Tube and solid state on a professional recording

[teemuk]

I don't know where these "doesn't clean up with volume knob", "doesn't interact with my playing" -ideas come from because I've been hearing plenty of people saying exact opposite and the youtube video I posted here pretty much demonstrates that either of those claims holds no water with a good modelling unit. Then again, I've also heard people saying that you need to install potentiometer brand and style x to your Trainwreck or whatever myth-ridden amp or otherwise it doesn't clean up with your volume knob and interact just right... So I'm willing to believe that a large part of this is in the head of the guy holding the guitar.

A tip: reduce the compression and gain that those modeling amp patches usually have way too much. Tweak the patch sound clean at clean volume settings of your guitar and then it automatically overdrives when you up the volume. Also, learn to control your pick attack. It really isn't rocket science.

[kg]

So Stan, would you agree with my viewpoint here? scopeblog » Less is More: Pay More, Get Less

I hear what you are saying about just capturing single points in the continuum of how an analog amp behaves. If I made a modelling amp, I would want to start with an equation that modelled a vacuum tube over its whole relevant operating range, and a mathematical proof that the equation had no audible discontinuities. Then string it up into a real-time thing that is more or less a Spice simulator.

I don't think that is feasible with consumer-level DSP technology just now, but the stuff is getting more powerful every day.

Oh, and a little module to simulate tube microphonics and the crash you get when you kick a spring reverb.

i think you're on to something.

rather than trying to model the whole amp, as a black box, model components (or at least groups of components).

[km6xz]

Sure, a dynamically varying simulator for the active sound generator would do it. There is the problem, the simple part is amplifying devices but all the others really a problem: wood, glue, body mass/damping, figure pressure, acceleration, string characteristics and all the things that play together in an analog system of creating a spectrum that the modeling amps thus far ignore. They focus on the transfer function of the amp. But the amp being modeled has a difficult to model interaction with all the above mentioned variables.
At a single gain level the identical harmonic spectra of any amp by digital modeling methods and even analog methods. That is the "easy" part. We still can't model the properties of a guitar being played by a human.

[Steve Conner]

That is the "easy" part. We still can't model the properties of a guitar being played by a human.

Who said we had to, though? This thread is about modelling amps.

They focus on the transfer function of the amp. But the amp being modeled has a difficult to model interaction with all the above mentioned variables.

Sure, but it must be possible to quantify the amp's interaction as a transfer function, albeit in the full sense of an arbitrary function that can be non-linear and stateful, which is a far cry from the linear transfer function theory we learnt in college.

If you argue otherwise, you imply that the amp is accepting some other input besides the jack on the front, or maybe somehow talking back to the guitar. Of course the amp does talk back to the guitar, see the controlled feedback vs. latency issue raised earlier.

[km6xz]

The amp is talking back to the guitar in any high gain situation, put the amp in an iso booth and see how different the amp and guitar interact. The problem is we do not know how to model the input signal, otherwise, by approaching it only from the amp side and ignoring their interaction, the modeling of response of the amp to the characteristics of player/guitar combination will neglect to primary goal of creating a system that mimics the playing characteristics and response of known amps and playing styles. But this is all a short term transition, eventually a new generation of players will only know modeling amps and string synth devices and so will not want old playing styles to be reproduced. How many people playing keyboards now care if it emulates the playing response of a piano? In 10 years few people getting into guitar will bother with non-programmable guitar(the modeling amp will not be needed since it is cheaper to build the tone generators into the guitar and use wireless links to the pa system). Old timers will still tinker with "old wood" or vacuum bottles but they will die off soon enough. In years past there was an effort to simulate tape machines when creating the operation and sound of old analog tape. That was just 15 years ago and now the concept is not bothered with and is not even remembered by 90% of recordists today. CD players are still designed to simulate manual manipulation of vinyl records but that is steadily being replaced by computer based DJ program generators which do away with the idea of manipulating anything physical. So this whole issue of modeling is to temporarily appease those who remember the old methods of playing but they will soon be gone.

[Ronsonic]

I don't know where these "doesn't clean up with volume knob", "doesn't interact with my playing" -ideas come from because I've been hearing plenty of people saying exact opposite and the youtube video I posted here pretty much demonstrates that either of those claims holds no water with a good modelling unitl

One of those complaints just came from me, so I can't see how you don't know where they come from. I'm simply relating my experience. You likely do hear people saying the exact opposite and probably all sorts of other things. I wasn't talking about what I hear other people saying, I was talking about my observations.

[retgaz]

I also am speaking from my own experiences. 14 years managing a busy music department in a college teaching up to degree level. We had three different models of the Pod, two line six amps and a Behringer Bass V amp. We also had amplitube and POD farm in every studio. These were used heavily for recording and songwriting sessions.
However when it came to performances the students (by their own choice) would bring in their own amplifiers, which were usualy valve amps and never modeling amps. The less well equiped students would usualy use our Laney VC30 or one of the Marshall valvestates. The line 6 amps would be sat in the wings unused. Students owning modeling equipment tended to be students that did a lot of recording and had no real desire to perform, or were first year students, who would often replace their equipment as their playing style developed and they became limited by the modeling amps.
My distrust of modeling amps is not just that of an old guy who is stuck with the same amp he bought 100yrs ago, it's based on experience, both as a musician and from working with developing musicians.
Modeling amps are great in the studio where the resolution of the recording equipment is barely different to the resolution of the modeling amp, but live, they just don't cut it. Any guitarist, who has developed his playing technique to be able to create feelings through tiny changes in the dynamics of his playing, would be a fool to want to limit that technique. The number of people still wanting to own valve gear, decades after it should have become obsolete, bears weight to my point. Not all of that desire is down to some sort of "snake oil kudos". Musicians want gear that allows them to express themselves. To a skilled musician the amplifier, speaker and room combination becomes part of their instrument. Keyboard players accepted the loss of that interaction because carrying a grand piano, clavinova etc to every gig was not practical (or affordable). For guitarists the difference in size of the amplifiers is negligable. The benefit of having a huge range of sounds is the major selling point for modeling equipment. However to many musicians this is no benefit because they feel it stops them developing their own sound.
I have no doubt that in the next few years modelling amps will become the equipment to have. But the technology isn't quite there yet.

[kg]

well along those lines retgaz i think the idea of the modeling amp replacing the ENTIRE amp (including speaker cabinet in many circumstances) is where the issue lies.

can we model the input stage of a tube guitar amp? sure we can. if we did ONLY that stage "digitally" would there be a big difference in ultimate tone? probably not enough to discern, even with the most golden ears. (of course the story may change if that input stage were heavily overdriven by a pedal.)

so perhaps the "best" modeling amp is a true hybrid--using some preamp stages in the digital realm, and then back to all tube output stage and real honest speaker cabinets. you give up some of the (theoretical) flexibility of a blameless/colorless power amp/speaker which merely reinforces the signal, but you get back some authenticity and liveliness that seems to be missing.

[Steve Conner]

I think this is an interesting point. Uncoloured speakers are always less efficient than guitar ones with their strong colorations. (the colorations are resonances that boost efficiency)

So, to model a guitar speaker with a PA speaker you will need to pump lots of power at the modelled speaker's resonant frequencies. This makes me uneasy, the same way as boosting bass to EQ room modes does. So in that sense I agree with Ken's point. A tube amp and guitar speaker naturally sounds like itself, but the boosted version might sound kind of weak and strained.

If I was designing a modelling amp, I'd use:
A tube first stage: the noisy tube adds some analog dither and compression, helping with the bit depth problem.
The new 24-bit ADC from That.
The same 32-bit floating point DSP hardware we use at work. (20 dollar chip with 1.8 GFLOPS, programmed in C)
Two 200 watt Blameless SS power amps. Heck, two big ones and two little ones so we can have an active crossover.
The Eminence Acoustinator drivers with coaxial tweeters.

of course it would cost a fortune and nobody would buy it!

[retgaz]

I think there are some good points there too. The problem that you would run into is that you would need different output stages and speakers for different models. Say swapping from a JMP 50 playing through 4 greenbacks to a Black face Fender Twin (EL34's to 6L6's etc.). I know these situations are modeled but I presume they are modeled as a complete unit. Perhaps they could model in a more serial manner (if they aren't doing this already), ie. a modeled pre amp followed by a modeled power stage followed by a modeled cab (or corrections for the speaker being used). As Steve mentioned though that could involve some pretty hefty peaks at certain frquencies.
The design Steve described sounds great but would have its own sound instead of being the perfect clone that modeling amps aim to acheive. Perhaps thats the way companies like Line 6 need to start thinking though. So that they give customers slightly reduced cloning but add the benefit of a truly responsive musical "instrument".
I suppose at the moment it's like most things in life. Perfect clones haven't been developed yet. If you want a really close facsimile that will do a lot of stuff get modeling gear. If you want the little bit extra that the real gear will give you but without the extra bells and whistles get the real gear.
I'm sure these discussions have been had in many design meetings by people who are hopefully a lot more intelligent than me. And as I've said before I think I may live long enough to see amps that can trully do the job. I hope its not too soon though, I'd miss working on all those old valve amps.

[kg]

it sort of reminds me of the "jack of all trades, master of none."

me, personally, for my use, i'd MUCH rather have a three awesome channel amp than a three hundred okay channel amp. part of that is because when i do play, it is all original material, and it's for my OWN enjoyment (and the other guys in the room, too, i guess.. ).

however, were i regularly gigging as part of a cover band, then i would seriously rethink that. i'd likely have the swiss army knife amp, hell, let's add a midi pickup too and get true synth sounds out of it. why? because the goal is reproduction for the enjoyment of the AUDIENCE, and having a lot of different tones--even if mediocre--is more important.

[retgaz]

With you 100% on that kg. I'd probably go for a modeling pedal so I could stick it in a cupboard when I could play my own stuff and drag it out for the cover gigs.

[MWJB]

I think we also have to temper this discussion on modelling amps (which has veered somewhat away from "tube vs SS") with the motivation for making them...it's largely to hoodwink kids into the idea that for a hundred bucks they can buy a gadget that makes them sound like they are playing through a tweed bassman through a closed back 2x12 cabinet miced up with a RCA ribbon, or a JTM45 through a 4x10 open back cab mic'd with a Neumann tube mic, etc., etc., whilst the kids in question and the guys who build the modellers have never probably heard these amps first hand & certainly never heard the combinations that the modellers claim to access.

I know that they can be useful tools in the studio, as can flexible tube or solid state amps, but in my experience most guys cannot tell amp wattage/speaker configuration/power tube type/PP vs SE/brand of amp from a recording alone. The reality being that the modellers might be useful tools in their own right, but claim to offer something quite intangible. Most builders who offer reissues & reproductions of vintage amps actually source/audition several amps of that same model in order to try and find a benchmark, before deciding on the characteristics that they deem desirable for their interpretation...I think it's fair to assume that Behringer & Pod do not have a secret facility somewhere, packed to the rafters with hundreds of the world's most desirable & collectible amps...with untold speaker cabinets & hook up options ;-) If they did, their products would cost nearly as much as one the amps in question.

Players can often determine cathode bias vs fixed, SS rectified vs tube, significant variances in preamp tube plate voltage/cathode resistor value. A small PP tube amp with a few options (cathode/fixed bias switch, rectifier options, input stage cathode resistor option etc.) can probably cover 99% of the ground, no less authentically than a modeller (wet effects apart)....it'll still cost several times what the modeller costs however.

[Steve Conner]

I think it's fair to assume that Behringer & Pod do not have a secret facility somewhere, packed to the rafters with hundreds of the world's most desirable & collectible amps...with untold speaker cabinets & hook up options ;-) If they did, their products would cost nearly as much as one the amps in question.

I agree with all the other points, except Line 6 do claim to have a room stuffed with vintage amps, and I've seen shots of it in their publicity material. It's not all that big though.

But yes, modelling amp makers maybe aren't selling a tool for players, they're selling an image or dream on the McDonalds model.

With analog solid-state circuitry it is quite hard to replicate the exact touch sensitivity of tubes. However with what we know now, I'm convinced we could get close enough. I'm sure it would be possible to make two 5E3s, one with the correct innards and the other with a bunch of semiconductors, and have them sound close enough that no player could tell them apart reliably.

Along these lines you could make something that was more of a player's tool than a modelling amp could be, at a competitive price point. It wouldn't be much cheaper than the average boutique amp, because it would still need a good speaker and nice cabinet, but with no tubes needing changed and much improved reliability, the total cost of ownership would be attractive.

Sadly modelling is seen as the way forward, and has taken up all of the new ingenuity and research effort. And, I don't think the boutique segment will ever see beyond tubes.

Overall the market is probably getting the amps it deserves.

[jmaf]

With analog solid-state circuitry it is quite hard to replicate the exact touch sensitivity of tubes.

What is what folks call touch sensitivity? I've heard/read this expression time and again and I don't know what it means.