Vacuum Tube Amps vs. Solid State?

[Noein]

It seems that amps using vacuum tubes are becoming more and more popular with musicians these days. They're a bit expensive, at least the models I've seen, and the claim is that these vacuum tube amps produce a better sound or a richer sound. I was wondering if there was a physics-based justification for this claim or if it's just some psychological thing.

 

[Sojourner01]

I believe it's to do with the shape of the signal they produce. Semiconductor signal processing (I believe) introduces spikes and jagged edges into the signal. Valves are apparently preferable because of the configuration of the noise they produce. It's hard to describe, but valve amps sound slightly 'fuzzy'.

 

[turbo]

I have played electric guitar for ~40 years now, and I have used tube amps exclusively for all that time. I also tune up, repair, and restore tube amps. Solid-state amps have an immediacy that can sound harsh and edgy. The best tube amps for my kind of music (blues and rock) have tube rectifiers as well as tube power and preamp stages. I keep the power tubes biased a little on the hot side, for more breakup, and I hit the input pretty hard, so the rectifier has a hard time keeping the B+ steady, and the B+ rail sags. This allows the note to "bloom", and the interaction between the guitar and the amp during the attack, sustain, and decay of the note can be exploited, if you know what you're doing.

 

[rcgldr]

Tube amps supposedly create even harmonics, while solid state amps create odd harmonics. Other than marketing hype, it's all moot, as a few years back, some clever guy created a solid state circuit that duplicates all of the characteristics of the tubes used in amps. There's no reason that an amp composed of these tube emulating solid state circuits could not be made, other than marketing (or maybe such an amp is made but I'm not aware of it).

Solid state circuitry is more accurate than tubes, and can be made to eliminate harmonics, then sound modules are used to put selected harmonics and other effects back in.

One issue with older stye tube amps, is there is little negative feedback to control high frequency harmonics, which leads to hearing damage. The low frequency stuff (popular at disco clubs with high end equipment) doesn't cause damage unless the decibles are very high. Your chest cavity will resonate before your ears get damaged with low frequencies.

 

[chroot]

The transfer function for a good-quality solid-state power amplifier is almost perfectly linear through its entire output range, until it hits the power-supply rails. When the output hits the rails, the signal is "clipped."

It's as is someone came by with a pair of scissors and just cut the tops and bottoms off your nice, pretty sine waves. In other words, the clipping introduces discontinuities in the derivatives of the signal, which introduces unwanted harmonics. For some reason, these harmonics sound ugly -- or even unnatural -- to human ears. None of the classical instruments made of strings and pipes are capable of making such jagged waveforms.

On the other hand, tube amplifiers have quite non-linear transfer functions, particularly as their otuputs get closer and closer to the rails. This is a very unwanted phenomenon in communications, sound recording, and many other situations, which is most modern amplifiers are solid-state. When you pick up a guitar, though, you're not aiming for linearity.

Since tubes more gently "round" the tops of large signals, you don't get the strong, unwanted individual harmonics. Instead, the gently rounded sinusoid spreads that unwanted harmonic energy over a large bandwidth. There's second-, third-, fourth-, and even fiftieth-harmonic energy in that sound -- it becomes a rich tapestry of tonal qualities all mixed together. Then the amplifier manufacturers provide a couple dozen knobs to turn, each gently shaping that harmonic envelope, and give musicians reason to spend rainy weekends "working on their tone."

So why does the tube amp sound better? Psychoacoustically, humans are not capable of detecting a small sound with a frequency close to that of a large sound. Futhermore, the ear's response is logarithmic -- even if a single harmonic contains half as much energy as the fundamental tone, is it not perceived by the ear as being anywhere close to "half as loud." As long as none of the harmonics compete fiercely with the fundamental, you'll probably enjoy the sound.

As a result, even a very heavily clipped (over-driven) tube amp continues to produce a pleasant sound. Heavily-clipped sounds with lots of harmonic energy all over the map may be considered "dirty," or "gritty," or "fuzzy," and are even the subject of intense envy if you're trying to sound like Jimi Hendrix.

If you drop your Marshall stack out a window and have to settle for a Crate, you'll be in for a different experience. An overdriven solid-state amp just assaults you with a very strong, very unflattering second or third harmonic nearly as loud as your fundmantal tone. Instead of creating a rich, full sound reminiscent of the warmth of sunlight's continuous spectrum, you're staring at the monochromatic ugliness of a naked fluorescent tube.

- Warren

 

[Sojourner01]

Nice one, Warren. Better answer than I could ever have come up with.

 

[3trQN]

It seems that amps using vacuum tubes are becoming more and more popular with musicians these days.

Thats not a new trend by the way, its allways been that way.

 

[Dense]

Great answer, Chroot!

 

[Noein]

Thanks for the replies all.

The transfer function for a good-quality solid-state power amplifier is almost perfectly linear through its entire output range, until it hits the power-supply rails. When the output hits the rails, the signal is "clipped."

What does the term 'rails' mean?

 

[chroot]

What does the term 'rails' mean?

'Rails' is an old term for the power supplies of a circuit. If your circuit is supplied with, say, 20V and ground, the "rails" are 20V and ground. No linear amplifier can produce an output voltage greater than its supplies.

The term 'rails' is used largely because early circuit were made with a bar of metal for the positive supply and another for the negative supply, and devices were connected between the rails.

- Warren

 

[rbj]

i sort of work in this business: DSP algorithms for audio/music effects. i haven't specifically worked in tube emulation but have, at Audio Engineering Society conventions (another one is coming up this week in San Francisco) had multiple discussions with many other audio signal processing persons about this very issue. DSP emulation of these classic tube amplifiers is a "holy grail" issue sort of like perfect, artifact-free time-scaling of audio signals (that is speeding up or slowing down a sound signal without changing the pitch). the issue is ultimately one of mathematics and psychoacoustics and the physics of the vacuum tubes and the other components (in the amps that we like) just happen to process the signal with the mathematics that we like. it doesn't mean that the physics of the vacuum tube is the only way to do such mathematics to the electric guitar signal.

if we put a nice vintage Marshall (or a Crate or a Fender Twin Reverb or a Mesa Boogie or whatever is your favorite) amp (cranked up to arc-weld or some other good setting) and bottom (loudspeaker) in a good anechoic room that's not too small, put a nice B&K instrumentation mic (i don't remember the model numbers anymore) in front of it (or maybe it should be a Shure SM-57 mic which is the sort of standard mic they use on stage to mic guitar amps), pull the amp input terminals and the mic output terminals out of the room, close the door, and put this whole thing into the metaphoric "black box", what you would have left is a "signal processor" that takes the input signal (whatever comes out of the guitar), performs some mostly repeatable and causal mathematical operations on that signal, and outputs another signal (from the mic) that is a deterministic mathematical function of the input signal (along with two possible additional input signals, the 60 Hz hum from the less than perfectly filtered power supply and the tube shot noise, which even though those toocan be emulated, but I'm not sure we want to include those effects in the output).

the problem is, and the sole reason that tube emulation is still the unfound "holy grail" of some audio DSP heads (like myself), is that we do not know the entire nature of that mathematical relationship even though we know it must exist. now, electrical circuits (and causal DSP algorithms) have, among other things, two basic and orthogonal attributes: linearity vs. nonlinearity and memoryless vs. reactive (or "non-memoryless"). the easiest combination to deal with is linear and memoryless (e.g. a broadband amplifier or a resistive voltage divider). linear and non-memoryless (filters and EQs that aren't overdriven) isn't so hard either given all these tools (Fourier, Laplace, Z transform) we learn in undergrad EE classes. nonlinear and memoryless (e.g. a resistor - diode circuit at low frequencies) isn't too hard either.

however, when you include coupling capacitors, interelectrode capacitance, non-linear tube curves, power supply regulation (or more aptly the lack of regulation, which is one reason using tube rectifiers affect the sound), output transformer behavior (frequency response, non-linearity, and hysteresis), loudspeaker and cabinet behavior to boot, you get a whole slew of nonlinear and reactive behavior that interacts in so many subtle ways that it is extremely difficult to measure it (but might be psychoacoustically relevant). it is truly a nonlinear system with memory and the only straightforward way to mathematically describe such a system with complete generality is with a mathematical technique called "Volterra Series". there is even a method that is very similar to the "training" of an LMS ("least mean square") adaptive filter that you can use to get the coefficients of the Volterra series (when given an unknown system in a "black box") but i shudder at the number of significant non-zero terms (i think there would be tens or hundreds of thousands of terms you would have to worry about).

so instead, some folks are trying to accurately emulate the smaller subsytems (like the output transformer or poorly regulated tube-rectified power supply or, more difficultly, the speaker and enclosure) of the tube amp and then piece these emulated subsystems together. no one has gotten there yet (not even Line6 or SansAmp or Yamaha or Korg, although they might want you to believe they have). but i believe we may before 2010 and then we will have the desireable part of the tube sound without the hum (of a poorly filtered/regulated power supply) or the shot noise (from the random thermioc motion of electrons in the tubes) and without the eviction notice for keeping the neighbors awake.

 

[turbo]

no one has gotten there yet (not even Line6 or SansAmp or Yamaha or Korg, although they might want you to believe they have).

Yeah, it's all marketing hype aimed at kids with more money than good sense. It seems that guitar magazines carry a breathless announcement of a breakthrough in tube-amp emulation with each new issue. A good friend of mine owns a music store, and when I visit him, I try out any new SS amps that claim to emulate particular tube amps (some even claim to be able to emulate many tube amps) - and they are not even close to emulating the easy ones. If any tube amp can be emulated by SS, it should be the Fender Twin Reverb, with it's all-diode rectifier and fairly stiff filtering. Or perhaps one of the bigger amps in the Ampeg Portaflex series - they were sooo clean that they even had "accordion" inputs. It still hasn't been done, though. For the ultimate test, the tube-emulation guys should look to the little tube-rectified Fender Tweed amps, like the 5E3 circuit Deluxe. These circuits are very simple, but their complex tones, lively interactions with the guitar, and their sensitivity to variations in the input signal would make them the some of the most difficult amps to emulate with SS. If they could even get close to emulating one of these little tone monsters, they'd have a handle on the problem, and could take a serious shot at emulating some of the clean amps like the Twin Reverb, or perhaps the AB763 Super Reverb with its fairly stiff GX34 tube rectifier. I don't anticipate that convincing emulation of even the cleanest tube amps will happen in my lifetime - the gap is huge, and it is not closing very fast.

Edited to add: And I fully expect that like in most real-world problems, the last 10% of the progress will take 90% of the time and resources available.

 

[Sojourner01]

The question is; do these kids who buy fancy SS equipment know how to balance and tune their kit to make having an expensive tube amp worthwhile? I'm no musician, but I've fiddled around with guitars and tweaking them to sound the way you want is indeed a huge pain in the arse.

 

[turbo]

Totally disagree. It is possible to produce tube/valve charateristics in a solid state amp.

go to http://uk.youtube.com/watch?v=lgGA6qx2NzQ

pure, gorgeous Marshall plexi style tone using FETs.

More info here... http://members.aol.com/bluetoneamps/vvt.html

You are not even close. Show me a SS amp that can can simulate the sag, dirt, and channel interactivity of Fender Deluxe 5E3. You cannot. I've been in this for a few decades. The holy grail of the SS manufacturers is to claim that "my SS amp sounds as good as this tube amp" and it has never happened - not even once. I am not claiming that people playing SS amps cannot sound good BB King and David Hidalgo put that foolishness to shame. BUT There is no curretnly- available SS amp that can duplicate the response of a Fender Tweed Deluxe, a Super Reverb, or any of the other iconic tube amps of the 50's and 60's.

If you are not convinced, bring me a run-down piece of crap tube amp that has been neglected, and I will return you an amp that will sing.

 

[turbo]

Wrong, wrong and thrice wrong!

Try this... http://uk.youtube.com/watch?v=dN2To3CSuGE&feature=related

and this... http://uk.youtube.com/watch?v=lgGA6qx2NzQ&feature=related

and this... http://members.aol.com/bluetoneamps/vvt.html

seems like you guys missed it!

read the reviews/testimonials.

No harshness, pure valve tone from a solid state amp. See/hear...it did happen in your life time.

trust me!

Not even close. A Fender 5E3 Deluxe or a Fender Bassman (the prototype for the Marshall) would slaughter those SS amps. I resurrected a '65 Fender Twin Reverb and even then I had to ramp down the output and tinker with the rectifiication to get a natural tone. It's easy to emulate notchy distortion with SS - it's a lot harder to emulate the sag and distortion from a tube amp.

 

[rbj]

Wrong, wrong and thrice wrong!

...

seems like you guys missed it!

read the reviews/testimonials.

i think I've heard this before (Monster Cable).

trust me!

they said that, too.

Don't need to, mate. You are truly hard of hearing.

there are all sorts of solid state amps that have some sort of grunge or crunch when cranked up. we are not going to be able to tell the difference by listening to our computer speakers playing some utube.

BTW, i am not at all opposed to the possibility of accurately emulating a variety of tube amps. it's a matter of measurement, modeling, and math. it could certainly be done with FET and op-amp circuits, but they would likely hafta be sort of complex. i think a simple FET amp stage could get you a reasonably good sounding "tube-like" distortion (with 2^nd harmonic distortion due to the asymmetric FET volt-amp curves), but i wouldn't dare say that it would be indistiquishable from the tube amp (perhaps with a tube rectifier DC power supply with low regulation). not until after lots of blind listening tests directly with the gear, not via some youtube.