New amp: Need help

i don't particularly care. my point is that your claims were unsubstantiated. i try not to get involved in threads where there's some petty "X is better than Y" thing going on, but when posts take some fundamental rules of physics and turn them on their head...

"gain" and "distortion" are two different things, physically and functionally. if you're saying that an amp can have the same amount of power from a fully distorted preamp and from a clean preamp, you are correct. get rid of the master volume, and everything changes.

whatever the case is, you cannot argue with one empirical truth: the power output (or current gain) in the output stage is very dependent on how much voltage is driving the grids. biasing and voltages is the other part of it. take a 100W multichannel amp, turn it up to a decent level and then turn up the gain on the clean channel. you can FEEL the power increase, not just hear it. this is the reason why multichannel amps have separate volume controls per channel - if you jumped from one to the other without setting that right, you'd be slamming the PI/power section in a big way and would probably go deaf going from clean to full bore distortion.

end of discussion.

 

Isn't it possible to take a real t00b amp, feed signals with a wide variety of input you would get from playing a guitar, and analyze the outputs. [you know by actually making someone play a real guitar into a real valve. Split the input signal going into the amp, and compare that to the output signal coming out from the speakers.

If nothing then at least you would be able to cover the gain, and dynamics part of it.
And behavior that comes close to the actual amp.

 

while your concept is correct, that's the most primitive way of doing it, actually, and the result would be extremely different and massively disappointing.

the actual approach is to analyse signal behaviour at each functional block of the circuit.

we haven't even discussed one of the most important parts of the amp - the output transformer, and modelling that in the digital domain. inductances are notoriously difficult to begin with, then there's the question of interleaving, capacitances etc.

the whole point of amp modelling is to get the tone of THAT marshall, fender, vox, (insert brand name here)., and doing that is a lot more complicated than simply comparing an output to an input and deriving a static algorithm out of that. there's a lot more going on there, and a good reason why all those amp emulators, software and hardware, charge good money for their products. you're essentially trying to digitalize each and every component, piece for piece... and that's anything but simple, particularly when you want a linear system to react in a very specific non-linear way. they're probably still tearing their hair out at fractal audio trying to improve the "bloom" of their trainwreck model...

like i said, i'm not an expert on modelling, i just understand the signal path and system requirements - you would need to join one of the forums to learn more, maybe acmebargig seeing that it's an open source project.

analog vs. digital, dynamic vs. static, complex vs. discrete... within 10-20 years and we should be seeing 100W digital modelling amps that nail the tube sound, have 95% power efficiency and weigh 1kg.

hope this helps.

 

I guess somebody who knows the exact answer to that has the potential to become a millionaire...

 

FWIW, i think emulators are getting closer with each new product on the market and drop in the processing powerrice ratio.

 

I guess thats what they do when they sample overdriven guitar tones for keyboards. You could easily tell an actual guitar tone from a keyboard tone because it sounds so fake (inspite of the added chorus and delay).

 

What about PC based programs?
Or they are even more crappier?

 

I guess guitar as an instrument is way more complex than a keyboard. The only variables in a keyboard note is the velocity with which it is struck.

But in a guitar, a lot more things affect the overdriven tone: how hard you hit the string, how far from the bridge you pick the string, the pick angle and also picking at different nodes on a string give different overtones. For that matter, if you play the same E note on 4 different strings you get very different sounds. So I don't think you totally can simulate the sound by using some kind of trigger to set off a pre recorded tone.

PC based programs have a whole new issue: the lag between input and output due to A/D and D/A conversion. So I think you are better off waiting for a modelling amp or something.

 

I already have Tech21 stuff, but it was more in terms of general musing.
Why digital has not caught up ...