Tube versus solid state: The truth behind tube distortion
Nov. 9, 2018
- Revised on Dec. 20, 2020 ('Introduction' and 'About amplifier listening tests...')
In this article I will examine some of the myths and hype related to tube guitar amps and tube distortion. In my view the real answer lies in guitar amp modeling and not in tube emulation or 'how tube works'. In fact, we don't need tube emulation to get smooth tube distortion.
One of the main reasons why guitar amps and vacuum tube distortion are misunderstood is that the most common approach (harmonic distortion analysis) is not suitable for characterizing the distortion of complex waveforms in multi-stage circuits. The guitar signal is a complex wave with many harmonics and not a simple sine wave. Without the right approach there is no right answer, and all methods based exclusively on harmonic distortion analysis 'just digging in the wrong place'.
It's important not to cherry pick. For example, when a practice combo (a transistor amplifier) sounds awful then it is always 'blamed to the transistors'. However, the awful sound of a transistor amplifier is caused by the poor circuit design (mainly by the poor filters) and not the components themselves.
Brief historical overview
Tube amplifiers came about long before solid state amps. In the golden years of rock n' roll guitar players had no too much choice: tube amp or no amp. Later, in the early 70's solid state guitar amplifiers appeared, but these first attempts sounded too "cold" or lifeless (one good exception is the Marshall Lead 12). This gave the solid state guitar amps a bad reputation.
In the 80's and 90's lot of "solid state emulating tubes" type circuit were born (and patented). These can be divided into two main groups: one group of circuits that emulates the asymmetrical clipping of tube stages and the other group of circuits that emulates the power amplifier frequency response with speaker load. An example of the latter is the current feedback in the power amplifier which is used nearly in lot of solid state guitar amplifiers nowadays (Peavey TransTube series, Marshall MGs and AVTs, Crate amplifiers, Laney TFX/LV/LX series).
While the main interests of patents and academic research are how to model vacuum tubes in general, the real goal is to capture the tone of certain guitar amplifiers. This second approach is just more convincing.
Some remarkable analog amp modelers from the past and from the present (some of them are out of production):
- Tech21 SansAmp Classic was the first amp-in-a-box (1989). It has no 'dedicated' tube emulation, instead it has clever filtering and op-amp gain stages and an analog speaker-microphone simulation on the output. SansAmp Classic was followed by the GT2, TRI-A.C. and the rackmount SansAmp PSA-1.
- Marshall DRP-1 (1994) another op-amp based design with separate preamp and power-amp emulation.
- Other amp emulators for classic british tone: Zvex Box of Rock, King of the Britains, Boss OD-3. These pedals require tube power amp simulation, guitar cabinet and some tweaking...
To make it clear, despite its name the Tube Screamer pedal and its clones (Boss SD-1, Fulltone Fulldrive etc...) have nothing to do with tube emulation. Tube Screamer was designed for replacing transistor based treble boosters and not for tube amp emulation.
About amplifier listening tests and demos
Guitar amplifiers have many internal parts, stages and all of these stages affect the sound. The preamp, the tonestack and the speaker(s) have the greatest impact on the tone. This makes testing difficult, especially if we know nothing about the internals of the amps. This complexity is a good source of confusion and cherry picking as we can't create a general judgment about the technologies by a simple test.
Usually the devil is in the details. For example, Peavey Bandit (TransTube series) has an input impedance about 250 kOhm, which is much lower than the 'standard' 1 Mega Ohm, and low impedance sucks out the high frequencies from passive pickups. Now put a tube amp next to a Bandit and you can say that solid state amps have a darker sound (the remedy to the low input impedance is to turn down the volume on the guitar to 3-4, or use a clean booster..., or fix it). The other important factor in listening tests is the speaker. Different speakers may sound very differently. Comparing a solid state amp equipped with a low budget speaker with a tube amp equipped with a high quality Celestion V30 or G12-M is again nonsense.
There is another common misconception about guitar amps and power amp distortion. It is well known that classic tube amps have to be played at high volume to get power amp distortion. They have to be cranked to get their specific tone (Fender Tweed Deluxe, Bassman, non master volume Marshalls). The power section of solid state amps designed to be stay as clean as possible, but it doesn't mean that they cannot be cranked. In solid state amps the power amp stage starts to distort when the volume is turned up to 8 while a tube amp starts to distort around 3-4. This power amp distortion can give a sweeter tone to a solid state amplifier too. So this is not specific to tube amplifiers.
Misconceptions about tube distortion
These are the most common myths and misconceptions about tube amps and tube distortion:
- Vacuum tubes are slower than transistors.
- Vacuum tubes have soft clipping (signal is rounded), while transistors have hard clipping. Or in other words: tube amplifiers have smooth transition from clean to overdrive.
- Vacuum tubes generate even order harmonics, transistors generate odd order harmonics.
Vacuum tubes are not slower than transistors. In fact, some old germanium transistors are really slow devices and they are 'slower' than tubes (e.g. AC128). On the other hand it does not make much sense to talk about the frequency response ("bandwidth") of an electronic component because the frequency response depends on the circuit (or at least on the actual stage).
And the same rule applies to the distortion: an amplification stage's distortion depends on the whole stage and not just on the type of the components. Most important factor is the amount of feedback. Because global feedback is generally much larger in solid state power amplifiers and in operational amplifier stages than in tube stages they clip the signal sharper. But this is caused by the feedback only and not the components themselves. It doesn't make sense to talk about "tube type distortion" or "transistor type distortion", because they can be formed in a variety of ways. Only crossover distortion is trickier, but not impossible to emulate.
Marshall JMP 1987 power amp distortion - the signal is hard clipped
(power tube distortion only, no phase inverter distortion)
But why do push-pull tube amplifiers clip the signal hard and not softly, even though they have very little amount of negative feedback? The answer to this question is that vacuum tubes may have three different types of distortion (grid clipping, saturation, cutoff) and in power tubes (in a push-pull arrangement) grid clipping and saturation happen simultaneously. Both sides of the signal are soft clipped at the anodes of the power tubes (this is the power tube saturation) and at the inputs of the power tubes (grid clipping). They have diode-like soft clipping characteristics separately, but their combined effect added together and what we get is hard clipping (plus some crossover distortion, see figure above).
This is true for all famous vintage guitar amplifiers, such as Marshall Super Lead 100, Fender Tweed Bassman and Vox AC30. Power stages of these amps clip the signal hard and not soft - as commonly believed. The smooth tone comes from the topology (design) of the amplifier and not from the power tube distortion. We can measure the harmonic distortion of power tubes, but it is pointless because it will be the same as an op amp with a LED clipping at the output.
Another widespread myth that vacuum tubes clip the signal more asymmetrical than transistors (BJTs). In fact the opposite is true! A typical class A transistor stage clip the signal more asymmetrical due to the larger bias shift, than a typical common cathode tube stage (even if they are biased to mid). Just look at the fuzz pedals and what they do with the guitar signal. Actually, designing a simple transistor gain stage with symmetrical clipping is a real challenge without using diodes.
And finally the most popular myth: asymmetrical clipping generates even-order harmonics, symmetrical clipping generates odd-order harmonics. Why is it a fallacy? Because, harmonic distortion analysis is valid only for sine waves, but not for complex musical signals. The guitar signal itself contains both even-order AND odd-order harmonics, so it does not make sense to add even order harmonics to it... What is important that asymmetrical clipping have different intermodulation distortion characteristics (louder fundamental, less high frequency content). Everyone can hear the difference between symmetrical and asymmetrical distortion, but the difference is not the higher versus lower even order harmonics. Sometimes asymmetrical clipping sounds smoother than symmetrical clipping and in some types of circuits symmetrical clipping will produce a smoother distortion. This behaviour depends on the actual shape of the waveform - which depends on the filters before the clipping stage.
The sound of tube amps
Good distorted guitar sound is not connected directly to components or technologies (tubes, transistors, op amps...). It is about how the guitar signal is filtered before and after the distortion.
Tube power amplifiers have a hidden EQ (extra bass and presence) which is caused by the speaker's impedance curve and the relative high output impedance of the power amplifier. This has to be included in the model otherwise the sound will be too flat. The current feedback is a simple solution to emulating the non even response of a tube power amp.
The above response (transfer function) is modified by the distortion of the speaker, the output transformer and the power tubes. E.g. speaker distortion (cone excursion generated distortion) and output transformer distortion reduce the amplitude of the resonance peak. Power tube saturation flattens the overall response. This response changes can be measured or simulated directly too.
The real difference between the bluesy vintage tone and the modern "scooped" rock tone come from the position of the tone stack. In non-master volume amps (Vox ACs, Fender Tweeds, Marshall Super Leads, JTMs) the tone stack comes before the main distortion stage and not after as in modern amps. But this is basically again a filtering issue not a tubes versus transistors question.
There is no question about that asymmetrical distortion is characteristics of tube amp sound, but the amount of asymmetry varies greatly with both the amplifier design and the amplitude of the input signal (and even with the frequency of the signal!). In classic 50 Watt and 100 Watt Marshalls the most asymmetrically clipping stage is the phase inverter: the duty cycle of a clipped sine wave can be as high as 60%. This is much higher asymmetry that occurs in 12AX7/ECC83 preamps with normal bias, where the maximum value of the duty cycle is about 55%. But when a vintage Marshall fully cranked, the distortion will become less asymmetrical due to preamp distortion. And there are amplifiers where the phase inverter distortion is nearly symmetrical: e.g. Vox ACs and clones.
So that preamps are not left out, the famous JCM800 2203/2204 preamp has nearly symmetrical distortion, in spite of the fact that the second stage is rather cold biased.
If someone asked me why tube guitar amps sound better than solid state amps my possible answer would be the following. Professional tube guitar amps are designed for professional musicians while solid state guitar amps with tube emulation are designed for the mass market. The majority of solid state amps are intentionally "under-designed", so the differences are hidden in the design and not in the technology used.
This article reflects my point of view, which is based on my experience in upgrading amps and pedals, measurements, studies of amp schematics, tube data and realistic SPICE simulations of guitar amps that made up realistic (tested) tube models and realistic speaker impedance models. Amp design books (and sites) may give us a good electronic background on this subject, but their conclusion is almost always questionable or missing...