aristotle
New Member
- Joined
- Apr 26, 2012
- Messages
- 771
It's all an interesting discussion in any case. This business of how waveforms get distorted in a particular amp is top of mind for people who model such things. The designer of an amp that attempts to duplicate a tube guitar amp has at least a couple of choices on how he / she might want to do it.
One type of modeling amp tries to take a look at all aspects of the original amps design, warts and all...then tries to duplicate that with solid state circuitry (not digital circuitry...analog transistor circuitry). This is how it was originally done, and it is exceedingly hard to get right. In principle, you can take a gain stage for example, figure out exactly what is going on, and there is no reason on earth that you can't duplicate how that gain stage is working by adding the sorts of non linearity that the original design had. The problem is that there is so much feedback amongst components, in even the simplest amp designs that it's near impossible to get right. Just too many variables to account for that aren't represented in component data sheets. So, you try to measure away and hope that you are accounting for as much as you can. The results are usually less than pleasing.
For the other type of modeling amp, the designer takes the attitude that he doesn't care if the amp is composed of transformers, tubes, leaky capacitors, inductors and resistors, or if the amp has little green Martians inside making the amp do what it does. It treats the amp (or the amp-speaker combination) as a black box. The modeler measures the input / output relationship of the black box under a variety of input conditions to the point that an accurate representation of the original box can be made. This requires all sorts of digital signal processing, so it is by its nature a digital (sampled data) thing. So for those of you who can hear the digital nature of a CD for example, you'll hear the same thing with these. But these things do work surprisingly well, and they have the advantage that a single amp can model any amp that you want. They do (in my opinion) a very decent job of including all of the power supply / transformer types of things that are hard to accomplish with an analog model, to the point that I find it hard to tell in a recording which is the original and which is the modeled amp. Live, I do think that I can feel a difference. I attribute it to the fact that the modeled amp isn't developed from my particular crappy playing, and no model has infinite fidelity, so I can just feel the difference. The amp doesn't respond to me like the original. Or maybe it does, and I just think it doesn't. It's pretty close in any case.
One type of modeling amp tries to take a look at all aspects of the original amps design, warts and all...then tries to duplicate that with solid state circuitry (not digital circuitry...analog transistor circuitry). This is how it was originally done, and it is exceedingly hard to get right. In principle, you can take a gain stage for example, figure out exactly what is going on, and there is no reason on earth that you can't duplicate how that gain stage is working by adding the sorts of non linearity that the original design had. The problem is that there is so much feedback amongst components, in even the simplest amp designs that it's near impossible to get right. Just too many variables to account for that aren't represented in component data sheets. So, you try to measure away and hope that you are accounting for as much as you can. The results are usually less than pleasing.
For the other type of modeling amp, the designer takes the attitude that he doesn't care if the amp is composed of transformers, tubes, leaky capacitors, inductors and resistors, or if the amp has little green Martians inside making the amp do what it does. It treats the amp (or the amp-speaker combination) as a black box. The modeler measures the input / output relationship of the black box under a variety of input conditions to the point that an accurate representation of the original box can be made. This requires all sorts of digital signal processing, so it is by its nature a digital (sampled data) thing. So for those of you who can hear the digital nature of a CD for example, you'll hear the same thing with these. But these things do work surprisingly well, and they have the advantage that a single amp can model any amp that you want. They do (in my opinion) a very decent job of including all of the power supply / transformer types of things that are hard to accomplish with an analog model, to the point that I find it hard to tell in a recording which is the original and which is the modeled amp. Live, I do think that I can feel a difference. I attribute it to the fact that the modeled amp isn't developed from my particular crappy playing, and no model has infinite fidelity, so I can just feel the difference. The amp doesn't respond to me like the original. Or maybe it does, and I just think it doesn't. It's pretty close in any case.