How does one model an amp?

I'm not an amp designer, but I'll give an interpretation. Modeling is a way to simulate (or mimic) the operation of something using something else. There are simple, coarse models and there are complex, detailed models. Simple filters, as you are guessing, could be used to make one sort of crude "model". It can get a lot more complex, depending on how much engineering and cost is put into it. One can try to find the entire "transfer function" between an amp input and the acoustic output of the speaker system. This involves measuring gain and phase, and get even get into transient phenomena and heating effects if one wants to go nuts with it (which, for the technically-inclined, means a time-variant transfer function). Note, too - especially with tube amps - that such modeling has to produce a non-linear transfer function to account for overdrive, etc.

Models can be developed in different ways. One can measure the transfer function and design circuits (or software) to approximate it. Or one can also feed these signals into a learning algorithm (artificial intelligence) which generates a set of "patterns" or heuristic rules for the model to follow. The advent of inexpensive and powerful digital signal processing gives designers a lot of ability to do neat things. Those multieffects pedals one can buy to do reverb, wah, compression, distortion, phasing, etc. are a few examples of this. I think amp modeling is probably using some of the same technology. That's my impression, anyway. I think how well it works depends on the level of sophistication put into the model vs. how difficult it is to simulate the behavior of the real amp.

- Mike

 

Hm, I'm no expert, but I think that the electronic components of an amp are simulated by computer algorithms.
That means, you have mathematical formulas that describe the behavior of e.g. a resistor or transistor or tube. A model, so-to-speak.
So, when you feed data into it (digitized audio signals, which are nothing but numbers), you get (approximately) the results that a real circuit would give you.

You need a lot of processing power to get the model as accurate as possible, and low latency, meaning minimal delay between input and output result.

 

JMX makes excellent points - I agree.
- Mike

 

I gladly return the compliment, because I'm really not sure about the exact methods.

I only remember an interview of the engineers of Hughes & Kettner's Zentera amp, who measured the sections of the original amps and and use 2 SHARC 32bit floating-point DSPs in the Zentera amp to simulate them.
The algorithms were coded by Spectral Design who also program FX plugins for Steinberg and other companies.

 

I seem to remember Line 6 doing a long write up in my POD book about the modeling process. I couldn't find it but if I do I will send you a link or something. It was a fairly well written piece and very informative.

 

The Sans Amp BDDI has a booklet of knob settings that simulate basic amp types; SVT, Bassman, Acoustic, Etc.

But a friend of mine has the Johnson Mellinium modeling amp. It sounds great--in the living room. He doesen't use it on stage, probably would be good in the stuido.-noisless operation.
I believe they are made by Digitech because they use the "S-Disc II" Engine, and so does the Digitech BP-8 Bass Pedal.
I have talked to my guitar player a lot about this, and he says that the digital domain for effects are too linear for live stuff. He uses all analog effects.
I have a friend that has the BP-8, and he loves it, and he sounds great, but he only uses one or two patches.

 

I have been using a bass POD pro for a little more than a year now. I have had a chance to A/B it against some of the models. Some you would be amazed, others maybe not.

For the most part, it gets close enough to the SS and hybrid stuff that you probably couldn't tell the difference within the mix of a loud house system or recording.

The tube stuff is a little more difficult to pull off.

It NAILS a GK 400RB. Blindfolded, you'd never know the difference.

It getting amazingly close to the SWR SM400 and the Eden Roadrunner.

Very good job with the Acoustic 360.

As far as the tube stuff goes, well. I guess it at least is close enough for you to know what they were trying to do.

The 400+ model is probably the best tube model. At least it is the closest.

The Fliptop would be second.

The SVT is a distant third.

The Fender sounds like cheese.

I have played through the real versions of all of these amps. I actually have done a real A/B test with a GK, SWR and the Fender. Using the same cabinet, a Carvin 410.

The other models on the POD I can't comment on because I am not familiar with the originals. They include a Vox model, a couple of Marshalls and a few others.

Chas

 

I could see a POD for studio use, where cabinet choice might not be an issue. OTOH I've had people try to convince me that a POD Pro running into a WM12 sounded like an SVT.

 

I don't think you are going to get a SVT sound out of 1 12, even if you use an SVT.

You are absolutely correct that the cabinet choice is a huge factor. It significantly effects your stage sound. I chose the POD because I don't use a cabinet. I play straight into the board with the POD and use the monitors. Even when I use my amp, I make sure I keep the volume low enough that the house is carrying my sound.

Basically, I am getting that studio sound coming straight out of the house.

Chas

 

I did some primitive modelling on my computer..

if you use some RC filters and equalisers, and mix the signals correct, you can get a huge load of possibilities.