The Mega Percussive Synthesizer Project - Theory
An Introduction Percussive Voice Synthesis
Analog synthesis has gone through its own unique evolution. Its primordial roots extend back as far as the late eighteenth century when people were beginning to get a grip on why feathers stuck to a glass tube after it had been rubbed on a piece of fabric. Things gathered steam in the late nineteenth and early twentieth century as vacuum tube technology was born. Through the thirties, forties and fifties, analog synthesis became something bright, shiny and new as development overtook development. New development reached its peak with the popular and commercial offerings of early synthesizers designed by the likes of Buchla, Moog and Bode in the early to late sixties.
During this period all of the basic ingredients for analog synthesis were invented. That's not to say things stopped there; it just means we have been handed a quite complete set of building blocks for implementing analog synthesis and, for anything truly different to come about, we need simply to arrange these elements in new and exciting ways.
This applies to all electronic analog synthesis equipment, including electronic analog drum instruments as well. After all, an analog drum synthesizer is really just a task-specific analog synthesizer. Even the simplest analog drum synth typically contains some form of oscillator and some variation on a method to control the amplitude of the oscillator over time to simulate both the impact and physical resonance of the particular percussive instrument being synthesized.
The simplest form of drum synthesizer is a filter at near self-resonance that is unbalanced briefly by an electrical pulse, causing it to "ring" for a short period of time before the oscillations are fully damped. Other designs use more sophisticated methods that employ some sort of noise, a function generator, a VCA, pulse processing, envelope generators, and oftentimes simple filtering to form the drum voice the design will generate. A percussion synthesizer must also have some sort of input telling it when it has been "struck". Simply put, a typical percussive voice will simulate the impact of one object striking another, and it will simulate the harmonic content of the struck object's physical vibration as a result of the impact. The analog circuitry for accomplishing these tasks have long since been in existence; in order to develop a better analog drum synthesizer, all that remains is refinement of the process.
For over thirty years Thomas Henry has designed a lot of analog synth modules and effects, the diversity of which reflects his many interests in all aspects of analog synthesis. Of course, among these interests, is the analog drum voice. In fact, he's written a thorough book on creating analog drum voices, from very simple "Twin Tee" designs through very sophisticated voltage controlled models. This book, "The Electronic Drum Cookbook", which I highly recommend, is available through Magic Smoke Electronics.
All of this blather leads to the subject of this page, a little device dubbed the "Mega Percussive Synthesizer" or "MPS" for short. The MPS is one such instrument I alluded in my preceding notes in that the elements of it are all part of the established pantheon of analog synthesizer ingredients handed down to us from the pioneers who preceded us. These carefully selected ingredients are combined and uniquely implemented in the MPS to create, literally, a monster of a percussion synthesizer. And - being designed by Thomas Henry - the elements of the MPS are derived ecomomically and with great thought to topology. This translates into a wonderfully versatile and unique percussion voice module without presenting a prohibitively complex build process for the average synth DIY enthusiast.
The MPS is truly an inspired design. One very important element of the MPS, the impact generator, was derived directly from a modular percussion patch by one of the great synthesizer masters. In fact, the MPS contains three seperate tunable oscillators. In addition to the oscillators, the MPS utilizes a ring modulator, it has a noise generator, a voltage controlled resonant filter with two selectable responses (low pass or band pass), three VCAs, and a mixer section for mixing all of the elements together to fine tune a very realistic percussive sound.
The MPS will deliver anything from a convincing snare to toms, to bass, to a sizzling high hat, to a glorious cowbell. When Christopher Walken demands "more cowbell", the MPS can deliver. Not only can the MPS deliver standard percussive sounds, it can easily be tuned to the "out of this world" percussive sounds. It can easily supply voices ranging from the electronic disco drums used in modern day hospitals to empty stomachs of their contents to the sound of an alien craft taking off with each drum beat. In fact, at the flick of a switch, the MPS will transform from a percussive synthesizer to a full-on sustained..sound..generator capable of fully rendering the Neptunian Planetary Anthem in full four part Neptunian harmony.
What does it take to build this thing? Nothing exotic or hard-to-get. In fact, it requires only a total of six ICs and a handful of other parts. And pots...lots of pots, because the MPS provides *lots* of control of the voices it will synthesize. Oh, yes, there is some amount of switches. And connectors. One thing you will save on is trimpots, because there is nary a one in the design.
The MPS is an incredibly versatile device - suffice it to say it would take an extremely large body of work to faithfully illustrate the range of percussive instruments (and other sounds) it can produce. The best I have to offer at this writing is a sort of composite of out-takes from various examples and compositions I've recorded. The sample goes as follows:
00:00 - The sample starts out with the first recording I made of the MPS. The MPS is being triggered by a square wave LFO and the output is run through a delay. As this portion progresses, I tweak a number of controls slightly for variation. It's a good example of the "oomph" factor of the impact circuit.
00:54 - I fade in a portion of a sample I made using the MPS in locked mode. The Shell VCO is controlled by the Klee sequencer. It starts out with the MPS noise through its filter and I use the MPS mixer to mix in the ring modulated shell output.
1:35 - This is the beginning portion of a Klee composition that used the MPS as a cowbell. Towards the end of this section I used the mixer control to bring in the noise as a sort of additional percussive voice.
2:14 - Another portion of a Klee composition. Here, the MPS is providing the sort of grainy cymbally/high hat sound.
2:30 - Yet another Klee composition fragment - here the MPS is acting as a sort of cross between a low floor tom and a bouncing basketball.
3:07 - Once again, a portion of a Klee generated sample. The MPS is providing the menacing high tom - the shell is modulated by the impact oscillator.
3:33 - In this frenetic Klee fragment, the MPS is controlled by a voltage output of the Klee to provide a range of high tom pitches. Bass drum is a UD-1.
4:00 - Hmmm....more Klee. I suppose the MPS sound could be best described as some variation of a timbale perhaps?
4:27 to end - This is a sample taken by tweaking the controls while the MPS is being triggered and all of the envelope generator decays are set to max. It's sort of like being in locked mode, only the envelope generators are still triggered (unlike lock mode, the sound would eventually fade away once the triggers stopped).
Before getting into the nitty-gritty of the MPS, a bit of review of exactly which elements make a percussive voice a percussive voice is in order. This is only a light treatment placed here in order to give the sections of the MPS context. For a more in-depth treatment, I highly recommend Thomas' book "The Electronic Drum Cookbook", available at Magic Smoke Electronics.
Typically percussive sounds are very brief, lasting anywhere from a fraction of a second to a few seconds long. But, in that brief amount of time, a lot of information is transmitted to the brain. This information tells your brain if you are hearing a clave, a cowbell, a snare drum or a bass drum, for example. This information is remarkably detailed - it tells the brain more-or-less if the struck object is solid or hollow, or if it is large or small. It gives information that allows you to perceive what the object may be made of, be it wood, glass, skin or metal, and it imparts at the same time some modicum of intuition as to what object was used to strike the percussive instrument - whether it was a stick, mallet, a fist or the palm of the hand. This information allows you to discern if the object was struck forcibly or softly. That's a lot of information your brain picks up and processes in a very short amount of time.
This leaves a fair amount of work to the synthesist - just as there is a difference between simulating a basoon and a baboon, it takes just as much consideration synthesising the difference between a snare drum or a cowbell, only you have to pack that information into a tiny slice of time. As with any synthesized sound, there are certain elements that comprise the sound - pitch, loudness, and harmonic content, and the change of these three parameters over time.
So, what gives a percussive instrument its unique sound? First of all, the typical percussive instrument has to be struck by some object in order for it to make any type of sound. So, we can deduce right away the instrument must produce the sound of that intial impact - the stick striking the drum head, for example. The second part of the percussive sound is what happens after it is struck - if it's a hollow drum, the sound will resonate for a comparatively longer time than if it is more of a solid object, such as a woodblock. If the drum has a skin that is struck, like the snare or tom, that skin actually stretches for a brief time, which causes a short bend in pitch as it is struck. If the percussive instrument is metallic, it may be expected to produce some dissonant harmonics, or "pitched noise" as it rings. And, if there are snares for example, the drum will produce a wider set of frequencies more closely resembling classic white or pink noise.
So, we can break the percussive sound down to three general ingredients: Impact, Shell (the body of the instrument) and Noise. Not every percussive voice will use all three elements, but they will all use at least a couple of them.
Impact is the element of the percussive sound derived from the drum stick, hand, mallet or other object striking the percussive instrument, and therefore forms the intial portion of the sound itself. The information contained in the impact portion of the voice not only suggests the type of object striking the instrument, but also the type of surface that is being struck - a skin, wood, metal, etc. The pitch of the impact will vary from instrument to instrument. The impact envelope is generally of much shorter duration than the shell envelope. Most percussion synthesizers actually derive the impact portion of the percussive sound by filtering the pulse that is used to initiate the drum sound - the pulse may be filtered to eliminate, for example, the higher harmonics. In other words, it may be made "brighter" or "duller".
The MPS does not use this method. Instead,the MPS uses an actual tunable swept pulse wave oscillator to generate the impact portion of the percussive sound. The impact oscillator was inspired by a patch described by the great Roger Powell in which he actually dedicated a VCO to generate the impact portion of a percussive sound patch. This method creates a much more realistic impact effect, and is one aspect, if not the most important aspect, responsible for the sheer realism in percussive sounds the MPS is capable of generating.
The effect of an impact to the physical body of the percussive instrument is simulated by the shell parameter. A physical percussive instrument will typically resonate, ever so briefly, or even not-so-briefly, when it is struck. The frequency, duration and harmonic content of this resonance is determined by the size, shape, and materials that make up the body of the instrument. As an example of the electronically simulated sound of a briefly resonating object, consider the woodblock-like ticking of the atomic clock on the shortwave WWV stations - that sound is actually made up of a five millisecond burst of either a 1,000 Hz sine wave or a 1,200 Hz sine wave (depending on the location of the transmitter).
The drum heads of many percussive instruements, such as bass drums, toms, snares, etc, will actually flex as they are struck - this causes the pitch to intially bend up briefly as the head is struck. The shell section of a percussion synthesizer typically provides variable amounts of pitch bend. This bend is usually more subtle in standard physical drums, but can be exaggerated, as in the previously mentioned electronic disco drums. If this parameter is set for an excessive pitch bend, you will be transported to the land of Donna Summer at Studio 54, which you may or may not want. The MPS utilizes a voltage controlled, tunable swept oscillator as the shell oscillator.
What I sometimes think of as a sub-function of shell is something Thomas refers to as "Clank" - this is a short duration of a harmonically rich signal and is vital for the metallic, hollow percussive instruments such as cowbells. The effect of clank is to give a sort of tuned, metallic sound to a percussive voice. The MPS generates this parameter through use of a separate tunable oscillator ring-modulated with the shell oscillator.
Shell envelope durations are typically, but not necessarily, of a longer range than impact oscillations. The MPS provides a wider envelope generator time range for the shell envelope generator than the impact envelope generator.
Noise tends to get a bad rap. Very few people make it through their adolescence without at least once hearing an irritated adult loudly directing one to "Turn that noise off"; specialized headphones are manufactured to cancel out ambient noise for those unlucky enough to have to deal with traveling by aircraft on a regular basis; all types of audio devices will invariably brag of some sort of low noise specification or other. Yet, here you are considering building a device that purposely inserts noise into the sound it produces.
The subject of noise can run very deep - I've read dissertations and explanations of noise that range from the very technical to the quite philosophical, but the long and short of it is anybody who spends much time inside an anechoic chamber could tell you that your life would be close to unbearable without a bit of noise here and there. Our brains use noise at often a subliminal level to constantly process our surroundings. For example, with the introduction of digital technology to cellular phones, the cell phone providers realized that they actually had to insert what is known as "comfort noise" into the signal - the noise of analog technology had been completely removed, and the manufacturers realized that without some sort of audio cue, their customers had a difficult time determining if they were still connected to the party at the other end of the connection.
Noise plays an important role in the brain's interpretation of a percussive sound. As an example, the snares of a snare drum will emit "wideband" noise, similar to the noise a radio produces when tuned between stations, but only for a very brief time. This simple burst of information tells the brain it is, in fact, listening to a snare drum. Perhaps the burst of wideband noise is even much shorter - short enough to represent the impact phase of striking a percussive instrument. Here, the noise would be delivered in a very short, yet identifiable burst produced by a wooden stick glancing against a skin surface.
One reason the sound of striking a cymbal or high hat differs from the sound of striking a bell is that the overtones generated from striking a bell are fairly constant in frequency and "purity", whereas a cymbal will generate a series of overtones that more resemble noise. This clangorous sound can be duplicated by frequency modulating an otherwise pure tone with a noise signal. This is one method of producing "pitched noise" - a noisy signal that appears to have a general pitch, such as the noise of a vacuum cleaner or a trash can slammed to the ground at 6:00 AM on the morning you get to sleep in late.
When the "static" noise of a radio tuned between stations is passed through a bandpass or lowpass filter, the output of the filter may resemble perhaps wind, surf or thunder as the cutoff frequency of the filter is varied. When the filtered noise is not sustained, but instead is heard as only a very brief burst, it sounds instead as some sort of physical impact.
The MPS provides a white noise source that is passed through a resonant voltage controlled filter that is switchable between low pass and band pass responses. As with the static "hiss" example above, the filter serves to remove a portion of the noise in order to mold it to the purpose at hand. Through control of the filter's resonance, a specific range of specra may be emphasized. One obvious application of this section of the MPS would be to reproduce the effect of snares. Another use of the filtered noise might be to enhance the impact by providing a brief burst of noise spectra - lower noise frequencies may indicate a larger instrument whereas selectively higher spectra indicate perhaps a smaller instrument. The filtered noise envelope may be lengthened while emphasizing the higher frequencies. This signal could then be used to modulate the shell VCO to turn the "clean" shell signal into a shimmering rendition of a nice open high hat or cymbal.
The frequency content of a percussive instrument can vary over the duration of its voice. This applies to the noise component as well. When first struck, there may be a predominantly larger amount of higher spectra that diminishes as the voice reaches the end of its envelope. Therefore, both the center frequency of the filter and loudness of the VCA the filtered noise passes through are controlled by the noise envelope generator. The noise envelope generator has the same time constant as the shell envelope generator (the impact envelope generator is the only envelope generator that has a purposely shorter time constant). This arrangement allows dynamic noise frequency content over time, and also controls just how long the noise portion of the programmed instrument will last, thus helping to define the instrument.