Tube Guitar Amplifier Overdrive Webpage

[Usable Thought]

You won't find a better book on this subject than Neumann and Irving's Guitar Amplifier Overdrive A Visual Tour. The book includes a lot of graphical content to help make it understandable and math is kept to a minimum.

Yeah but this guy Neumann is dangerous. I went to his supposed "Strings and Frets" web site - it seemed innocent enough at first, articles like "Why Valve Amplifiers" etc. - until I got trapped by the page on the Princeton AOT amp project. I think this is what they call a "gateway amp project," sort of like a "gateway drug." Too many neat pictures & sound clips & ideas!

[Mike Sulzer]

@teemuk
IMD - Something New

Your second reference is applicable to a low distortion situation. That is, it shows sum and differences are much smaller than harmonics with symmetrical clipping and an input of two sine waves for distortion that is just audible. With electric guitars, a much higher level of distortion is often used. in this case, the sum and differences have higher levels. The attached plot used two sine waves of amplitude one (generated in Python) with the sum clipped to .1, fftd and plotted with this command:
plot(20*log10(absolute(fft.rfft(clip(s0 + s1, -.1, .1))))).

[teemuk]

And specifically, if your view is correct, and it's all "subjective," then how do you account for the confluence of interest in particular tones - indeed for your own activity in furthering the development of SS circuits that produce such tones?

I do believe our taste and preferences over vague concepts such as "musicality" are indeed subjective. It does not mean there couldn't be vast number of people sharing similar - if not identical - preferences with each other, but it also doesn't mean we could isolate specific key characteristics of "musicality" of which every individual on this planet would agree on, in each and every context of music. In fact, already if we focus on single musical genre only we will find that different individuals have somewhat varying tastes concerning music within the genre. Some people may like thrash metal music, yet they may still remain unthrilled by band such as Metallica although their music captures essential characteristics of thrash metal. Sharing similar views with other people is different to all people thinking alike and always preferring the same things in equal proportion. It gets even more complex when we put this in perspective of evaluating guitar tones, although we often find them sharing striking similarities within specific generes.

Yes, there are definitely certain trends in guitar music enjoying wide popularity (which naturally also means certain groups of people share somewhat similar preferences). One popular trend, established long time ago, is effecting the natural timbre of the instrument so that it starts to resemble something else. Typical "effects" excluded this is achieved mainly by two means: 1) altering envelope and 2) altering harmonic content of the signal. We can also objectively quantify methods and schemes that produce certain types of effects in this regard.

Like...

Sustain: Instrument signal having an envelope with distinct peaky attack, and rather fast decay does not meet objective definition of a sustaining tone. It's in fact the opposite. For signal to sustain it has to have more "even" envelope throughout, meaning much slower rate of decay. In signal processing means this is achieved by limiting signal dynamics, commonly by means of gain compression (which generates very low amounts of harmonic distortion) or by means of clipping (which by nature generates greater amounts of harmonic distortion). As result the envelope changes, which has distinct effect on instrument's "sound". This part of theory can be objectively agreed upon. But my point is that it is harder, if not downright impossible, to objectively agree upon which specific type of envelope is most "musical". I would say that is very much content-related, as it is a matter of individual preferences.

Distortion:
Timbre of instrument also changes when we modify its harmonic "colour". Almost no instrument produces pure sine waves so in addition to envelope "tone" of an instrument is defined by the harmonic content of the signal (it is usually dynamically variant). By either adding, amplifying or attenuating certain harmonics the colour of the instrument thus can be altered, sometimes radically. We can objectively conclude that in order to create distinctively distorted tones some method to introduce harmonics to the signal that weren't present in it before is required. That's also basically definition of harmonic distortion. IMD expands the theory to cover interaction of complex waves made out of several frequencies. By nature, most methods to alter signal envelope also alter harmonic charcteristics. (e.g. clipping).

We have compiled very valid theories about waveform arithmetics and harmonic frequencies that explains the relation of harmonics and waveform's shape. None of these issues are subjective. All follows strict, defined laws and is thus objective information.

It is much more difficult to agree on what type of harmonic pattern or signal shape actually is more "musical" than another. Many have tried and failed. Just the mere fact that "musicality" of even harmonics vs. odd harmonics has been found to be nothing but a myth (and does not, for example, even to begin explain difference of tube or solid-state amp sound) is a pretty good example of this. If we could agree that certain harmonic pattern is more "musical" than another we wouldn't need a vast range of effect processors / amps, which all produce their own unique patterns of harmonic distortions and altered signal envelopes. In fact, we would have already likely developed the most "musical" instrument that exists and would be contempt in playing no other instruments but it.

Equalisation:
Since harmonics (that is, frequencies in addition to fundamental) define timbre of an instrument adjusting magnitude of frequencies throughout the effective bandwidth of the instrument also has distinct effects to timbre. Not only frequency response of the system but also magnitude of IMD created by harmonic distortion can be affected by PRE distortion equalisation by usual means of hi, band -and low-pass filteltring. POST distortion filtering will in turn affect harmonic frequencies created by the distortion process. This equalisation process, especially in interaction with scheme that introduces harmonic distortion, has distinct effect to "tone" of an amplifier. The effect may be even more distinct than minor differences in harmonic distortion characteristics (e.g. soft vs. hard clipping, symmetric vs. asymmetric clipping).

While we may compile a set of rules to explain what equalisation methods result to specific outcomes (it follows strict laws and is predictable), and while we even find many amplifier designs sharing strikinly similar equalisation schemes overall, no one still has discovered -The- equalization scheme that would be universally considered so "musical" that it would have obsoleted all other, and different, schemes of equalisation. In practice, EQ is probably the most powerful tool to shape amp's signature tone by means of using different equalisation schemes than what other amps use. If this isn't a matter of artistic preference (subjective) then what is?

In practice we live in a world where some people like the tone of Fender Champ while some may prefer tone of Vox AC30 instead. Many people also like both, although - and perhaps because of that - they are very different amps from one another in terms of the effected "tone" they produce inherently. We could isolate, objective parameters and circuit behaviours that would explain why these amps have the characteristic tone that they have. (In fact, you can find a great deal of such research). That still doesn't mean we could conclude that one amp is better or more "musical" than another and regard it as an universal truth everyone agrees on. It ain't happening. People are still building / selling widely different amps. For reason.

In terms of electronics and signal processing we can find solid, objective facts how these circuits work and how they affect the signal that passes through in objective, indentifiable terms. We can actually pinpoint these things to great detail. Thus we can compare these things and learn from them. We can find objective information that explains why Champ amp sounds different than AC30, but I'd be very hesitant to draw any universal concusions about expected "musicality" of these amps from any of that.

Add to that the context-dependability. We would practically have to evaluate such things in every imaginable musical context. We can likely find a large group of people who will agree that tone of Vox AC30 is very "musical" and fitting to 1960's -style instrumental music, unlike overdriven high gain tone of ENGL Powerball amp. Lesser amount of people would likely agree that AC30 has the most "musical" characteristics in context of technical death metal. In that context the entirely different tone of Powerball amp might instead have more fans. These are just traditional expectations of tonal characteristics interlinked to specific musical contexts.

There is no denying that guitar tones accustomarily employed in specific musical contexts often share a lot of similarity in certain characteristics. But effectively we couldn't even file musical compositions under specific genres unless the compositions within that genre didn't likewise share certain degree of similarity. Key characteristics of certain musical genres are usually pretty well defined in objective terms so we can universally agree on what is, for example, country or metal music. In guitar-oriented music such defining factors usually happen to cover also how instruments are supposed to sound (in order to not be too "odd" for the genre). Perhaps not in exact detail, but more or less.

But that's just a guideline. We can define that characteristics like sustain and high amount of distortion are rather essential parts of modern rock guitar tones. That's easy. The jump to subjective realm would mean something along the lines of claiming it is also universally accepted fact that Jimi Hendrix sounded more "musical" than Van Halen... because while both gentlemen employed tones with similar characterstics of sustain and high amount of harmonic distortion, Jimi Hendrix's tone happened to have those characteristics in specific ratio, which must be universally accepted as the most "musical" one over any other example of similar nature. Reality is that we can't make such universal conclusions. We can only have individual preferences. Your's is as good as mine. Such individual preference may naturally be shared by others, but with all of those individuals it still remains a matter of preference and hence a matter of great deal of subjectivity.

Here is an analogy: Musical theory may explain popular and archetypal characteristics of music genres such as, say, rock music. Typical chord arrangements and progressions, riffs, rhythms, characteristics instruments employed and their characteristic tones. We can objectively agree that if we employ such parameters in musical composition then by good chance we end up with something at least resembling rock music. But we can't effectively evaluate those parameters and pinpoint which ones of them would be ones leading to composition that each and every individual out here on this planet would regard as the most "musical" composition ever made in the genre of rock music.

While it's easy to define some objective aspects of things like what is "rock music" or "distorted signal" in gneral, it's a rather large jump from that to objectively define "musicality" in manner which can be accepted by each and every individual under each and every circumstance. That no longer meets the "objectivity" criteria.

[ThermionicScott]

In guitar amps, players using a lot of distortion often tend to use 'power chords', which include octaves and 5ths, but avoid major or minor 3rds. One reason for that may be to minimise IM products.
However, I'm wondering if IM can sometimes be 'good' for distorted guitar and it occurs to me that a common technique is to play two notes, where the lower note is a tone below the higher note, and then ‘bend’ the lower note up to unison with the higher note. In this case the difference frequency will start low and go sub-sonic, while the sum frequency will bend up towards the octave above the two unison notes. Could be a nice effect?

There's a particular Rush song -- I'm blanking on the name since I'm not a huge fan -- featuring a break where Alex plays distorted oblique bends and does exactly that before the rest of the band comes back in. It would sound pretty awful outside of the context of rock music.


ON EDIT: It bugged me not to know which song that was, so I found it! It's toward the end of "Working Man" (6:38):

[Mike Sulzer]

Thanks, but do not just lurk too long; your contributions are very useful.

Hmm, I think we are talking at cross purposes now. I can't understand a word you say because I just don't have your technical background. If I look at what you are saying in a slightly different way, however, I think you are clearly onto something - I just don't know what it is! As I say I used to be an editor & writing coach for professionals; and I developed a sense early on for the relative degree of BS (and I don't mean bachelor of science!) vs. credibility in someone's writing even if I did not share their technical background. I get a sense of credibility from you so that's good.

Anyway I don't think right now I have any more to contribute to the discussion so I'll go back to lurking. For me to really grab onto these topics I'd have to immerse myself for a long time in them on my own . . . and my problem is, there are so MANY topics I would like to do that with, that I have to pick and choose. Learning math; learning about basic circuits; learning jazz progressions on the piano; learning better classical piano technique; and much more. And due to health reasons I have only a couple of good hours per day - though I shouldn't kvetch about it. I just wish there were time enough for me to really do it all.

However this is definitely a topic I hope to hang onto and come back to - whether sooner or later.

[printer2]

Intermodulation is a 'bad thing' in general, as new frequencies are created which are not harmonically related to the original signal frequencies (and therefore sound 'unmusical').
Certainly for hi-fi, PA systems, etc. IM is avoided as far as possible.
In guitar amps, players using a lot of distortion often tend to use 'power chords', which include octaves and 5ths, but avoid major or minor 3rds. One reason for that may be to minimise IM products.
However, I'm wondering if IM can sometimes be 'good' for distorted guitar and it occurs to me that a common technique is to play two notes, where the lower note is a tone below the higher note, and then ‘bend’ the lower note up to unison with the higher note. In this case the difference frequency will start low and go sub-sonic, while the sum frequency will bend up towards the octave above the two unison notes. Could be a nice effect?

Can't spend much time here again, just a couple of comments. I recall reading about some experiments of cleaning up a Bassman or Marshall amp, can't recall which but generally the same circuit. Basically filtered out all the hum from the PS. And the result was an amp with less balls (in not so many words). Seems the IMD of the hum added to the sound. And recordings in North America and England sounded different due to 50 Hz and 60 Hz in use.

While it is interesting to hear about the psychoacoustic effects of distortion this might go a little farther than Rob is looking for.

And one other. My brother works designing electronics, somehow it came up that I was messing with tubes and guitar amps. The new guy there from China nods his head in understanding and replies, 'For the high fidelity'. My bother realizes there is a great cultural gulf to cross. Tries to explain about distortion, decides to grab their production manager. He explains the concept and looks to the PM for conformation. She goes 'Really?' She thought it was just the guitars that sound like that naturally.

The Chinese guy is trying to decide if my brother is just trying to pull his leg or not. No, I am not making this up.

[teemuk]

I recall reading about some experiments of cleaning up a Bassman or Marshall amp, can't recall which but generally the same circuit. Basically filtered out all the hum from the PS. And the result was an amp with less balls (in not so many words). Seems the IMD of the hum added to the sound. And recordings in North America and England sounded different due to 50 Hz and 60 Hz in use.

You must be referring to Project Wildcat...

https://www.ampbooks.com/mobile/clas...ass-AB-ripple/

...If I understood correctly, the point of that particular project was to devise a power supply that retains "sagging" characteristics but minimizes power supply ripple. You can crudely think of tubes clipping to the B+ voltage limit, but because B+ has ripple component it equally affects the clipping threshold. In essence the ripple, along with sag, dynamically modulates the B+ voltage and during clipping these components thus get superimposed to the signal. Ripple is side product of rectifying so it has frequency of 100 Hz if mains frequency is 50 Hz, or 120 Hz in case of 60 Hz. In generic power supply, the more current gets drawn the lower the B+ sags and the higher the ripple amplitude also becomes.




Project Wildcat site gives utmost importance to this modulation, Pritchard solid-state amplifiers deliberately introduce similar effect (they call it "fat") and it's hard to ignore wittness accounts about amplifiers sounding different with different mains frequencies. So I tested the thing myself, likewise by introducing similar modulation to clipping thresholds but to my great surprise heard no distinct difference in comparison to same clipping arrangement sans modulation.

Ok, with practical signals envelopes are such that it's somewhat rare for clipping overdrive condition to sustain for even moderately short period of time. We don't generally drive amplifiers with a continuos sine tone so effects like "sag" or "ripple modulation" are not as apparent as they are when we feed a continuos sine wave input to the amp and begin to overdrive the amplifier in continuos manner for moderately long periods of time. No, in practice and with practical input signals the duration to clip is moderately short and this does not leave much time to introduce effects that are very much time dependent... like gradual voltage sag or 60 - 100 Hz signal component. I don't say these effects don't exist (they do) but typical visual representations of these effects (like waveforms above) really give somewhat unrealistic image of them. Amplifiers are seldom exposed to such continuos signals in practice. However, it was evident from waveforms ripple modulation had -some- effect, yet personally I had to struggle to hear a difference between clipping and clipping with ripple modulated threshold.

That doesn't mean no one else couldn't hear the difference, in fact given devotion of some people to capture these effects I think there are people who perceive the effects very, very clearly. It left me wonder the very good possibility that we humans tend to perceive clipping and distortion in very individual ways, perhaps we focus to some details of it more than we do on others.

[Malcolm Irving]

.....'For the high fidelity'.....

It's interesting that some hi-fi enthusiasts prefer tube amps to solid-state. Solid-state hi-fi amps can have much lower distortion figures, but it seems that some people prefer to hear their music with a bit of added distortion.

[Justin Thomas]

I choose my "hi-fi" based on what is most aesthetically pleasing to my ear, not on specs. And although neither of these counts as a true "hi-fi," I preferred the sound of my crusty old Zenith tabletop radio to the sound of my much "better" truck stereo (a Pioneer, I believe). It was the compromises that made it pleasing.

Justin

[teemuk]

Relevant link to aforementioned Pritchard "fat" scheme:
https://patents.google.com/patent/US5761317A/en
There's actually plenty of more in the patent (it covers the entire tube power amp emulation scheme) but in nutshell the idea is to use a ripple voltage source as one of signal sources that modulate gain of the OTA stage. The magnitude of modulation is made dynamically varying.

An interesting side note is that while Pritchard implements little details like "ripple modulation" to his tube amplifier emulation scheme he intentionally omits modelling bias shifts at grid circuits and resulting crossover / blocking distortion (typical to generic PP tube amps) as he deems the effect detrimental to tone. (An opinion shared by many IF crossover / blocking happens to be too excessive). Peavey, along with many other manufacturers, on the other hand, regards that as an essential part of push-pull tube amp tone and several of their designs have strong focus on capturing the characteristic. They, on the other hand, commonly omit modelling "ripple modulation" because side product of the modulation are "ghost notes", which are often disturbingly discordant.

So, one man's "fat" can mean discordant ghost notes to others and one man's preference for dynamic crossover distortion effects during sustained overdrive may mean disturbing blocking distortion to others. I'd say that the mere existence of such differing viewpoints regarding particular distortion characteristics supports my hunch that people do perceive distortion in widely individual manner.

[uneumann]

Here is an example of ripple modulation and bias shifts at the speaker output of a max'd out Bassman.
The input is a burst sine wave (220Hz) with a background 2200Hz sine wave at ~10% of the burst amplitude.
The DCCF, PI, and output stage are all overdriven by the burst. Their accumulated shift decays and gain variations are visible after the signal burst.

I'm just pointing out that circuit behaviors get really complex when overdriven stages interact in a complete amplifier system. It's beyond me how any theory or measurement can relate this behavior to a subjective sense of "like it" or "don't like it"... Even if you could measure something and correlate it to a subjective like/dislike - simply turning the tone stack knobs will produce very different different looking waveforms.

That's not to say it will never be done... I've learned to avoid the "never" word. Somebody may come up with something tomorrow.

[uneumann]

I think you are too close to your particular clump of very interesting trees to see the trees that someone else is looking at; or the forest that still someone else sees because they are farther back.

Or to use a different metaphor: You are looking through a very particular analytic lens. So you can only see what can be seen with that lens. It is great for your purposes but useless for many others. Outside our own chosen domains, we are all laypersons & usually unaware of it. So it is dangerous to say what others can or can't see when you don't even know what lens they would use to see the very interesting trees they are looking at. Likely you would have to start off from scratch learning their field - even with significant technical overlaps; e.g. they too might be interested in waveforms, yet have very different criteria for analysis and classification.

I will tell you one thing I really like: your co-author's openness & curiosity towards interesting questions w/out bothering to get all heady as I have just done. He seems to enjoy playing around. "What if we did X? What would happen? Maybe it would be interesting!"

I also like your web site very much, BTW, but need a complete new brain to understand the technical stuff.

Not the first person to suggest I'm too rigid in my thinking... ask my wife and kids.

A bit off topic, but definitely related... I have a long-time interest in how creative people see things and do what they do... how do they think about what they do? Anyway - you may find this interesting... "Get Past The Facts"? FRESIA
Jerry has a way of explaining what can't easily be explained.

[Usable Thought]

If this is just an ad seems out of place.

[Enzo]

There is a time limit on editing posts. I think it might be as much as 24 hours.

[Justin Thomas]

No, it's a red eft.

https://en.wikipedia.org/wiki/Red_ef...otted_newt.jpg

And it says so in post 29, right after the one you quoted.

Justin