Physics explains why rock musicians prefer valve amps

February 8, 2017, Institute of Physics

For many guitarists, the rich, warm sound of an overdriven valve amp – think AC/DC's crunchy Marshall rhythm tones or Carlos Santana's singing Mesa Boogie-fuelled leads – can't be beaten.

But why is the valve sound so sought after? David Keeports, a physics professor from Mills College in California, looked at the science of valve amps for the journal Physics Education, to explain why their sound is 'better' to the ears of so many guitarists.

Professor Keeports said: "Although diodes and transistors are cheaper, more practical, and technologically more advanced than glass valves, valves survive because so many guitarists are exacting about their tone, and prefer the sound a valve amp gives them."

"At its most fundamental level, this is because a moderately overdriven valve amp produces strong even harmonics, which add a sweetening complexity to a sound. An overdriven transistor amp, on the other hand creates strong odd harmonics, which can cause dissonance."

Professor Keeports explored the physics of why even harmonics enrich a sound, and why the timbre of the sound from a valve amp changes when a guitar is played more loudly.

First he ran a 200Hz sine wave – a pure wave with a single frequency– through a small Bugera hybrid amplifier, featuring a valve preamp and a solid state power amp. He tested both 'sides' of the amp; first turning up the gain knob, which controls the valve preamp while the master volume knob (controlling the solid state power amp) was set low. He then repeated the process with the preamp set low and the master turned up.

Using Logic Pro X music production software, he examined the resulting sound waves in both frequency and time domains.

Professor Keeports said: "The output from the amp showed that a moderately overdriven valve preamp produced prominent 2nd and 4th harmonics at 400 and 800 Hz, and only a very weak 3rd harmonic at 600 Hz. For the solid state power amp, this pattern was reversed. All of this behavior is consistent with the common claim about the harmonics that valve and solid state amplifiers produce. But the story is not quite so simple. Overdriving the valve preamp harder produces strong odd harmonics."

"The shift toward odd harmonics at increasing gain is a characteristic of valve amplifiers that further explains their appeal. An electric guitar player can overdrive an amp two ways: by turning up the amp's gain control, and by attacking guitar strings more strongly. Experienced guitarists don't just play their guitar – they also play the amplifier. By striking the strings harder or softer, they can change timbre along with volume."

And why does a valve amplifier behave this way?

"The simple physics of triode valve function explains everything," said Professor Keeports. "Valves have two ways to to flatten a sine wave. Overdrive a valve moderately, and it flattens just the top of the wave to make an asymmetric wave that is rich in even harmonics. Overdrive the valve harder, and it also flattens the bottom of the wave to produce a symmetric wave full of odd harmonics."

"The even harmonics provides the complex, warm, rich that so many guitarists desire. Add to that a valve amp's ability to produce somewhat dissonant yet driving sounds when a guitarist attacks strings harder and turns rhythm playing into lead playing, and function creates just the harmonics a rock guitarist needs."

Explore further: Approval expanded for sapien artificial heart valve

More information: The warm, rich sound of valve guitar amplifiers" Keeports D 2017 Phys. Educ. 52 025010. iopscience.iop.org/article/10. … 088/1361-6552/aa57b7

Related Stories

Approval expanded for sapien artificial heart valve

October 22, 2012

(HealthDay)—U.S. Food and Drug Administration approval for the Sapien Transcatheter Heart Valve has been expanded to include additional people with aortic valve stenosis, the medical term for a narrowing of the aortic valve ...

FDA expands approval for 'valve in valve' aortic replacement

March 31, 2015

(HealthDay)—The U.S. Food and Drug Administration said Tuesday that use of the CoreValve "valve-in-valve" aortic replacement has been expanded to include people at extreme risk for serious complications of traditional ...

Recommended for you

New insights into magnetic quantum effects in solids

January 23, 2019

Using a new computational method, an international collaboration has succeeded for the first time in systematically investigating magnetic quantum effects in the well-known 3-D pyrochlore Heisenberg model. The surprising ...

Rapid and continuous 3-D printing with light

January 22, 2019

Three-dimensional (3-D) printing, also known as additive manufacturing (AM), can transform a material layer by layer to build an object of interest. 3-D printing is not a new concept, since stereolithography printers have ...

Scientists discover new quantum spin liquid

January 22, 2019

An international research team led by the University of Liverpool and McMaster University has made a significant breakthrough in the search for new states of matter.

Researchers capture an image of negative capacitance in action

January 21, 2019

For the first time ever, an international team of researchers imaged the microscopic state of negative capacitance. This novel result provides researchers with fundamental, atomistic insight into the physics of negative capacitance, ...

31 comments

Adjust slider to filter visible comments by rank

Display comments: newest first

ab3a
4.3 / 5 (6) Feb 08, 2017
Old news. By the way, a decent tube (valve, if you're British) amplifier also has a very heavy, significant audio transformer to take the high impedance audio output and play it in a low impedance speaker. There are resonances and unusual saturation modes that occur when the transformer input reaches certain levels.

Basically, as the article points out, you're not just playing a note in to the pickup on an electric guitar, you're playing the amplifier as well.
cgsperling
Feb 08, 2017
This comment has been removed by a moderator.
TechnoCreed
5 / 5 (4) Feb 08, 2017
...Carols Santana's singing Mesa Boogie-fuelled leads – can't be beaten.


Goofing on the name of such an eminent artist takes the credibility out of the article by outlining the lack of culture of the writer.

The name is Carlos Santana... Hope this get corrected soon.
gkam
1.6 / 5 (7) Feb 08, 2017
If he gets flattening on the top before he gets it on the bottom of the waveshape, he has a DC bias on the circuit.

But it is true reducing one side of a sine wave will generate even harmonics, while the flattening creates odds. A square wave is an infinite number of odd harmonics.
ExIowan
3 / 5 (2) Feb 08, 2017
The audio engineers figured this out in the late 1960's.

Not mentioned is another phenomena: Tubes handle ultrasonic inputs better than transistors. In the real world, there is a lot of electronic noise that has a frequency greater than 20Khz. Those ultrasonic frequencies cause intermodulation distortion particularly in amplifiers that use a lot of feedback (such as transistor amps). If you filter out frequencies above 20KHz going into an audio amplifier and don't run it close to clipping, it will sound as clean as a good tube amplifier - and will have better bass to boot because of the higher damping factor and higher drive current capability.
koitsu
5 / 5 (2) Feb 08, 2017
And now, huge armies of transistors mimic these very behaviors in amp modeling software, which is quickly approaching the point of being able to fool most if not all guitarists in terms of tone and dynamic response. The experience of sitting next to a nice tube amp and jamming away may never be successfully modeled, but it's still fascinating to see what one can do with just a smart phone or a computer these days.
Whydening Gyre
5 / 5 (9) Feb 08, 2017
"Physics explains why rock musicians prefer valve amps"

Cuz they are the only ones that go to - ELEVEN...
randomcyborg
2.6 / 5 (5) Feb 08, 2017
This also speaks to the reason that jazz listeners prefer vinyl records played through tube driven amplification instead CDs played through transistor amplification, especially if the music was recorded using transistor powered equipment. The chord structures in jazz are built around dissonance, and the odd harmonics of transistor amplification will add unwanted odd harmonics.

I've also found that keyboardists (including me; I'm also a scientist) generally prefer tube amps for pianos and organs, but transistor amps for monophonic synthesizers. I can't give hard data for electric bass, but I think it's about half and half (I'm in the tube amp preferring group of bass players).

Tangential to this is something all musicians know, even if they don't know they know — the wood of the instrument also produces its own set of harmonics, even if amplification is completely transistorized. This is also true for speaker enclosures (for CD and tape player enclosures, as well).
24volts
4 / 5 (2) Feb 08, 2017
Maybe I misunderstood at the time but I thought amp companies figured out how to use fet transistors to mimic tube amp characteristics a number of years ago.
Eikka
3 / 5 (4) Feb 09, 2017
If he gets flattening on the top before he gets it on the bottom of the waveshape, he has a DC bias on the circuit.


Talk about pointing out the obvious. That's how a class-A valve guitar preamp works.

The valve amplifier works by turning a varying voltage input on the grid into a varying current output through the anode and cathode. In a class-A amp, the current is put through a resistor to turn it back to a voltage signal for the next amplifier stage.

http://www.learna...bias.gif
http://www.bartol...sfer.png

In operation the tube is DC-biased at the input at around half the linear operating range of the output. From the graph you notice that the characteristic curve has a knee at one end: when the signal amplitude grows it hits this non-linear knee that makes the even harmonics, and then the current supply limits at the other end which makes the odd harmonics.

Eikka
1 / 5 (2) Feb 09, 2017
the previous examples were of a transistor amp. Here's a better picture of the transfer characteristics of a vacuum tube:

http://earlyradio...vac3.gif

It's mostly the same as a transistor, but you'll notice the smooth changes as it goes into cutoff or saturation at either end of the transfer curve.

Maybe I misunderstood at the time but I thought amp companies figured out how to use fet transistors to mimic tube amp characteristics a number of years ago.


They did. The difference was that the transistor was "too perfect", it had hard and sharp transitions which cut the peaks of the signal right flat when the amp goes into saturation, which makes for stronger odd harmonics and turns the signal into more of a square wave than a squished sine.

Another reason why tubes are preferred is because they're loud. They tolerate being overloaded so a small amp can make a big sound. It just distorts the sound, whereas a transistor amp would burn.
manfredparticleboard
2 / 5 (4) Feb 09, 2017
God...mosfet valve synthesis circuitry has been around for a decade or more. I still like my audionote valve amp though. It's a pity that trying to find good recordings in a lossless format has become harder in an mp3 age. Yes- there's lots out there, but not what I like to listen to.
Eikka
1 / 5 (2) Feb 09, 2017
There's a guitar effects pedal called "fuzz" which does sort-of the same thing as a distorting tube amp by exploiting the non-linear properties of diodes (or transistors).

It's basically an amplifier with negative feedback through diodes, so when the signal amplitude grows to the point where the diodes start to pass current, the negative feedback comes in and limits the signal. That causes it to distort in a non-linear way that makes even harmonics, until the point you overdrive it and it starts to produce odd harmonics.

You got two diodes, one for each half-cycle, so the selection of diodes and input bias point affects how the signal gets distorted. By clever selection of the feedback network, you could create an effect pedal that mimics the distortion of a tube amp.

dirk_bruere
2.3 / 5 (3) Feb 09, 2017
There's nothing that a DSP running at 24/192 cannot reproduce or mimic to the limit of the listeners ability to hear.
eagleon
3 / 5 (2) Feb 09, 2017
I love the way this paints musicians as hapless feelie know-nothings, as if we didn't invent the tools Prof. is using to do the analysis that any rudimentary attempt at valve amp emulation requires. No acoustic or DSP engineers in our ranks, we just blindly flail at Spice until out pops a guitar pedal.
Shootist
3 / 5 (2) Feb 09, 2017
There's nothing that a DSP running at 24/192 cannot reproduce or mimic to the limit of the listeners ability to hear.


crap on a stick
skystare
2.3 / 5 (3) Feb 09, 2017
Question - how much can the physical shaking of the tube by the sound waves affect the output?
Da Schneib
4 / 5 (4) Feb 09, 2017
As a guitar player myself, and an EE, this makes a lot of sense and I'm surprised no one has done the research to determine exactly how this works before now.

It makes a lot of sense; the guy's right, even harmonics' "smoother" distortion sounds better with desired tonalities that lean toward the "clean" side, and odd harmonics better with desires toward the "dirty" side. This also explains why I've had better luck with tube than transistor amps getting the preamp level set to where I can stay clean by playing softer and get dirty by playing louder. With an electric guitar, this works from both the dynamics on the strings and the volume knob (it's easier of course to modulate one's attack at the strings than to have to reach away for the knobs).

There's a lot more to the story, I think, but this is actually pretty valuable to me as a player trying to get things to a state where I can get the sounds I want with the least effort.

Nice article!
randomcyborg
2.6 / 5 (5) Feb 09, 2017
Skystare, I don't have numbers for you (my reference times are buried above me in the attic, and today I'm not up to doing a web search), but physical vibrations affect output a lot. This is true for tube amps, transistor amps, record players, CD players, mag discs in computers, and I'm sure I'm leaving out some stuff. Vibrations are picked up by device housings (along with more obvious forms), some of which affects vacuum tubes in the system — energy is getting fed back, and some goes right back into the electronics.

Every time I set up a sound system, I always cranked up the system, without connecting any microphones (I used synthesizers to create the signal) to see what feedback the system caused in a particular venue so that I could tone it down or crank it up appropriately. Sometimes it was so pronounced the building would shake; other times I'd get really bad ultrasonic feedback that most people couldn't hear, but would unpleasantly reinforce lower frequencies.
Da Schneib
5 / 5 (3) Feb 09, 2017
@Eikka of course has a bunch of comments that are irrelevant to how actual musicians use real amplifiers.

Learn to play beyond novice level first, dude. I mean, really.
Da Schneib
5 / 5 (3) Feb 09, 2017
So, it's the infidelity that is the desirable feature. Interesting....
Been true since the 1960s. Some folks will tell you it was Hendrix, but you can reach back further and note that saxophones produce these same types of tonalities and show differences in tonalities when played gently and when pushed hard.

I listen a lot to sax players when looking for pleasing tones for my axe.
Da Schneib
5 / 5 (3) Feb 09, 2017
Question - how much can the physical shaking of the tube by the sound waves affect the output?
Best not much- this leads to tube failure. It's part of why heads are separate from cabinets. Tubes don't like being shaken around.

Tubes handle ultrasonic inputs better than transistors. In the real world, there is a lot of electronic noise that has a frequency greater than 20Khz. Those ultrasonic frequencies cause intermodulation distortion
Ummm, not. IMD is caused not necessarily by harmonics, but by the sum and difference frequencies of the signal with noise. Ultrasonics are clipped by competent amps (like the dominant Lin amp topology for transistor power amplifiers) and do not produce IMD. Sums and differences of ultrasonics with audible signals are still ultrasonics and excluded by the caps between stages in competent power amps.

Seriously, folks, please learn the difference between Class A preamps and the dominant Lin topology for power amps.
Da Schneib
5 / 5 (3) Feb 09, 2017
Maybe I misunderstood at the time but I thought amp companies figured out how to use fet transistors to mimic tube amp characteristics a number of years ago.
Without the understanding of the types of harmonic distortion that tube preamps produce as opposed to the types transistor preamps produce, it would be difficult to figure this out, and you can't see it even with a high resolution frequency analyzer.

I am aware of certain builders from the 1970s and 1980s who made modified transistor preamps that could produce the sorts of sounds that were considered desirable by tube amp aficionados, and rather sought after as a result due to better reliability and lower cost, but these generally involved a "black box" with impenetrable electronics whereas this analysis lends itself to reproduction with transistors by all comers.
Da Schneib
5 / 5 (4) Feb 09, 2017
I love the way this paints musicians as hapless feelie know-nothings
Hate to say it but most musicians are hapless feelie EE know-nothings. They know sounds; they don't know EE. And that's fine, but don't dis them and don't dis the EEs either. The best musicians know both sides and get the best sounds.
randomcyborg
Feb 09, 2017
This comment has been removed by a moderator.
Da Schneib
5 / 5 (2) Feb 09, 2017
Yep. I had a friend who had one of his boxes. That's one of the ones I was referring to above.
randomcyborg
3 / 5 (6) Feb 10, 2017
Referring to the comment asking if vibrations can affect the sound produced by the electronics, Keith Emerson got some almost reverb-like effects (and some others) by hitting, kicking, and dropping his C-3, in studio and on stage. If add energy to any electronic system, even if it's only in the form of vibrations (kinetic energy), some of that energy will get into the electronics, and produce interesting effects. The main problem is replicating those effects. Kicking is definitely a legitimate signal processor...
ab3a
5 / 5 (3) Feb 10, 2017
As a guitar player myself, and an EE, this makes a lot of sense and I'm surprised no one has done the research to determine exactly how this works before now.


It HAS been tried. Even with some fairly advanced DSP. The problem with replication of the distortion modes is that there are many many effects to compensate for. There are microphonics of various components such as the paper capacitors, there are resonances between the decoupling capacitors and the output transformer. There are distortion products of the tube itself, and the sagging of the high voltage supply filtration.

In other words, it's like trying to figure out exactly what makes a Stradivarius violin sound so good. We know what it does, but we don't really know how to replicate it.
Estevan57
4 / 5 (4) Feb 10, 2017
For my two cents the future of amp modeling is already here: https://www.kempe...overview

I haven't seen anyone who can tell the modeled sound from the original.

By plugging into the amp, it DOES model the various electronics of the amp being modeled, down to the noisy carbon resistors, caps, transformer sag, etc.

I was very skeptical of claims for this, but after direct comparison of amps, modeler, same amps with different tubes, etc. it convinced me.

It can have different models of the same type of amp, and they models all sound like that particular amp, from whisper quiet to OMFG loud.

Sidenote: the Roland JC-120 is a solid state amp, and people seem to love it. I do.

Craig Anderton is still my hero. Anyone build their own effects?

randomcyborg
3 / 5 (4) Feb 13, 2017
Send an organ's output into an overdriven tube amp, and press the A-220 key. The overdriven odd harmonic is at E-330 (perfect 5th). The note produced by the E key, however, isn't E-330 — it's E-329.6. Playing both keys produces A-220 (the key's note), E-330 (the A-220 harmonic), and E-329.6 (the key's note) — two different frequencies for the same note. This is not an artifact of the organ's signal, or of the amp; it's an artifact of modern tuning.

Modern instruments use "even temperament" (based on binary logarithms – for real). In even temperament, every note must be slightly off from "just temperament" (which is based on the integer ratios of harmonics), because just temperament is key specific, and even temperament is not. If an instrument is in tune using just temperament in a specific key, it's out of tune in every other key.

Has anyone compared overdriven harmonics using just temperament with these harmonics using even temperament?
Whydening Gyre
5 / 5 (3) Feb 13, 2017
Wow! Just Wow!
You guys are taking this brain where it's never been before...;-)
"Lucky Man", "Hang Man", "Sympathy for the Devil", "Dark Side of the Moon" and "Where will the Children Play" (not to mention a WHOLE HOST of others) will never sound the same, again...:-)
Am I Experienced?
More all the time...:-)
Thanks!

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.