The Art of DSP Saturation

I’ve discussed compression in depth over the past few articles, and while we pointed to saturation as either a desirable or undesirable byproduct of dynamics processing, we didn’t dive into how and why intentional saturation is also an essential part of music production.

Overloading the physical components of analog equipment results in audio saturation. Like a sponge to water, a given component has a threshold of energy tolerance before it saturates. Technically, saturation is distortion. Colloquially, since distortion often has a negative or specific connotation, saturation usually refers to a more nuanced level of harmonic distortion. For example, passing signals via transformers, tape machines, or tubes often results in a sonically attractive output because the phenomenon of component saturation results in harmonic enrichment and dynamic compression. Used creatively it can subtly or dramatically alter the character and color.

From the historical perspective of the electronics engineer building professional-grade or audiophile analog products, all types of distortion were considered to be problems rather than features. The goal of professional audio capture and reproduction was for every component from microphone to playback to be as transparent and “true” to the “source” material as possible unless intentionally pushed into distortion by the producers or engineers to achieve a specific sound. Manufacturers competed to achieve the best specifications: low harmonic distortion, minimal cross talk, high signal to noise ratios, low self-noise, minimum intermodulation distortion and beyond. When digital audio emerged in the late 70s and early 80s, its potential represented a “holy grail” of sonic purity.

Under the hood of a 1960s eight channel EMI REDD.37 console developed at Abbey Road studios available from Vintage King audio, we find a meticulously hand-crafted assembly of tubes, transformers, potentiometers and more. The size of a large office desk, its practical functionality is surpassed by some solid state mixers under $500 today, but the range of harmonically enriched sound it was capable of is incomparable.

From a mathematical perspective, digital audio is a blank canvas. It adds no color. It adds nothing. If we generate a perfect set of sine waves within the frequency range of human hearing in the digital domain (below the nyquist frequency, or half of the system sample rate), and store the summed waveform in the digital domain as PCM at 16bit 44.1, it will be mathematically identical to the original on playback. That digital representation of waveforms may also be altered mathematically with precision.

Conversely, all analog systems, including analog recording mediums, amplifiers, microphones and speakers have inherent non-linearities. They all, no matter how expensive, add some imperfection and character. They color the sound.

As soon as digitally stored or generated audio leaves the digital domain via D/A conversion, it’s going to be colored by the D/A conversion, and whatever amplification circuitry and transducer is used to move the air that ends up entering your ear holes. None of these systems are perfect, and most consumer equipment for playback, including car stereos, headphones and earbuds are varying levels of imperfect junk from the perspective of fidelity. Most of the time consumers are listening through something “good enough to hear the tune” and many aspirational consumers spend (x) thousand dollar audio systems but place them in a room with the acoustics of an echo chamber.

It’s Too Clean

As digital audio systems encompassed more of the recording process, from capture, processing, storage and mixdown, it became apparent to many music producers that, despite finally achieving that holy grail of sonic purity, the character of digital was too perfect — that something essential was missing. Audio guru and Crane Song founder, Dave Hill, once said that, “Pro Tools sometimes has the problem of being too pure.” For the folks who make audio recordings professionally, vintage analog equipment continued to be prized because of the pleasant color of saturation it imparted. Saturation was no longer perceived as an inevitable and often undesirable byproduct; it was an intentional artistic option in an otherwise flawless capture medium.

The possibilities of digital signal processing are effectively boundless and only limited by their designs. DSP is perfectly capable of saturation, but due to the limitations of how digital audio is implemented, it’s surprisingly CPU intensive to faithfully recreate the “dirty saturation byproducts” inherent to most all analog circuitry. The past 30 years have been an exploration of this medium, and much of it has involved the reverse-engineering of analog non-linearities to satisfy the subjective tastes of artists and consumers.

This is where things get interesting. Few analog systems made up of physical components are consistent to their specifications. A circuit may contain hundreds, thousands or even millions of components, each with a range of variance and tolerance. I used to collaborate with a microphone company that sourced their bespoke designs to Chinese factories. Their office and warehouse had boxes upon boxes of defective samples. Precision in mass production is hard without robotics and rigorous quality control standards. Molecular compounds degrade over time. Real world equipment may be subjected to electromagnetic interference, vibration, quality of power sources and a range of temperature just to name a few of the many wildcards at play.

The biggest difference between working in pure analog systems and digital systems is that most analog signal chains from microphone to tape offered a forgiving operational range of overload and gradual saturation. Pro systems were clean within optimal operating ranges, but as you pushed more signal into the components, they would “break up” or saturate progressively, adding harmonic energy and compression automatically. With digital systems, there’s none of that built in – either the signal is flawless or it hard clips past the mathematical boundaries, and that clipping is usually unpleasant.

Noise Or Color?

Noise is a term used in electronics and information theory to describe random, unpredictable, and undesirable signals. Noise, by this definition both random and unpredictable. Harmonics generated by saturation may be complex, but may also be mathematically predictable in the digital domain. A signal clipping formula will predictably generate the exact same harmonic output and “noise” imprint on the same waveform input no matter how many times you repeat it. Is this noise?

The predictability of DSP is an advantage for the modern producer. Imagine an effects chain that never produced the exact same results on playback. Each playback, print or export would be unique, which could be maddening in situations where you wanted to change a specific variable in your mix, but dozens of other variables were randomly different.

Today’s electronic hardware engineers building physical analog circuits use digital models in a software program like SPICE to test how something will work before they invest in prototyping them. Monte-Carlo formulas account for randomness of available components within an acceptable range of operational tolerance. It is improbable to perfectly replicate the randomness and unpredictability inherent to analog systems under all possible conditions in DSP, but engineers can certainly design them to operate reliably within the intended specifications with absolute precision. When it comes to physical circuit development, even with sophisticated computer circuit modelling, there is still discovery and happy accidents during the prototype phase. On the other hand, with digital perfection comes the ability to strictly eliminate or intentionally introduce a predictable range of randomness using the exact same approach if we’re trying to model an analog component as a DSP effect.

A DSP saturator, compressor etc. can operate with predictable precision, or it can have a range of randomness intentionally added (and some actually do without advertising it) to provide the esoteric flavor observed when measuring analog circuits with lab equipment. It’s whatever the designer or software engineers want to achieve. Given most hardware circuit design is born in a computer model, it’s no surprise that relatively primitive vintage circuits can be reproduced as software for ITB use. In fact, with increased CPU power now available in the average computer, reverse engineering classic analog circuits from schematics using something like LIVESPICE is an increasingly common approach to create faithful emulations of prized hardware.

Saturation is no exception, and DSP engineering has come a long way in implementing fantastic sounding saturation designs for the ITB mix engineer. To understand saturation, let’s look at the flavors of saturation we encounter in both analog and digital domains to gain a better understanding of how and why they sound.

Developed by the TBProAudio team in conjunction with GearSpace contributor feedback and suggestions, the GSat+ is remarkably flexible and free saturation plugin that offers a wide range of options.

Riding The Wave

Some mild to medium clipping of a waveform with rounded (soft) peaks may not even be noticeable and results in a gentle musical saturation.

Harder clipping deeper into the waveform results in more obvious distortion and odd harmonics typical of a heavy fuzz distortion circuit.

The symmetry of waveform determines harmonic content. When we observe waveshapes containing harmonics, we can see just how subtle symmetry and clipping are.

Simple waveshaping: The graph above illustrates an output wave in blue which is the SUM of the raw sine input and the resulting differential odd or even harmonics, both natural number multiples of the input frequency. The left represents an asymmetrical waveshape that contains a perfect even order harmonic: the northern peak of the wave at left is flattened and the southern valley of the wave is extended. On the right we see a perfect odd order harmonic where the wave is symmetrically flatted both north and south. For guitar pedal enthusiasts, this may be familiar. Fuzz and Distortion pedals came first, were relatively simple circuits and produced basic symmetric clipping. Overdrive pedals like the TubeScreamer and OD-1 are relatively newer and more complex circuit designs that approximated the sound of tube amps using clever diode configurations combined with op amps to achieve softer asymmetric clipping.

Tube Saturation

Without splitting hairs about the fact that both solid state and tube amplifiers can generate many variations of waveshaping saturation depending on their topology, we’ll broadly categorize tube saturation in simple single-ended Class A amps as “asymmetric soft clipping.” Because classic tube circuits frequently exhibit asymmetric distortion, they significantly increase even order harmonic distortion. If you’ve played around with simple subtractive synthesizers, you may have noticed that asymmetrical patterns like sawtooth waves produce a mixture of both odd and even harmonics, while symmetrical ones, like square waves, are extremely dense, harsh and contain just odd harmonics.

In a triode vacuum tube, a cathode emits negatively charged electrons, which are attracted to and captured by the anode, which is given a positive voltage by a power supply. The control grid between the cathode and anode regulates the current of electrons reaching the anode. Negative voltage on the grid repels electrons back toward the cathode. Positive voltage on the grid allows more electrons through, increasing the anode current. A given change in grid voltage causes a proportional change in plate current. In a simple amplifier application, when grid current is a waveform the anode output will be an amplified waveform.

The 12AX7 is a popular tube found in many devices. Low cost “tube color” preamp designs and guitar pedals may intentionally “starve” the plate current so that its operating range is in a non-optimal range and results in more saturation.

The relationship from grid voltage to plate voltage with a fixed plate power supply and load resistor is non-linear power law function. Tubes can achieve maximum values with softer curves. The tops of too-large waveforms are softened before being eventually limited. This shape creates a “musical” saturation transfer curve. Second order harmonic distortion, or musical octaves are the source of the “typical” tube tone.

ART’s best-selling low cost single-tube preamp employs an intentionally starved plate for harmonic saturation. These have been extremely popular since they were introduced in 1995, and utilized by many hobbyists seeking to impart “tube tone” to their sounds. However, it’s not the same as high voltage tube saturation of a properly designed preamp: it’s more like fizzy diode distortion.

If you enjoy understanding how tube amps work, Rob Robinette’s now classic webpage is still up and running. Below is a 1940s Westinghouse film he annotated for educational purposes.

Transformers

Myth or fact? Enthusiasts opine that Carnhill transformers manufactured in the UK are a key part of the sound of the classic Neve 1073 console strip. Rupert Neve is a transformer designer/builder himself and collaborated with UK manufacturers Carnhill and Mariner to mass produce components to his particular specifications for his products. Neve transformers are a Neve specification for Neve products, regardless of the manufacturer and can be reverse engineered and measured.

I like to think of transformer saturation as low end and midrange “power chord” saturation that adds “weight” and “girth” to signals. Years ago I read an interview with Pete Townshend where he remarked that power chords through overdriven amps sound larger than they are because the harmonic distortion of the amplifier or distortion pedal adds additional harmonic information to the three note (or two if you discount the octave) chord you’re playing on the guitar, and the harmonics are so complex that human hearing tends to “fill in” whether it’s major or minor based on the broader context of the music. The music of Nirvana is a perfect example. Kurt Cobain usually played simple barre power chords through heavy distortion for a “big” sound, and the chord progressions provided the implied context of the key and mood.

Transformers are unique in that they usually add the majority of their saturation below 500 Hz. Core size and material impact this. As a result, loads of third and fifth harmonics are added to the lower mid-range and low end, adding tones that are a third and fifth higher than the fundamentals, not unlike a power chord.

Transformers use primary and secondary coils that are wrapped around a core to induct current. Three common metals used in transformer laminations are cobalt, nickel, and silicon steel.

Induced magnetic fields are used to transfer the signal from one winding of wire to the next. This is called the flux field. The transformer core generates the flux field, and bigger magnets generate larger fields. The amount of flux that a core can tolerate increases with size. Transformer saturation is mostly symmetrical because magnetic flux has the tendency to shift polarity like an alternating cycle (A.C.) saturation, which is the point at which the transformer can no longer tolerate more flux generation.

Transformers produce low levels of distortion when the core is initially magnetized. The reaction of a transformer is also impacted by a number of parasitic factors like capacitance or stray capacitance that exists between the parts of an electronic component or circuit simply because of their proximity to each other. An easy way to think of capacitance is a build up and release of charge over time, like an extremely short lived battery. This is important to a transformer’s program dependent characteristics. Short transient energy approaching or exceeding the threshold of saturation may not distort or compress in the same way that sustained RMS levels would. This means that a transformer may not generate the same level of saturation or harmonic energy on transient-rich signals, especially as frequency increases. Lower frequency waves are longer, slower and require more energy to generate equal loudness for human hearing. This is a key non-linear characteristic of differences between classic console channels like 70’s vintage Neve and API designs, where one or the other may have perceptively different transient punch.

The Tessla Pro mkIII by Variety Of Sound offers a flexible DSP interpretation of subtle transformer saturation that is easy to use and sounds great. A earlier lightweight variation called TesslaSE mkII is also available that I use often.

Magnetic Tape

Tape recording improved dramatically through the twentieth century, but it was always plagued with some fundamental “problems” that many of us love the sound of. Since tape is a magnetic recording medium, it’s similar to transformer-related magnetic flux field saturation. In the 1970s, with most studio signals passing through many stages of transformers and landing on tape as the capture medium, there was certainly a significant accumulation of magnetic saturation character on every record. For example, a typical vintage Neve had ten transformers from the mic or line input to the bus output. Add inserts for outboard compressors and you’ve got even more. The tape machine had input and output transformers and the tape itself was magnetic. Pushing all that signal chain hard could get really juicy.

When tape is pushed to saturation, it produces a punchy low frequency odd-harmonic distortion, blurs transients, and has issues with coercive force (low level distortion). Tape bias (ultrasonic signal above 40k) is utilized to “activate” the magnetic particles and lessen low level distortion with a tradeoff of high frequency response.

In addition to having symmetric saturation, since tape moves across a magnetic head, it exhibits hysteresis. Hysteresis manifests in practice as a lag in the output signal relative to changes in the input signal. To put it simply, transient signals are “smeared” as magnetic devices discharge because they retain the signal that just passed.

The frequency response of the tape machine is greatly influenced by tape speed, head gap size, and width. In the end, even with proper alignment techniques, tape usually exhibits a slight bump in the low frequency range (50-100Hz) and frequently a slightly lumpy lower midrange with some extreme HF roll-off. This is why it’s a popular medium for capturing drum mixes, with many modern producers opting to record drums to tape, or processing their drum mix into a tape machine. It’s bad in a good way.

IC/Solid State Distortion

The characteristics of solid state component distortion vary greatly based on the circuit design, but for the purposes of this discussion, a solid state IC, op amp etc, is typically going to hard clip the signal symmetrically when pushed beyond its design limits and produces odd order harmonics. This is the principle behind a primitive fuzz or distortion circuit. Other factors aside, this saturation will affect the entire frequency spectrum equally, and within the digital domain, the amount and range of this added energy may extend well beyond the Nyquist limit, depending on the sample rate or oversampling.

The Aural Exciter

Given all of the interest today in reproducing the “good crap” that we had to deal with in the analog tape days, the Aphex Aural Exciter wraps all this up nicely in a bow. Major-label record sessions were becoming increasingly convoluted and drawn out by the mid 70s. The tapes would become worn and start shedding during extended recording sessions. Repeated playbacks, overdubs, and bouncing to make room on the most advanced recording medium available at the time deteriorated the sound quality — especially high frequency loss. Engineers could only compensate with EQ, which simply increased audible hiss.

Enter the Aphex Aural Exciter.

An illustration extracted from a 70s era article posted on the Gearspace forum from Recording Engineer magazine (defunct) that is a fascinating read.

The Aural Exciter was the first effects processor that was intentionally designed to specifically “fill in” high frequency signals that had been weakened by tape wear by adding harmonic saturation.

The discovery of the exciter was a happy accident when a hobbyist named Curt Knoppel unwittingly wired one channel of a Heathkit DIY stereo amplifier kit incorrectly. The defective channel was thin, distorted and unappealing, however he noticed that when he blended the good channel with the broken channel using the L/R balance control summed to mono, the harmonics and phase rotation in subtle doses produced a wonderful “excited” effect that added “sheen” and “sizzle”.

Knoppel approached the Aphex company with his discovery. Although the Aphex engineers weren’t quite sure of exactly what the unit was doing, they recognized its potential, and the company realized that this could be marketed as a solution to the complaints by many recording engineers that their final mixes sounded like crap if their big budget artists wanted to do 50 bazillion overdubs on multitrack projects.

Hilariously, since Aphex didn’t initially understand how Knoppel’s device actually worked, they couldn’t file a patent on the circuit. So, they rented the machine for 30 dollars per minute of program fed through it. This costly “fairy dust” was such a sensation among among the record labels that they even credited it in the liner notes of albums and mentioned it in press releases. Early albums to feature the original technology included Linda Ronstadt, Fleetwood Mac, Paul McCartney, James Taylor, Jackson Browne, Led Zeppelin, Dolly Parton, Hank Snow, Grateful Dead, Neil Sedaka, Johnny Mathis, The Four Seasons and many more.

By 1979 the Aphex engineers had reverse-engineered the circuit, identified and replicated the exact properties and filed a patent on the technology. By the 1980s the first units were made available for a mere $3,000 US, which is around $11,500 US in 2024.

The way that the Aphex works is actually relatively simple. Knoppel’s original unit was a tube amplifier. His wiring mistake high-passed the signal and generated even order harmonics on it. When judiciously blended into the original signal, you get that wonderful sound that adds sparkle to the top end.

The psycho-acoustic effect of phase rotation, to quote the aforementioned 1977 article by Howard Cummings in Recording Engineer Magazine was a fascinating discovery at the time,

The random phase characteristics of Aphex related to the source turns out to be the most useful… If main and sub-carriers were in a perfect in-phase relationship, the much weaker Aphex imprints would become absorbed and would their independence and unique identity, and for Aphex to add the dimension by apparently widening the source, it is creating an ambience effect that encompasses the listener instead of emanating directly from the source.

A byproduct of the slight delay and phase rotation of the saturated signal was “widening” of the stereo image.

Thus, we can attribute the ubiquitous “sizzle” of most major label records from the late 70s and 80s to the introduction of this magical/cursed device. As Aphex lowered the price point and competitors figured out that high-passed blended saturation could be achieved without explicitly infringing on the patent, the excitement surrounding exciters diminished, and producers began using it more selectively on certain sources – particularly vocals. Today, exciters are garden variety tools that are often used to restore dull signals, or to compensate for dynamic de-essing on troublesome sources.

A scanned archive of the now defunct Recording Engineer magazine is available courtesy of David Gleason’s website Worldradiohistory.com.

Another fantastic product by Variety of Sound (Herbert Goldberg) that takes the principles of harmonic excitement into the realm of multiband. It can achieve the Aphex exciter effect and much more. I’ve been reading Herbert’s papers for the past twenty years and feel like I practically know the guy.

Stateful Saturation

One of the primary challenges of DSP engineers working to achieve convincing emulation of analog saturation is that simple waveshaping is not enough. As we’ve observed, both transformers and tape exhibit attributes of “memory” or statefulness that impact the sound. Few vintage box tones are a single component, and almost all have some stateful qualities given that the usually have one or more transformers. Herbert Goldberg was probably not the only DSP engineer to observe this, but he was one of the few independent software developers not working for a proprietary manufacturer under NDA who openly wrote about his research and development in his spare time and released a series of VST plugins that were extremely well received by the audio engineering community, even achieving an award for his Ferric tape emulation plugin in 2009.

Herbert has maintained a series of “TechTalk” blog posts over the past many years including interviews with many legendary audio engineers and DSP developers.

I was pleased to see Herbert return to his passion projects following the pandemic of 2020. Just days after this article was initially posted, I noticed that he released a fabulous new plugin he calls the FeenstaubTX which is a unique offering for use on full range material like summing bus and mastering duties!

Below is his description of the latest work.

“FeenstaubTX is an exciter and expander effect based on audio transient processing. It enhances and refines signals in a mid/side configuration with saturation and expansion effects, allowing for unique signal coloration, transient management and glue effects, as well as sublime control over the stereo image.

Its designed to work on complex full program stereo material for very gentle balancing audio in three dimensions: Timbre, dynamic structure, and stereo field. While all areas are touched only slightly, it creates a musical consistency that brings some analog feel back into the digital realm, and the signal appears subtly colored and sprinkled with a little magic.

Practical Application

Now, let’s tie this back into the practical application of saturation in a mix. Saturation waveshaping is probably the least understood aspect of ITB mixing. I don’t imagine any of us have personally coded and tested every processor we use in a mix; we’re just playing with products to see how they sound. If we like it, we use it. After all, that’s part of the fun.

Saturation potentially benefits the signal chain of any channel in a variety of ways:

  • Harmonic excitement and richness
  • Added density and perceived loudness without vanilla compression that could otherwise squeeze out the lively natural microdynamics.
  • Better control of compression downline from saturation

While their are no rules, the best application of saturation is often subtle, almost inaudible on many channels or subgroup busses in a mix that have a cumulative impact on the overall sound.

Hands-On DIY Examples

We don’t necessarily need a complex set of tools to achieve a wide range of saturation styles, but slapping the same broadband saturation plugin on everything isn’t necessarily the solution either. Different sources benefit from different options. There aren’t that many dimensions of basic saturation waveshaping:

  1. Hard clipping (odd harmonics)
  2. Soft clipping (even harmonics)
  3. Frequency specific saturation
  4. Strength or amount of harmonic energy, directly proportional to threshold of clipping.
  5. Capacitance – or the tolerance of transients before saturation
  6. Hysteresis (if you desire)

For the sake of demonstration, let’s utilize three free basic DSP effects to approximate a range of saturation schemes.

  1. FreeClip2 by Venn Audio – which features a variety of waveshapeing and ample oversampling. If you don’t have this, look for a basic clipper, saturator or distortion plugin. It’s critical that it supports oversampling.
  2. Your DAW’s flexible stock compressor
  3. Your DAW’s flexible stock graphical EQ

The setup is simple, create a send from your source track to a bus/group channel or effects channel where we’ll essentially create a parallel effect.

On the channel you’ve created, insert plugins in the following order:

  1. Equalizer 1
  2. Compressor
  3. Clipper
  4. Equalizer 2

Example 1: Transformer style Saturation

First, let’s experiment with building a simple transformer saturator effects chain, which is perhaps the most interesting and useful example.

Step 1. On your equalizer apply a medium slope low pass filter down to around 500hz or lower. The result will sound dark and muddy. Steeper slopes will cause more dramatic phase rotation, delay and result in cancellation, so for more precise control consider a linear phase EQ such as the free TDR Nova.

Step 2. Use your compressor with a 6:1 or 8:1 setting with fast attack and slow to medium release to attenuate only the transient peaks of the signal. Don’t over-compress the signal. We’ll do this in order to reduce clipping of the peaks into the clipper and approximate the program dependent behavior of a transformer’s capacitance.

Step 3. Set the clipper to a hard clip option and begin reducing the ceiling/threshold so that saturation starts to become quite audible. A spectrum analyzer is actually quite helpful for this, but you should hear it clearly on lower frequencies. The amount of saturation is up to you. Aggressive settings may sound extreme while you’re setting this up, but will produce more harmonic content which may be useful for blending.

Step 4. Using your second equalizer after the clipper, you may want to use a gentle shelving filter to remove some of the lower sub-harmonic frequencies that are already present in the source. This is optional to taste. See the advanced methods below to solve this problem.

Step 5. Using your mixer faders, blend in various amounts of the saturated signal in with the dry original signal. Adjust the settings of your effected signal to taste.

Step 6. Because you’ve utilized some fairly heavy equalization on the effected channel, phase rotation has occurred that will result in phase cancellation. Flip the polarity of the effected channel to see if you prefer the sound and watch out for loss of low frequency information.

Like all parallel effect strategies, the beauty is that the original signal is still present with clarity and transient detail, but is now reinforced with the added girth and density of the effected signal. Since the saturated signal will increase the overall volume, reduce both until you’re satisfied.

Example 2: Aural Exciter/Vocal Air

This is a similar but opposite saturation strategy to add air and sizzle to a signal. This works particularly well on signals like vocals or acoustic guitars that have substantial treble information.

Step 1. On your equalizer apply a medium to steeper slope high pass filter up to around 5-8khz for a vocal air type effect. The result will sound very thin and airy. Steeper slopes will cause more dramatic phase cancellation, for more precise control, consider a linear phase EQ. The Aphex used a gentle slope starting at 500hz, so you can also play around and see what sounds good.

Step 2. Use your compressor with a 6:1 or 8:1 setting with fast attack and fast release to attenuate only the transient peaks of the signal. Don’t over-compress the signal. We’ll do this in order to tame sibilants and transients like pick noise on an acoustic guitar.

Step 3. Set the clipper to a soft clip option and begin reducing the ceiling/threshold so that saturation starts to become quite audible. The amount of saturation is up to you. Aggressive settings will produce more harmonic content which may be useful for blending.

Step 4. Using your mixer faders, blend in various amounts of the saturated signal in with the dry original signal. Adjust the settings of your effected signal to taste.

Step 6. Because you’ve utilized some fairly heavy equalization on the effect channel, phase rotation may result in phase cancellation. While this is less common for high frequencies, flip the polarity of the effected channel to see if you prefer the sound and watch out for loss of low frequency information. Like the Aphex, the phase rotation may actually sound good and generate some stereo widening.

Example 3: Tube Vibe

This is by far the easiest case involving no EQ unless you want to tune the target frequencies. For example, you may want to add clarity to a bass guitar or vocal by using a steep bell curve centered around 700-1k. But first, just try to bypass the EQ.

Step 1. Light peak compression will even out the peaks similar to our transformer, but tubes don’t have quite the same capacitance behavior. Aggressive compression will allow you to maximize the signal so that harmonics stand out and can be blended in.

Step 2. Set the clipper to a soft clip option and begin reducing the ceiling/threshold so that saturation starts to become quite audible. The amount of saturation is up to you. Aggressive settings will produce more harmonic content which may be useful for blending.

Step 3. Here you could optionally add a fairly aggressive compressor to squash the transients and really bring the harmonics forward. Try it to see how it sounds on your source.

Step 4. Blend the effected signal with your dry signal to taste and keep in mind that the more compression you’ve added the more overall density will be present. Since tubes saturate based on input, you might want to disable the first compressor and rely on the second to have more control over the dynamic saturation.

Advanced Methods:

The downside of all of the above approaches is we’re feeding a substantial amount of the original signal back into the blend, which increases the overall volume of the source, and if using compression, this happens to double as parallel compression, which is a common strategy and potentially useful even without saturation. However, there’s a technique to work around this and strictly control only the harmonics.

  1. Bypass all effects on your effected channel, and create a second bus or effect channel send from the source track/channel. You’re now working with three channels that are all essentially duplicates.
  2. Send the output of both effected channels but NOT the original source to a new summing bus / group. Ensure that the levels of both effected channels being summed are equivalent.
  3. Flip the polarity of the second effected channel. This will null the summed signals completely. If both of the effected channels are the exact same gain, no sound will pass through the summed bus group channel.
  4. On the first effected channel routing into the new bus group, engage and apply only the clipper and start reducing the ceiling distortion/clipping until you start to hear the harmonics emerge from the summed bus/group. Depending on the accuracy of your DAW’s delay compensation, you may want to add the clipper with the same oversampling settings to the dry effect channel as well , but with no clipping engaged to ensure that both effected channels remain phase aligned.
  5. Using the new summed bus, blend in the effect with the original.

What we’ve done in this case is nulled the original source signal so that only the generated harmonics remain in the sum as the difference between the inverted waveforms. This is quite effective if you want to strictly control only the harmonic content vs. your original signal.

Saturation In Effects Chains

While it was an in-built feature of many classic analog circuits, digital audio workstations lack any inherent saturation character by default (however ProTools and Studio One introduced solutions for this with HEAT and Console Shaper), and that is what often leads to the conclusion that digital recording ITB with no effects may sound “sterile” vs. their analog counterparts. Clipping and saturation waveshaping added to your signal chain has many benefits:

  • Greater density and excitement thanks to enriched harmonic content
  • With transformer type distortion, an improved sense of rich density may be achieved in the critical low to low midrange areas where the “body” and “muscle” of many sources reside.
  • Transient attenuation that helps reduce the workload of downline compressors and limiters
  • A natural and highly musical type of “compression” that has fewer artifacts than the typical VCA, FET or Optical compressor approaches where a mechanistic attack and release control is used.
  • On low frequency instruments, greater clarity is achieved on lower quality playback systems by adding harmonics above the fundamental frequency. This is the principle behind many bass enhancement plugins.
  • On higher frequency instruments, a sense of air, excitement and polish without harshness when applied judiciously.
  • Digital audio may contain very small transient spikes that are inaudible to human hearing (our ears are not fast enough to perceive them), but impact everything downline. Shaving these “useless spikes” down with a clipper is a common practice in mastering to reduce compressor or limiter pumping with virtually no audible side effects, and even when side effects occur (in cases of longer loud transients), the results are often exciting and musical.

Where you place saturation in your chain will depend on taste, but it’s not uncommon to place subtle amounts of saturation on every subgroup buss or even every channel.