Intermodulation Distortion in The Mix

In our last article, we discussed the art of DSP saturation, but with saturation comes side effects, one of the most important being Intermodulation Distortion (IMD). This is a type of audio distortion that occurs when two or more signals with different frequencies in a waveform encounter a non-linearearity like an overdriven amplifier, speaker, compressor, or similar. IMD are new signal artifacts at frequencies that are not part of the harmonic series of the original signal’s fundamentals. As we’ll see, where you place effects like saturation, distortion and even aggressive compression in your signal chain can have a major impact on the quality and clarity of your mix.

Understanding IMD

Monophonic content like a single note from a horn, guitar, bass, piano or voice can be a relatively straightforward waveform. For example, a sine wave from a test tone oscillator is the purest form.

A simple 440hz sine wave
440hz sine wave on a frequency spectrum analyzer.

Adding any type of non-linear process, particularly flavors of symmetric or asymmetric distortion to monophonic signals will result in mathematically predictable resonant harmonics above the original fundamental.

Hard clipping distortion applied
Clipping results in predictable harmonic enrichment.

Polyphonic content (like chords) or any signal with natural harmonics, wide spectrum energy, overtones etc. are much more complex waveforms. Even something as basic as two pure tones of equal energy sums to a noticeably distinct pattern:

The waveform of a 440hz and 550hz or a perfect third.
440hz and 550hz or a perfect third represented on a spectrum analyzer
Our two test tones are pleasant.

If you play guitar and enjoy using heavy distortion or overdrive sounds, you probably have noticed that simple power chords sound great, but more complex voicings like major 7ths, minor 9ths and the like don’t translate well or seem to get “lost in the dirt”. This is the unpleasant side effect of IMD.

One of my favorite guitarists, Andy Summers of The Police used advanced jazz chord voicings in a rock context, and leveraged clean tones with modulation and delay instead of distortion to preserve the clarity of the notes and avoid IMD. Photo credit: Wikimedia Commons.

Adding any type of non-linear process, particularly flavors of symmetric or asymmetric distortion to polyphonic signals, or any signal (such as drums) with wide spectrum tonal energy, will result in IMD artifacts, which are both consonant and dissonant harmonics above AND below the fundamentals. These frequencies are the sum and difference of the fundamental frequencies as a byproduct of non-linear wave shaping of a more complex waveform.

The sum waveform of 440hz and 550hz hard clipped
On a spectrum analyzer we see complex harmonic content above and below the fundamental frequencies, including dissonant frequencies.
Distortion added to the summed tones is dissonant and unpleasant.

Pro Solutions

As a point of comparison, let’s instead add the exact same distortion to each individual signal before summing them together.

Our two tones are now displaying consonant and pleasant saturation.
The sound is rich and free of intermodulation distortion.

A more radical example is what happens when you add distortion to a more complex chord.

Four note chord using a sine wave from a synthesizer with no saturation.
Sounds like a simple clean synth.
Here’s a frequency analysis of distortion on the summed signal. That’s a lot of junk, including a surprising amount of dissonant energy and energy in the subharmonic region. The colloquial term is often “mud”.
Since our synth in this example also offers a square wave oscillator, which is very similar to the dense harmonic structure of a clipped sine wave, let’s hear the difference when each of the frequencies is individually generated already “clipped”.
Note the abundance of harmonic energy from the square waves, but no obvious intermodulation distortion. This is why square wave synths sound rich without egregious IMD.
Square wave generation is very rich, but there’s no dissonant or low frequency junk.

Practical Implications

Intermodulation distortion is common in both the analog and digital domain and is not necessarily a problem if the frequency build up is well below masking levels, but WHERE you choose to apply saturation or distortion in the signal chain of your mix may mean the difference between a great sounding mix, and some odd unpleasant dissonant tones or “garbage” that you can’t quite identify the source of.

The diagram of a typical channel -> subgroup buss -> master buss above indicates where saturation or distortion risks unwanted artifacts (red), optional in subtle amounts (yellow) or preferred (green). Note that the monophonic signals or percussive signals like drums are usually more forgiving of saturation to taste.

Key Points:

  • Avoid Group Channels – With buss/group channel submixing (sometimes known as top-down mixing) where you have multiple instruments of different frequencies summed, heavy saturation may result in unpleasant dissonant harmonics that build up into your final print. For example, saturating a backing vocal submix buss with multipart harmonies, or multiple polyphonic instruments may seem like a good idea, but it would be better to add the same saturation to each channel prior to downline summing. Group summing channels are ideal for equalization (keeping the phase rotation of the sources aligned), compression (glue) and limiting.
  • Monophonic Good – Single note (solo) signals take both saturation and distortion well.
  • Polyphonic Risky – While we’re accustomed intermodulation distortion on hard rock guitars, we can see where problems could be non-obvious on other sources. Distortion and saturation on any polyphonic source, like acoustic guitars, pianos, etc may lead to undesirable consequences.
  • Saturate close to the sources – Adding saturation effects like tape, tube, or clippers in small doses on each source is preferable, especially if they are monophonic. This is why analog mixing consoles sound good – there are many small stages of gentle saturation on each channel. Slapping a heavy tape or console effect on your master channel to compensate for a sterile mix doesn’t sound like something more sophisticated like each source with a unique saturation imprint, or solutions like ProTools HEAT and Studio One’s Console Shaper.
  • EQ before and after – Sometimes distortion on polyphonic signals, warts and all is the sound you’re looking for. Equalization before and after saturation may help clean up the sound. For example, high passing after saturation or reducing certain frequencies before distortion can make a big difference.
  • Try compression instead – While it’s not uncommon to apply saturation or tape emulation to drum mixes, in some cases, compression or multiband compression may get a very similar effect without the ugly dissonant byproducts.
  • Avoid it in Mastering – The closer you get to your final sum in your signal chain, the complexity of your waveform increases and the more non-linear processes may cause unwanted byproducts. Adding broadband saturation on your master channel or in mastering, even in small doses, may lead to subtle dissonant garbage that’s not obvious, but reduces clarity and definition. Workarounds include multiband enhancers or saturation tools that allow you to apply saturation to narrower bands of frequencies. I’ve experimented with the enhancer saturation tool in Ozone for years and rarely find it improves the sonics except in very specific cases, and most often only on low frequencies. I find that when I stick to EQ, compression and limiting on mastering I get better, more transparent results.
The multiband exciter featured in Ozone offers greater control and blending in the mastering phase, but I rarely prefer it to adding saturation upstream in the mix.