Mats Eriksson wrote: ↑
Fri Sep 13, 2019 8:58 am
This with the power amp behaviour is an elusive thing to model. The main culrpit or problem with modelling is they're trying to make something totally chaotic, random, unlinear.... linear. They try to calculate and "read-ahead" chaos and take a guess what is supposed to happen.
I firmly believe in the close-enough-for-rock'n'roll standard. Seems like it would be easier to just generate similar randomness for some things. I'm not sure yet if this example applies to power-amp simulation at all, but Destroyfx had this really cool and simple effect called dfx Skidder
(looks to be still available — open source no less!) that was basically just an advanced tremolo that let you dial in as much randomness on different parameters as you wanted, and it was amazing to me how much life just a touch of "random" can add. I can just imagine that having several instances running through parallel gain stages would be awesome, even if they were set at almost subliminal effect levels. It would still sound natural, but more fluid and 3d, not like an "effect" at all... maybe. I intend to set up a computer to fart around with this stuff again. Maybe even resurrect the old P4 Precision if I can find my ancient but quite low latency IEEE 1394 audio interface, just so I can use energyXT again.
Mats Eriksson wrote:The "flyback" current and interplay occuring between speakers and power amp are yet to be simulated properly. When you tax a real amp to beyond safety levels. I am very well versed in Neil Youngs live sound on "Rust" especially on those two tunes. It is too many stochastic things going on at the same time... I know very well what you mean. But the thing is, is it THAT sought after except of you and me...like "I know of only God and one more, that would like that in a sim, but I've forgotten his name..." sort of thing.
JCJR wrote: ↑
Fri Sep 13, 2019 2:04 pm
Depending on the nature of the speaker, you might even get some "power sag" effects due to speaker coil resistance change because of voice coil heating.
Different speakers would have different heating and cooling time constants depending on design.
Driven by a clean solid state amp (which is not typically real picky about output load impedance except that power = rms volts ^ 2 / impedance) voice coil heating might be mostly clean power compression.
But a tube/transformer coupling might do funnier interactive things as temp modulates speaker impedance?
Again, stuff like flyback current
and motional impedance
and other behaviors are mostly ignored now, so how bad could it be to "subtly" fake it somehow? Motional impedance is usually just vaguely described as something like a random enforcement or cancelling of certain frequencies, almost like nearly nothing is known about it (though I did once find reference to a white paper from the '70s about the subject) — hell, creatively utilized Skidder-type randomizer thing to the rescue! And as minor as something like that might seem, Ted Weber thought that representing motional impedance was the key to realistic speaker simulation, so it seems like it should be given at least some attempt.
Less subtle would be the wave shaping and "envelope generator" type behaviors, at least when taken beyond a certain point. Here again, I think parallel gain stages with varying levels of the simulated behavior would sound more "natural", less overtly like an effect, even if the simulation is on the crude side.
Neil Young is just the best example I can think of to hear power-amp overdrive because it is extreme. I do think a lot if people would enjoy that type of experience if it could be contained in a sim, even if incompletely, but the same type of behaviors dialed back would benefit all styles, IMO, experienced as improved "feel".
I am just hoping that there are ways to improve amp sim feel right now using regular old audio processing. Worth a try, anyway. Of course, it would be much, much better if someone who actually knew what they were doing would give priority to feel and dynamics in amp simulations.