A Question on “modeling”

[EricKnabe]

So, tube amp emulation and modeling has been around a long time. It seems like, from what I've read, it can be done in both the analog and digital realm. Now, correct me if I'm wrong- I very well could be on this point- but from what I've gathered it seems like analog modeling units like the sans amp, maybe quilter,  and other products like that try to recreate the voltage and current characteristics of a tube stage with op amp stages and SS technology.
So if that's correct, this is essence is the very same thing digital units attempt with DSP: recreating the sonic character of a tube amp on a stage by stage basis. (Before they started component level modeling that is...)

So, which is better? Which method achieves a better emulation of the tube sound? (Don't say tubes do, we all already know that.)

Tube amps have this... smooth character. The way they breakup is very smooth compared to their transistor counterparts. I'll refer to if as "The Tube Thing." That to me is what makes the tube sound superior. The Sansamp actually replicates that well imo, but a lot of SS stuff does not, at all.
Every digital emulation I've ever heard as far back as the original POD does The Tube Thing as well. Usually with the digital stuff it's a matter of frequency response and dynamics that make or break the sim.

But as the Sansamp proves, The Tube Thing CAN be achieved with SS emulations. So my question is which method- analog or digital- is better at it?

[Fancy Lime]

Here we go again 

First, you need to define "better".

Second, you need to specify what tube amp. Not all tube amps do the "tube amp thing" equally well or even at all.

The biggest problem when trying to define "tube amp sound" is usually that most people think any sound they like must therefore be "tube like". But "smoothness" and breakup characteristics are influenced by a lot (and I mean a LOT) of things; the presence or absence of tubes is only one factor and usually not even a very important one. Power supply sagging, interaction between the speakers and the power amp due to low damping factors, frequency response and general circuit topography can all play huge rolls. Or not. It depends on the amp. A Marshall for example clips pretty hard, much harder than many solid state amps. Sunn amps have oversized power supplies to prevent sagging, which makes them pretty stiff and gives them high headroom, things we normally ascribe to solid state amps. But no-one would consider a Marshall or Sunn "un-tube like" would they?

If absolutely realistic emulation of a SPECIFIC amp sound is your goal, DSP is definitely superior. If you want something that sounds good and captures the characteristics that you subjectively find "tubey", it comes down to preference. I like tinkering with analog circuitry, others prefer digital. Both are just different vehicles to get to your goal but you still have to navigate yourself. The art of design remains fundamentally the same.

Cheers,
Andy

[EricKnabe]

I certainly find that DSP emulations seem to sound closer to tune amps than any solid state amp I've tried. But my experience is limited and that's why I asked. A simple answer really is all I was looking for, so even with those variables, here better=closer in sound to a tube amp. And in my experience DSP is a winner, but I had to ask maybe some more seasoned players, perhaps people with knowledge of the inner workings of the tech, hence why I came here. Thanks!

[EricKnabe]

Does anybody else have any input on the DSP vs analog emulators?

[Fancy Lime]

Hi Eric,

yes, I get that you are looking for a simple answer but I am afraid there are not always simple answers to complicated questions. But how complicated the answer is also depends on why you ask. If you want to know which kind of tube amp simulation to BUY, I would certainly say DSP is far superior in this day and age, if your concern is accurate reproduction of all characteristics of a certain well liked tube amp. But if you want to DEVELOP your own "tube amp sim", I can promise you that the answer to your question will change almost every step of the way. It's a fun journey but an endless one. And one that involves that you will question your perception of what "tube like sound" really is. There is a reason why there are so many so dramatically different tube amp simulation circuits (analog and digital) out there, and it's not that everybody minus one person suck at designing circuits. What "that tube thing" is, depends on who you ask. And even if I think I know what you mean by your description of it, what I hear in my head when I imagine that sound may be completely different from what you had in mind. Therefore, I cannot deliver a simple answer.

When you say "better=closer in sound to a tube amp", the question really is: which one? The differences between tube amps are enormous. The difference between a Vox AC30, a Dumble, a Fender Bassman, a 1960's Marshall, a 5150, a Diezel, an Ampeg SVT... is huge. Finding any "tube likeness" these tube amps all share is something I have never seen anyone do and explain convincingly. I would bet a modest sum that most experienced guitar players could not tell which ones are solid state amps and which are tube amps, if you set up a blind listening test with those above plus, for example a Sansamp and a Sunn Beta Lead. The difference between different tube amps is just so much bigger than the systematic difference between tube and (good) solid state amps.

Another advantage of DSP is that you often get a large number of models in one device or plugin and can eve change those in the future. That makes it much easier to experiment. That being said, there is also a reason why there are now even DSP models of Sansamps. Sunn Beta amps and other solid state amps that were originally intended to model tube amps with analog circuitry. These things can sound great. Better than (some? most? all?) tube amps to some people. Not to others. Taste matters more than circuitry when it comes to the magical thing we call "sound". Therefore, I would advise you to listen with your own ears and decide what YOU like. The people with knowledge of the inner workings of the tech (great band name, btw) will not be as helpful as your own ears, I"m afraid. Nevertheless, we're happy you found your way here. You have come to a really fun rabbit hole to fall into.

I sincerely hope that helps a bit. I'm not trying to be snotty but to be as helpful as I can. I have been in your shoes and I assume my answers may be quite frustrating. Sorry about that.

Cheers,
Andy

p.s. This question has been beaten to death for many people, I'm afraid.

[iainpunk]

tube amps can sound very "solid state" to people, if they are designed according to modern circuit design doctrine. A sunn model-T behind a curtain fools most people into thinking that its solid state, because of its bigass power supply and massive headroom, while a fuzz face sounds very tubelike according to some people because its battery powered and sags on the attack. the only thing that i find that tubes do actually different than solid state devices is the relative rate of change, it has a higher rate when going down than going up, this slightly skews the waveshape and gives wave a slight "sawtooth" shape to it

this video in general is really good, but in this instance, skip to 11:07 to see what a tube drive does

[EricKnabe]

Another advantage of DSP is that you often get a large number of models in one device or plugin and can eve change those in the future. That makes it much easier to experiment. That being said, there is also a reason why there are now even DSP models of Sansamps. Sunn Beta amps and other solid state amps that were originally intended to model tube amps with analog circuitry. These things can sound great. Better than (some? most? all?) tube amps to some people. Not to others. Taste matters more than circuitry when it comes to the magical thing we call "sound". Therefore, I would advise you to listen with your own ears and decide what YOU like. The people with knowledge of the inner workings of the tech (great band name, btw) will not be as helpful as your own ears, I"m afraid. Nevertheless, we're happy you found your way here. You have come to a really fun rabbit hole to fall into.

It's funny, contrary to most guitarists I actually feel inspired to play more using sims over "real amps" just because I have everything I need to get a good tone right there in the box. Sometimes when I try to mess with all my stompboxes and amps it gets frustrating trying to get a good sound and end up just going in dry. So I get a bit more creative with effects just due to the ease of use.

So, with regard to your comment on the Sunn amps and other SS amps that were meant to emulate tube circuitry, I am curious as to if I was correct about, are SS emulations generally made by approximating the characteristics of a tube stage, but with transistors or opamps or whatever the case may be?

I'm sorry if I'm not being clear enough, I have a layman's understanding of a lot of these things.

[r080]

Here is a pretty good explanation of one approach to analog modeling - essentially part for part translation of a tube circuit to an FET circuit:
http://www.runoffgroove.com/fetzervalve.html

Another common technique is more of a functional model, typically using some form of back to back diodes in varying configurations to approximate the way tubes clip. If you look at schematics for things like the Sansamp, they are essentially complicated distortion pedals.

[aron]

IMO, in the long run, DSP will be much better than solid state emulations - simply because there is so much work being done in this area. I would not compare a POD to the best DSP. OTOH, I have heard my friend have a stunning tone with a POD and a solid state amp. I couldn't believe it. It even made my dig out my POD to find out it sounds like a POD, not like his! hahaha

[Rob Strand]

IMO, in the long run, DSP will be much better than solid state emulations - simply because there is so much work being done in this area.

I agree.   

With solid-state implementations you are dealing with real parts:  transistor, JFETs diodes etc.   These have pre-defined characteristics you can only go so far to twist and bend things.  If you go through the patents, people have been doing this stuff for a long time with only limited success and things are unlikely to get much better.
 
With DSP you can pretty much do what you want.   Far fewer limitations.  The only limitation is processing speed and that is going up not down.    The proof is the ever-improving DSP stuff out there which has already overtaken the solid-state.

[aron]

All that being said, there's no way for the near future that overdrive boxes are going away. Too many people buying variations on the tube screamer!

[Rob Strand]

All that being said, there's no way for the near future that overdrive boxes are going away. Too many people buying variations on the tube screamer!

True.  It would also be crazy to build a DSP pedal which (only) emulated a TS-9.   A great load of circuit baggage with 100 times the power consumption.

[PRR]

> which method- analog or digital- is better at it?

Check your calendar.

Say it is 1899 again. What is "better" to ride into town? Horse-carriage? Electric runabout? Steamer? Spark engine?

20 years ago (1999), digital modeling needed costly chips and geeky thinking; any fool could wire a tube.

20 years from now (2040) more than ample processing will be about free, why fool with actual devices??

But here we are, between a horse and a steamer. Neither is the wrong answer.

[Mark Hammer]

Digital can model anything. The limiting factor is the quality and fullness of the model and algorithms used.  Among the earliest uses for digital modelling was delay and reverb, because all that had to happen was storage and output of the signal, regardless of the moment-to-moment properties of the signal.

Mimicking the properties of specific analog devices HAS improved as the speed and resolution of DSP devices has increased.  But the challenge still remains that a guitar signal is NOT an oscillator that provides a steady-state waveform to which clever transforms can be applied.  It changes by the millisecond, and often faster, and as a function of the spectral content provided and even how recently other notes/transients have been created.  The DSP units are probably up to the task at present.  The bottleneck is the "wetware" identifying just HOW to transform the signal based on those aspects, in a predictable orderly way.

Analog circuits have those hard-to-identify properties built into them, whereas DSP requires that one know about them and be able to identify, prioritize, and reproduce them mathematically.  We're getting there.....slowly.

[teemuk]

Operation of tube guitar (musical instrument) amps has been analysed quite thoroughly and I'm pretty sure design teams of modern modeling amps are fully aware of the most vital 'tubey' operating characteristics. To my knowledge, even the earliest generation of modeling units featured methods to model dynamic characteristics of analog circuitry.

[ElectricDruid]

I also agree with Aron and Mark.

DSP technology and techniques are constantly improving and getting cheaper. On the other hand, people aren't coming up with fundamentally new analog components, so the options there are much the same as they were thirty years ago. Yes, you can build something analog that models tube behaviour, and if you put the work in and throw a lot of circuit at it, it'll probably be pretty good.

But these days, DSPs are cheap and powerful, and they're only going to get moreso. As people develop ever more detailed models of amps and speakers in software, those DSPs will sound better and better. Unless someone invents a new JFET or diode or something, I don't see the same thing happening with analogue.