have the real thing anyway, surely measuring the real thing the first time would be the best / most ideal way to go (?)

By way of example ...... some 18 months or so ago I was banging-on about the missing EF86 Channel of the Matchelss DC30, and from memory, Cliff (?) chimed in to explain that they no longer had that amp and that the multi-stage "tone" control would necessitate a lot of modelling / measuring .... and this is despite the plethora of DC30 schematics online ...... so I suspect the real thing is needed if not critical

BTW ..... did I say real thing enough times :)

Ben
 
Last edited:
I have a fun discussion question: Is it worth Fractal revisiting older models and re-modeling since we have new DSP available in the Axe-Fx III? Some of those older models are fantastic but could they be improved? UAD for example revisits some of their older models and comes out with a new one that does audibly sounds better. I don't want to assume that updates are needed but wanted to know what you folks think.

What amps do you hear need remodeling?
 
I always assumed (?) they had the real amp opened up and measured it component to component (?) ....... if they did it solely from a schematic how would they know if it sounds like the real thing if they dont have the real thing on hand to compare it to (?) ... and if they have the real thing anyway, surely measuring the real thing the first time would be the best / most ideal way to go (?)

By way of example ...... some 18 months or so ago I was banging-on about the missing EF86 Channel of the Matchelss DC30, and from memory, Cliff (?) chimed in to explain that they no longer had that amp and that the multi-stage "tone" control would necessitate a lot of modelling / measuring .... and this is despite the plethora of DC30 schematics online ...... so I suspect the real thing is needed if not critical

BTW ..... did I say real thing enough times :)

Ben
He didn't say they didn't measure the real amp... He said they recreated the components from the schematics... And that remeasuring would not gain anything. That implies they DID measure it already.
 
From what I remember it's a combination. Most from the amps themselves, some from schematics, and some 'invented' from the mind of Cliff. Sure, there can always be improvements...FAS has proved that they just won't quit about a continued drive to make things better.

There are models that have been kept around from when they have been updated...but popular acclaim brought back the first tries (couple of models of those Friedman's).

My personal favorite of the inventions is the 'Buttery'. Just something about that sucker makes me happy.
 
My understanding it's both as well. The schematics gives the overall structure of what happens when in the signal chain, and then it's measured with a reference amp to see what's really going on because especially on older amps, tolerances was quite low when manufacturing resistors, capacitors,... 50 years ago so sometimes different amps sound quite a bit difference just because of different values. So there's schematics and then reference amps. I also read somewhere that Cliff also uses tone matching as a way of testing. If you tone match the reference amp with the model using the exact same knob values, the tone match profile should be completely flat. If it isn't, there's something that's different in the model compared to the reference amp.
 
None specifically. It's more of a question. As technology gets better and dsp becomes more powerful modeling can be improved. It's more of a question if re-models could provide a more articulate version of the amps.

But do you feel the current amps need more articulation?
 
If Cliff has any sense as a software designer, and I'll bet he has TONS: one would expect nearly every time one of the core algorithms gets updated it automatically back fills any past amp models. All of the models are constantly evolving as long as the FW continue to update the core functionality.... the algorithms that drive the AxeFx Ares/Quantum/MIMIC/etc technology that Fractal develops provide the Framework and are where most development is going to live.

This framework is going to require measured values to be input from each modeled amp. But there is little benefit to remeasuring them unless the algorithm updates dictate it or there is reason to believe one is wrong. A resistance value or amplifier circuit topology is not going to change (actually some component level values change over time (example caps are notorious but this is not a *good* thing you would have to perform maintenance on the physical and then measure it)).

So I am guessing, the only time he would need to 'revisit' a specific amp would be if:

  1. The algorithm update incorporates some new parameter or piece of data that must be measured or remeasured from every existing model. (This has happened before; I remember it was either the MIMIC or G3 update that required every single amp to be redone).
  2. There is an error/issue reported that leads FAS to believe the amp's original value to be mismeasured so it needs to be retaken. (This has happened numerous times).
  3. There is a new method for measuring that is believed to be more accurate for certain things. I'm guessing this is quite rare if they are measuring well understood things like resistance/voltage/current, etc.
  4. Some other thing happens I haven't thought of because I do not work in the music technology space
We have had tons of updates that have broken all kinds of folks presets because the algorithms fundamentally shift all of the amp models, regardless of what FW they were introduced in. Another example: Ares made all the clean/crunch model more articulate, all at once.

The only cases we have had where specific models have gotten huge overhauls are when Cliff discovered or obtained a physical specimen that was substantially different than a prior amp/revision. Ex: Friedman BE V1/V2. V1 was an early model that differed from the newer model (V2).

The idea of 'remodeling' an amp is not even meaningful in most cases, IMHO.
 
Last edited:
Hey guys, I develop and lead software teams in my day job. There are several reasons why accessing old models as is may be difficult/impossible.
1) 8 Years elapsed between AF2 and AF3. I'm almost certain a large part of their core code was re-written. You have to take these opportunities for re-writes because you learn lots of lessons and you make lots of messes along the journey of product development.
2) Old product upgrades do not often get rolled into new products. All the development of AF3 was likely done in parallel with AF2 updates.
3) Once fundamental architectural differences are introduced, which is inevitable with new hardware, new software and new capabilities, models connect to the software system differently then they did before. It's not like a guitar cable, it's like a modular synth patch.
4) The underlying connection between the models and the hardware (distinct from software) was likely re-written to utilize and parallelize the new DSPs. This would require a complete rewrite of the model to utilize the new hardware capabilities.
5) Business reasons. They aren't making any extra money from it, it's not a HUGE in-demand feature ( y not just use the old hardware if you love the old models so much ), and they've made a tool to help port AF2 patches to AF3.
 
Hey guys, I develop and lead software teams in my day job. There are several reasons why accessing old models as is may be difficult/impossible.
1) 8 Years elapsed between AF2 and AF3. I'm almost certain a large part of their core code was re-written. You have to take these opportunities for re-writes because you learn lots of lessons and you make lots of messes along the journey of product development.
2) Old product upgrades do not often get rolled into new products. All the development of AF3 was likely done in parallel with AF2 updates.
3) Once fundamental architectural differences are introduced, which is inevitable with new hardware, new software and new capabilities, models connect to the software system differently then they did before. It's not like a guitar cable, it's like a modular synth patch.
4) The underlying connection between the models and the hardware (distinct from software) was likely re-written to utilize and parallelize the new DSPs. This would require a complete rewrite of the model to utilize the new hardware capabilities.
5) Business reasons. They aren't making any extra money from it, it's not a HUGE in-demand feature ( y not just use the old hardware if you love the old models so much ), and they've made a tool to help port AF2 patches to AF3.
You understand that they are modeling the analog components in the digital realm, right?

The components and their values have not changed... Thus, "remodeling" AN amp makes no sense. They might change the modeling algorithm which applies to all amps but the values used (from the analog amp) are still the same.

I think @zenaxe described it well.
 
You understand that they are modeling the analog components in the digital realm, right?

The components and their values have not changed... Thus, "remodeling" AN amp makes no sense. They might change the modeling algorithm which applies to all amps but the values used (from the analog amp) are still the same.

I think @zenaxe described it well.

You can remodel interactions between components of the amp using the extra DSP. I'm not necessarily suggesting a full remodel.
 
You can remodel interactions between components of the amp using the extra DSP. I'm not necessarily suggesting a full remodel.
That is an ongoing process by Cliff.

You're pretty new here and the Axe Fx III is still a baby, unlike the Axe Fx II, so you lack some history and prior experience.

You'll see things like what you are talking about added in FW all the time. That is not "remodeling" AN amp, it's a change to the core modeling algorithms.
 
That is an ongoing process by Cliff.

You're pretty new here and the Axe Fx III is still a baby, unlike the Axe Fx II, so you lack some history and prior experience.

You'll see things like what you are talking about added in FW all the time. That is not "remodeling" AN amp, it's a change to the core modeling algorithms.

That's a really naive statement with unbelievable arrogance. I've been using modeling equipment for 20 years. I'm well aware of each generation of Axe.
 
Back
Top Bottom