Tube Amps vs. Algorithms #2: Equipment

First, the classic

So let's start with a classic amplifier. As we discussed in the last part, tube amplifiers are still very popular, even though their solid-state competitors have taken over some of the market from them. To compare, let's go with a classic tube amplifier design, which is the evergreen that digital modelling is trying to emulate the most. Such a "normal tube" consists of several basic blocks: preamplifier, equalizers, power amplifier and power supply.

1. Preamplifier

The function of the preamp is to modify the signal from the instrument so that it can be further used. First and foremost, it is about amplifying it to a sufficient level. At the same time, however, the preamp may create distortion and change the signal otherwise. At the guitar cable input (usually a 6.3 mm jack) there is usually an input circuit that defines, among other things, the input impedance. Some preamps have two inputs – these can vary in input level and input impedance. In the preamplifier, the signal passes through a varying number of amplification stages which amplify the signal so that it is ready to enter the power amplifier at the right level.

Capacitors separate the amplification stages in the preamplifier. Therefore, the preamplifier essentially consists of tube amplifiers and capacitors, which separate the DC component and whose value affects the transmission of the lowest frequencies. Behind the first tube, we usually find a device called "gain."

The tube stages are wired in such a way that it affects the setting of their operating point. That is, some are set to work in a linear part, some on the contrary to cause signal clipping, for example, in a "soft" or "hard" way. In this way, the tubes are lined up in a cascade and tuned for the desired effect.

Before proceeding to the equalizers, we should mention how tubes work in general. What exactly is a tube (or valve)? It is a glass, remotely bulb-like vacuum or gas-filled component used to amplify, generate or control electrical signals. The basic principle of its operation is to control the flow of electrons between electrodes through an electric field.

The cathode emits electrons due to the thermionic phenomenon whereby heated material releases electrons. The anode then captures the electrons emitted by the cathode. The flow of electrons between the cathode and anode generates an electric current. The grids in multi-electrode tubes control the flow of electrons between the cathode and anode by means of an electric field. And why does it actually amplify? Because a small change in the grid voltage causes a large change in the anode current. The output signal (from the anode) then has a larger amplitude than the input signal.

2. Equalizers

Equalization means frequency modification of the signal. That is that popular knob twisting, the purpose of which is to achieve a frequency characteristic that is pleasing to our ear. The simplest equalizers are single-band, and they are what we call passband. We use a potentiometer to cut off the lower or upper frequencies. Multiband equalizations are more complex and offer more precise sound adjustments. Over the years, various major manufacturers have developed types of equalizers that are typical for a given brand (or its imitators). That's why guitar nerds talk about Marshall, Fender or Vox-type equalizers.

3. Power amplifier

The power amplifier is the device that sends the amplified sound to the speaker. Its output power determines the "power" of the amplifier and the speaker creates the load.

Here we'll stop for a while to talk about types of amplifiers. In the tube guitar world, you'll most often find Class A, AB, and in modern solid-state amplifiers, Class G and D amplifiers. What does it mean? A type A amplifier has the operating point of the tube (or transistor) set at the centre of the linear part of the transfer. Thus, the transistor or tube amplifies the signal over a full cycle (360°). It has a lower distortion because the amplifier operates in the linear region all the time, and it also has low efficiency (typically 10-30%) because the system consumes power even without an input signal.

Type AB works by having two output devices (tubes or transistors) amplify the alternating halves of the input signal with partial overlap. Thus, each amplifies slightly more than half (180°) of the input signal, usually around 200-210°. Because of this small overlap, both devices amplify simultaneously over a tiny part of the sine cycle, minimizing distortion at the transition between the positive and negative parts of the signal (so-called transient distortion). Efficiency is significantly better than Class A, where the device is constantly heating up due to the continuous quiescent current. AB typically achieves 50-60%.

Class G and D solid-state amplifiers greatly improve the efficiency of amplifiers that are otherwise quite low, up to 80%.

In power amplifiers, we often find so-called negative feedback. It is routed from the winding of the output transformer to the cathode of the phase inverter (a circuit that splits the signal into two parts, reversing one phase). The feedback affects the gain of the amplifier, its distortion and also its frequency response. Negative feedback also reduces the gain of the amplifier. The well-known "Presence” is frequency-dependent feedback with the possibility of controlling its intensity by turning a knob.

4. Power supply

In tube amplifiers, the power transformer is no piece of cake. For tubes to work as they should, a much higher voltage is needed than for transistors. In addition, the tubes need to be heated. The transformer in tube amplifiers has a secondary winding to power the glow. The voltage from the transformer is rectified by diodes and a subsequent filter.

The design of tube amplifiers generally has its specifics. Transformers are quite large and heavy. Therefore, the amplifiers tend to be in solid metal chassis. The devices also get very hot, so sufficient cooling must be provided and materials that can withstand elevated temperatures are used. No wonder such "guitarist's delight" sometimes weighs a few kilos. Add to that a box with four twelve-inch speakers and steep stairs to the stage, and you're looking at a slipped disc.

The subject of discussion – often almost "metaphysical" – is the miraculous sound properties of tube amplifiers. If we ignore the explanation that it is voodoo or alchemy, we arrive at the fact that tube amplifiers have a higher proportion of harmonic components and that the sound is more dynamic. This then translates into practice by appearing louder.

Under certain circumstances, when overdriven, tube amplifiers have harmonic components distributed in the lower part of the frequency spectrum, and the second harmonic may predominate. Solid-state amps, for example, have harmonic components distributed throughout the spectrum when overdriven, with odd harmonics that are not pleasant to the human ear. Of course, the manufacturers of solid-state amps are aware of this and have ways to correct this "handicap" to get closer to the pleasant sound of tubes.

So much for the classic.

And now the imitators

And who are these "imitators" anyway? Digital guitar amp simulation technology – i.e. amp modelling – attempts to replicate as closely as possible the sound and response of analogue guitar amps, effects and speakers. What does it consist of? An input circuit, an ADC, a processor (DSP), a DAC, an output circuit and a power supply. In addition, control software is required to operate this device and there must be some user interface to control it.

1. Input circuit

The input circuit receives an analogue signal from the guitar and adjusts its level with a preamplifier for input to the ADC (Analog-to-Digital Converter). Impedance matching then ensures proper signal transmission, especially from passive pickups, and filtering suppresses unwanted hum.

2. Analog-to-digital converter (ADC)

An ADC converts an analogue signal into digital data (numbers) that can then be processed by a processor. Sampling rate and resolution are important here. Both have a direct influence on the quality of the result.

3. Digital Signal Processor (DSP)

This is the core of the simulator where the signal processing takes place. What happens in it? For example, the simulation of tube circuits (modelling harmonic distortion, dynamic response, interaction between stages), the simulation of loudspeakers and microphones, impulse response (IR), speaker boxes and microphone positions. In addition, analogue effects can be simulated.

4. Digital-to-analogue converter (DAC)

A DAC converts a digitally processed signal back to an analogue signal for output.

5. Output circuits

At the end of the signal chain, the output signal needs to be properly matched for connection to an amplifier, mixing console, headphones, etc.

6. Power supply circuit

Of course, there must be a stable power supply for all parts of the device.

Playing with numbers

And how does the "imitation" actually work? Through algorithms that "numerically express their real pattern". Algorithms have different definitions. For example:

1. Physical modelling

Physical modelling is one of the most accurate methods for simulating the behaviour of real amplifier components. This algorithm creates mathematical models of individual components such as tubes, transformers, resistors, capacitors and signal distortion. Each component has specific nonlinear behaviour and the physical modelling algorithm incorporates these nonlinear responses to the input signal. This ensures a reliable response at different volumes or playing dynamics, which is crucial for realistic sound.

2. Wavetable synthesis

This algorithm stores samples of sound waves recorded directly from real amplifiers. The samples are played back and combined based on parameters such as playing intensity or dynamics. Although it is less accurate than physical modelling, wavetable synthesis is efficient and less computationally demanding.

3. Impulse responses (IR)

Impulse responses are recordings of the acoustic response of a particular loudspeaker or box to a particular sound impulse. Recorded IRs contain details of the sound of the speaker, cabinet and microphone, including spatial effects and resonances. This is a very fashionable thing to do today. IRs are used for cabinet simulation because they allow a realistic reproduction of the character of different loudspeakers, which contributes greatly to the authenticity of the sound.

4. Wave-shaping and non-linear distortion

Non-linear distortion algorithms are used to mimic the distortion that amplifiers produce at higher volumes or different signal levels. Wave-shaping is a method that reshapes the input signal to mimic how real tubes or transistors distort sound.

5. Dynamic Filtering and EQ

Dynamic filters and equalization are used to mimic the tone circuits and filters of real amplifiers. Each amplifier has a unique response to different frequencies, and these algorithms allow accurate modelling of tonal characteristics. Dynamic filters change their effect depending on the input signal, allowing for natural changes in the sound, such as in response to playing dynamics or the interaction between the guitar and the amp.

6. Neural networks and machine learning

The big thing in recent years has been so-called "neural networks" that analyse and learn the sound characteristics of amplifiers and effects. Neural networks, such as in the case of the Neural DSP Quad Cortex, train their models on a large number of recordings of real amplifiers and can then generate a similar sound. Machine learning brings the ability to model complex non-linearities and variations in sound more faithfully.

7. Linear and non-linear phase delays

This algorithm simulates the time and phase differences between different frequencies, which is important for natural surround sound. Amplifiers have specific behaviours in the time spectrum that create tonality and depth in the sound. Phase delay algorithms can mimic this behaviour.

Real-time digital signal processing is a process in which recording, processing of the recorded signal and playback of the recorded signal occur "simultaneously". In fact, they cannot be completely simultaneous, so there is always some delay between recording and playback. It is technically called transport delay, but the term latency has come to be used for it.

Its magnitude depends primarily on the type of algorithm used and the characteristics of the target system on which the processing is taking place. In the early days of digitalisation, this was one of the key problems to be solved. Cheap modellers had a perceivable latency, and then playing with it was not much fun. Today's high-end devices have already dealt with this issue so that you don't notice anything at all when playing.

So what does this imply for players who don't care about cathodes and convertors and just want it to play nicely? What should they buy? There is no clear answer to this because it depends on many circumstances. And we'll discuss these in the next and last part of our series.