No matter what genre of music you listen to that incorporates synthesizers, chances are high that the function of the machine that gives its sound character and makes it engaging is the filter. Filters are the most integral part of synthesizers, and yet they're not often completely understood. Synth manufacturers dating back to the 60s have worked tirelessly to create their own distinct filters whose sounds stand out from the rest of the market.
With esoteric terminology like "pole" and "tracking", combined with an endless variety of types and arrangements, it's not hard to understand why many users have learned to get what they want from knob-tweaking without really understanding what's going on under the hood. The recent explosion in popularity of modular systems has only made things more complicated, with many manufacturers advertising filter types in promotional material.
That in mind, we'd like to guide you through the world of synthesizer filters and get you up to speed, so you can know your transistors from your diodes and your SEMs from your CEM3320s.
For readers unfamiliar with the general function of filters, scroll down to find our guide to the basic types.
The synthesizer's grand introduction to the general public was arguably the 1968 synthesizer-based album Switched-On Bach, which saw Wendy Carlos produce an album of Bach pieces on a Moog synthesizer.
As the catalyst of the synth craze, Robert Moog soon unleashed his Minimoog in 1970, a non-modular portable synthesizer with an exceptional filter design steps ahead of other manufacturers. Several took note and wanted to use similar filters in their own instruments, which only led to a domino effect of filter designs, a few of which have gone on to achieve legendary status. Let's now go through a few.
Ladder Filters (Moog vs. 303)
Originally developed in the mid-60s for their modular synthesizers, the ladder filter—named for the shape of the circuit—has become synonymous with the Moog sound.
The Moog Ladder was the first voltage-controlled low-pass filter of its kind, borrowing its structure from George Campbell's work with telephone systems at AT&T. With a 24dB/octave slope and an adjustable resonant peak at the cutoff point, it offered its own pleasing overdrive tones when attenuating the signal, contributing an unmistakable "fat" character to several basslines and lead sounds. Moog's patent for the filter was granted in 1969 shortly after Switched-On Bach's release, which protected his design from being lifted for every new company's attempt at a synth.
Another popular ladder filter design is the diode ladder, used famously in the Roland TB-303. Like the Moog Ladder, Roland's latter filter is a 24dB/octave low-pass filter. However, other differences in how the circuit is made—namely, swapping Moog's transistor-based design for a diode-based circuit—make this a different beast entirely, a perfect filter for the type of squelchy basslines so often played on the TB-303.
Tim Stinchcombe gives more insight in his expert dissection of diode ladder filters: "The move from transistors to diodes has implications in the way the circuit operates, and in one sense this leads to the loss of a certain amount of 'elegance' too: the resistor chain used to bias the transistors in the Moog ladder means that the voltages at each filter section are separated, which effectively means that the sections are buffered from each other; this 'isolation' is simply not present in the diode ladder."
That loss of isolation makes for a different pole placement and makes the poles move differently than Moog's when you tweak the diode ladder's resonance. While some commenters (and apparently even Roland for a time) have said the TB-303's filter is a three-pole, 18dB/octave filter—due to the filter's unique frequency response—Roland now says it is in fact a four-pole, 24dB/octave filter.
Steiner-Parker Filters
The 12dB/octave Steiner-Parker filter was originally designed for the Steiner-Parker Synthacon, a rare American monosynth from the 70s—the sound and design differs from the 24dB/octave filters favored by many Moog-inspired bass synths.
Taking a different track with its enormously popular bass synth series, Arturia opted to use a variation for its Brute line, which includes the original MiniBrute, MiniBrute 2, and MiniBrute 2S.
Because of the more gradual 12-decibel slope, the filter will attenuate less aggressively, but you can still get a range of filtered tones, including aggressive ones, in part because of the built-in Brute Factor control, which pipes the output of the synth's amplifier back into the input of the filter. Between that mode and the resonant filter's capacity for self-resonance, you need not fear that the filter will sound too tame.
ARP Filters
As Robert Moog's patent for the Moog Ladder Filter was approved in 1969, Alan R. Pearlman was busy developing an innovative modular synthesizer design that could be programmed with or without patch cables. His engineer Dennis Colin employed a pleasant-sounding 24dB/octave ladder filter that was suspiciously familiar, dubbed the 4012.
When the ARP 2600 was introduced in 1970, it took the world by storm—with its portable and powerful design, it quickly evolved a favorite of educational institutions and rock stars alike. However, when a Moog employee was tasked with dissecting the circuit to see what was inside, they found too striking a resemblance to the one they previously patented. After Moog paid a visit to Pearlman, ARP decided to create a new filter for the 2600. The 4072 was introduced in the upgraded synth, by now complete with a black-and-orange interface.
SEM Filters
During a stint as an ARP dealer, Tom Oberheim came up with an idea for a digital sequencer for synthesizers: at this point, the issue was that most synth players could not perform and simultaneously operate their sequencers. This gave him the idea for a standalone synthesizer add-on that provided extra functionality and the ability for the sequencer to play a part while the main synth was free for the user to play.
This is the origin of the Synthesizer Expander Module, known by the acronym SEM. In addition to its novelty as an add-on and its importance as providing a portion of one of the first fully-articulated polyphonic synthesizers, the SEM became iconic due to its own sound, courtesy of a 12/dB per octave filter that was unique for its time.
The SEM filter was designed by none other than Dennis Colin of ARP, but unlike the Moog-like filters that he previously pulled from, the 12/dB Multi-mode filter of the SEM had a different and desirable character that made it the ideal complement to the synths it was designed to expand. While it couldn't self-resonate, it did have a smooth sound and was multi-mode—meaning low-pass, high-pass, band-pass, and even a notch mode which allows all frequencies except a very small band (or notch) to pass.
Most recently, the SEM filter was used in Arturia's MicroFreak, and its musical response has a wonderful calming effect on the often harsh digital sounds the machine is capable of.
Korg Filters (35, OTA)
Another famous filter, and one that happens to be famously aggressive, is that of the Korg MS-20, which has both a resonant 6dB/octave high-pass and resonant 12dB/octave low-pass filter in series, allowing the creation of band pass filtering when used in tandem. Both filters were voltage-controllable and had resonance functionality, delivering a noisy, raw, expressive and almost untamable potential for sound design possibilities.
The original MS-20 actually had two different filter section designs. The first was a compact, chip-based filter Korg made itself called the Korg 35. Based on a classic Sallen-Key filter design, the Korg 35 chip at the center of the filter core, in combination with other components around it, helped define what players loved about the MS-20.
However, later models of the synth swapped the Korg 35 for OTA chips (operational transconductance amplifiers). While OTA chips can be found and are prized in the filters of the Roland Jupiter-8, Jupiter-4, and other synths of the time, people generally prefer the Korg 35–based filters in the older units. (You can find a comprehensive breakdown of Roland filter types here). In recent years, Korg has once again used Korg 35 filters in the MS-20 Mini and the even more diminutive Monotron.
OTAs also make an appearance in the famous Doug Curtis–designed span class="weight-bold">CEM3320, a 24 dB/octave multimode filter chip that was used in the OB-8 and OB-Xa, the Sequential Circuits Prophet-5 Rev 3 and Prophet-10, the Fairlight CMI, LinnDrum, and more synths and drum machines of the '80s.
Other Filters
While all the previous filters mentioned here are used in subtractive, East Coast–style synthesis, the low-pass gate (LPG)— sometimes called the low-pass gate amplifier—originally designed by Buchla, is an appropriately unique and musical filter from the West Coast synthesis pioneers. Its amplitude rises and falls in coordination with its frequency response, and, along with a gradual 6dB/octave slope, this can create natural-sounding filter sweeps. Found in Buchla's Music Easel and Korg's Volca Modular, LPGs are increasingly common in many companies' Eurorack module offerings.
Lastly, we'd like to mention the Wasp filter. The Electronic Dream Plant Wasp was an affordable and extremely popular synth in England in the late 1970s, and its 12dB/octave multimode design remains popular to this day, thanks to its inclusion in the Novation Bass Station and Bass Station 2 and Eurorack Doepfer clone A-124 VCF5.
Still asking yourself what filters actually do? Let's go back to the basics.
Additive Vs. Subtractive Synthesis
As electronic instruments first started to appear, what would come to be known as synthesis was achieved through a complex and expensive process of mixing certain harmonics at certain amplitudes—in other words, stacking sine waves of different pitches and volumes to build sounds piece-by-piece. This is now referred to as additive synthesis.
The engineers at AT&T came up with the electronic audio filter while searching for efficient ways to carry multiple telephone conversations by wire—they soon discovered that the filter was capable of amplifying or attenuating frequencies in an audio signal. When oscillators had developed enough that they could produce harmonic-rich waveforms like the sawtooth wave, they found that filters could shape the harmonic content of said waves into a wide variety of timbres. The process of elimination or removal they happened upon is now known as subtractive synthesis.
So, what is a filter? Broadly speaking in the context of subtractive synthesis, you can think of a filter as a general equalizer, able to remove—or more appropriately, filter out—certain frequencies from a synth's oscillator as dialed in by the user. The filter adds color and shape to the raw sound before passing through an envelope to give the amplitude some shape.
Basic Types Of Filters
The most common kind of filter is the low-pass filter, which allows frequencies below a certain point to pass. The point at which the filter starts working is the cutoff point, which can be set by turning your synth's cutoff knob. The opposite of the low-pass is the high-pass filter, which filters out low frequencies and allows high frequencies to pass through. Less common is the band-pass filter, which attenuates both low and high frequencies, allowing a range of frequencies around the cutoff point to pass. There are other types of filters as well, but these are the most common you'll see.
What about that knob that usually sits to the right of the cutoff knob? That sets the amount of resonance—sometimes referred to as Emphasis (on Moogs) or Q—and it's responsible for the squelchy bite that we associate with synthesizers.
Resonance creates feedback at the cutoff point, emphasizing the frequencies in a way that's pleasing to the ear. A filter can be said to self-oscillate if it's able to create a ringing sound when pushed far enough. This ringing can often be played as if it were a sine oscillator, hence the name. Not every filter on a synthesizer is resonant; many of Roland's analog synths have a resonant low-pass filter and non-resonant (or fixed) high-pass filter. This is useful for general tone-shaping, such as removing unnecessary bass from a pad or lead sound.
If a synthesizer has more than one filter, say a high-pass and low-pass, they will often be organized in series. That is, the sound will pass through one and then the other. Some higher-end synthesizers will offer parallel routing, which means the signal can be split and run through each filter independently. Another common synthesizer style is the so-called multimode filter, with a switch to select from a few different types of filter, usually low-, high-, and band-pass types.
Slopes and Poles
So far so good, right? You tweak the cutoff and it attenuates frequencies below, above, or around a certain point, and resonance emphasizes that point—but what about slopes and poles? What are they all about?
If you look back up at the low-pass and high-pass filter illustrations above, you'll see that there's a gradual curve around the cutoff frequency. Along this curve, on one side there's an area with no signal at all, and on another, there's an area where the signal is at its fullest. This curve is the slope of the filter, or how steeply or sharply it attenuates the frequencies past the cutoff point.
Most filters used in synthesizers are either a 12-decibel-per-octave filter or a 24-decibel-per-octave filter. As an example for what this means in practice: Say you have a 12dB/octave low-pass filter, and you've set the cutoff frequency around the equivalent of an A note. Between that A and the next A an octave above it, the filter will have cut 12 decibels of gain.
Thanks to this relatively tame curve, 12dB/octave filters generally have a rounder and gentler sound. Depending on the overall volume and harmonic richness of the passage or chord you're playing, you may still hear frequencies well past your cutoff frequency, because, while attenuated at a rate of 12 decibels per octave, the notes haven't been attenuated enough to be silent.
A 24dB/octave filter has a steeper cutoff point. To take that same example of setting a cutoff around an A note on a low-pass filter, the filter will have cut 24 decibels of gain between that A and the next A an octave up. This steeper climb or dip makes a more aggressive sound, and means that a greater amount of harmonic content not allowed to "pass" will be filtered out completely.
Poles refer to the amount of attenuation, with each pole responsible for six decibels of cut, and as a term, "poles" can and often is used as a substitute for "slope." A 12dB/octave filter is a two-pole filter, as it cuts 12 decibels per octave, and a 24dB/octave filter is a four-pole filter, as it cuts 24 decibels per octave.
However, as Gordon Reid explains in a 1999 Sound On Sound article (alongside the work of many other filter-heads), the exact position and behavior of poles will change for any number of reasons. This will lead to unique frequency responses, roll-offs, and other characteristics between different filters. While our illustrations above show clean, regular curves, in reality, a filter's poles will be irregular, meaning some parts of the curve are attenuated at different rates than the others.
That said, saying a filter is two-pole, 12dB/octave or four-pole, 24dB/octave will give you some basic idea of what you can expect from the filter. And while most synths will have filters that fit one of these two categories, some, like the Korg Minilogue, allow you to select between two-pole or four-pole settings. Some synths even use three- or even six-pole filters, though these are rather rare.
Envelope, Tracking, and More
There's more to a filter section than just cutoff and resonance, as anyone who has twiddled knobs will know. As with general amplitude, the shape of the filter can be controlled with envelope amount. The envelope amount knob (called "EG Int" on many Korg synths) controls how much of the envelope will be applied to the filter. A synth may or may not have a dedicated envelope section for the filter.
Keyboard tracking (sometimes called "keyboard follow") controls how much filter is applied to the notes as arranged across the keyboard, which is helpful for the situations in which you don't want your filter to shape, say, a low chord and a higher melody line in the same way. Keyboard tracking can also keep the resonance in tune to the key being played. If your filter self-resonates, you may be able to use keyboard tracking to effectively play the resonance, as on a Roland Juno-106.
Modulation (sometimes called "MG" on Korg synths) can add movement to the filter section, usually with an LFO (low-frequency oscillator). Using the LFO as an inaudible control signal, you can set it to raise and lower the filter's cutoff frequency at a certain rate, bringing variation to the way the filter clamps down or opens around your audible signal.
Lastly, don't underestimate the effect that drive can have on a sound as it passes through a filter circuit. Drive refers to a sound being pushed too hard, resulting in pleasing distortion and clipping. The Moog Minimoog Model-D is particularly well-known for this.
While you'll encounter many more filter types as you explore the wide world of synthesis, this should give you a nice foundation to understanding the many differences and uses of various kinds. Have a favorite you don't see here? Let us know in the comments.