Project 5-1: Filter Modes & Cutoff Frequency. Create a patch in your synth with the following characteristics:
• Use only a single oscillator.
• Select a waveform rich in harmonics. If it’s a sampled wave, choose one that sustains at a high level so you won’t need to restrike the key. For this experiment, a simpler wave, such as a sawtooth, is better than one that has a lot of built-in animation.
• Set the amplitude envelope for a fast attack and full sustain level.
• Set all of the modulation inputs to the filter (from envelopes, LFOs, velocity, keyboard tracking, etc.) to zero. Depending on how your synth’s operating system is designed, you may have to hunt around a bit. The modulation may be programmed in the source module, in the filter itself, or in a modulation matrix.
Begin with the filter in lowpass mode. If you have more than one lowpass mode, pick one at random.
Play a note in the middle of the keyboard. While holding the note, move the filter cutoff frequency slowly from its lowest to its highest possible setting, and listen to the changes in the sound. If your instrument has a knob for the cutoff, pay attention as well to how smooth or grainy your knob movements are. Try faster knob movements, again listening for graininess. (Graininess can be cool, but unwanted graininess is less cool.)
Add some filter resonance, and repeat the manual cutoff sweep. Add still more resonance, and repeat it again.
Go through this process with each of the filter’s modes (highpass, bandpass, etc.). This will give you a good feel for the range of tones you can get from your filter.
Project 5-2: Filter Envelope & Velocity. Starting with the patch you used for Project 5-1, do the following:
Program the filter envelope (or a general-purpose envelope that can be assigned to filter cutoff) for an instant attack, a reasonably quick decay, zero sustain, and a release that’s about the same length as the decay. Make sure the envelope is not being modulated in any way by velocity.
Set the filter to lowpass mode (with a 24dB per octave rolloff slope, if possible). Turn the cutoff down to a medium-low value, and turn the resonance up about halfway.
While striking a key in some suitably interesting rhythm, gradually increase the amount of envelope modulation of the filter. You should hear the envelope adding a resonant sweep to the beginning of each note.
After setting the envelope amount to an intermediate level, neither too high nor too low, start adding velocity modulation to the envelope amount. Play the keyboard lightly and then harder, and observe the sonic differences. If your synth allows velocity to modulate decay time, add this type of modulation. You may need to shorten or lengthen the decay time to keep the decay segment in a usable time range.
Try inverting the envelope, so that the beginning of the sound rises from the floor rather than descending from the ceiling.
Chapter 6
LFOs
The humble low-frequency oscillator, better known as an LFO, is found in one form or another on virtually every synthesizer. Once in a while a manufacturer chooses a different term: For many years, Korg called their LFOs modulation generators (MGs). By any name, an LFO performs the same basic task: It’s a source for control signals with which to add cyclic (repeating) modulation to the tone.
That’s a bit abstract; let’s see if we can make it more concrete. To begin the discussion, let’s assume the LFO is producing a sine wave. When this sine wave is routed so as to modulate the pitch of an oscillator, the pitch will rise and fall in a smooth, regularly repeating manner. In other words, the LFO will produce vibrato. When the LFO’s sine wave is modulating amplitude, the LFO produces a fluttering or stuttering sound called tremolo. This doesn’t actually sound much like the tremolo (Italian for
“trembling”) that a string player produces by scrubbing the string rapidly with back-and-forth motions of the bow, but it’s close enough conceptually that we can borrow the term.
When the LFO is modulating filter cutoff, the result can be anything from a subtle swell and fade in the high overtones, through a “wah-wah-wah” sound, to flat-out tremolo. (You’ll recall that in Chapter Five we defined a filter as a frequency-dependent amplifier. If the LFO modulation of the filter cutoff is turned up high enough to sweep the cutoff through the entire frequency spectrum, it will affect the amplitude of all of the frequency components of the signal. Depending on the LFO waveform being used, filter-based tremolo will probably sound somewhat different from amplifier-based tremolo, but both types qualify as tremolo.)
LFOs can be used for many other types of modulation — for example, changing the pan position so as to sweep the sound from left to right and back again, changing some aspect of the timbre via pulse width modulation or FM amount modulation, changing the delay time in a chorus or flanger module, or even changing the rate of another LFO.
In some instruments, an LFO will be “hard-wired” to the destination that its signal can modulate. You may see an LFO that’s simply labelled “vibrato,” for instance, because it’s dedicated to modulating the pitch. In other instruments, one or more of the LFOs may be general-purpose modules, capable of modulating various aspects of the sound. Sound designers tend to prefer the latter approach, because it gives them more flexibility, but depending on the intended market for the synth, the manufacturer may feel that offering a dedicated vibrato LFO makes it easier for the user to understand what’s going on. Most effects processors have dedicated LFOs, however, because the effect processor is usually monophonic — it’s only instantiated once, and is operating on the summed signal coming from all of the synth voices.
Because of this, using a voice LFO wouldn’t make much sense (for more on monophonic vs. polyphonic
LFO behavior, see “Trigger Modes,” below).
Waveforms
LFOs typically offer the sound designer a choice of several waveforms (see Figure 6-1 ). The basic waveforms, found in most LFOs, are handed down from the early days of analog synthesis. Certain of these waveforms are ideal for producing familiar musical effects such as vibrato, trills, and tremolo. In recent years there has been a drift (calling it a trend would be too strong) toward adding more complex waves suitable for special effects. Because the waves are stored digitally, adding them costs the manufacturer almost nothing, but the jury is still out on how useful they’ll prove to be musically. These odd waveshapes are not discussed below.
Waveform modulation, as described in Chapter Four, is available in some LFOs, but it’s not too common.
Sine. As Figure 6-1 shows, a sine wave changes smoothly throughout its cycle. It’s a good choice for vibrato and any other type of modulation in which you don’t want to produce any abrupt discontinuities (jumps) in the sound.
Triangle. An LFO’s triangle wave is similar in shape and function to its sine wave. In fact, the two are so similar that some synths offer only one type or the other, not both. The triangle wave has “corners” at the top and bottom of its travel, but I’ve never encountered a situation where these produced any audible glitching in the sound, because there’s never a discontinuity in the level of the LFO signal. I tend to favor triangle waves over sine waves for common applications like vibrato and panning, because the sine wave spends more of its time at the outer ends of its travel — the upper and lower regions of the waveform — and less time transitioning through the middle of its range. This is audible: A sound being panned by a sine wave, for example, seems to slow down as it approaches the left and right speakers, “sticking” or
“hovering” for a moment, and to whip rapidly through the middle of the stereo field. When panned by a triangle wave, the same sound will be in constant motion.
Figure 6-1. The waveforms found most commonly in LFOs (top to bottom): sine, triangle, square, positive-going sawtooth, and negative-going sawtooth.
Square/Pulse. When an LFO square wave is used to modulate an oscillator’s pitch, the result is a trill:
The pitch alternates between a higher and a lower value. Depending on how your synth is designed, tuning the trill to a musically meaningful interval may be easy or difficult, as we’ll see in the section below on
“LFO Amount.” When a square wave is used to modulate amplitude or filter cutoff, the result is some type of tremolo. Depending on the depth of the LFO modulation, the tone could be chopped off entirely during the lower portion of the square wave’s cycle, or the level could just drop a bit. Changing the amount of LFO modulation during the course of a note can produce a striking effect.
In some LFOs, the square wave’s pulse width (see Chapter Four) can be adjusted. Depending on what
else is going on in the sound — for example, if you’re creating a four-note pattern as described in Project 6-1 — a pulse width greater or less than 50% can be useful.
Square and sawtooth waves have abrupt discontinuities in their waveshapes. Unless your instrument smooths these transition in some manner, modulating amplitude or cutoff with either of these waves will quite likely produce clicks at the vertical edges of the waveform.
IDEA FILE: If you’re using a patchable analog-type synth, try processing an LFO square or sawtooth wave through a resonant lowpass filter whose cutoff is set extremely low (in the 5Hz range). Use the output of the filter as a modulation source for some other module, such as an oscillator. As you turn up the resonance, the vertical edge in the waveform will cause the filter to
“ring” at its resonant frequency, which will produce a burbling effect in the module that’s being modulated.
Sawtooth. Sawtooth wave modulation has no counterpart in acoustic instrument design or performance, but it has become part of the standard repertoire of synthesizer special effects. When pitch is being modulated by a sawtooth (or “saw”) wave, it will rise (or fall) smoothly for some period of time, jump suddenly back to the other extreme, and then start to rise (or fall) smoothly again. If the LFO rate is between 0.5 and 3Hz, the pitch modulation is fairly extreme, and the oscillator is producing a continuous tone of some sort, sawtooth modulation sounds unnervingly like an electronic ambulance, an air-raid siren, or a spaceship weapon.
Some synths provide rising and falling sawtooth waves (see Figure 6-1) as separate waveforms. In other instruments, only one sawtooth wave is provided, but it can be used as a rising or falling wave by changing the modulation amount from positive to negative.
If a synth has some form of LFO waveform modulation, you may be able to morph a single waveform from negative-going sawtooth through a triangle wave to a positive-going sawtooth, as shown in Figure 6-2.
Stepped Random (Sample-and-Hold). In order to explain the last of the common LFO waveform choices, we need to take a brief detour to explain what a sample-and-hold module is. The reason for the digression will become clear.
Figure 6-2. If an LFO is capable of waveform modulation, you may be able to morph the output from negative-going sawtooth through triangle to positive-going sawtooth.
Figure 6-3. The random input (curving line) and stepped output (horizontal lines) of a sample-and-hold circuit. Pulses from a clock source (shown along the bottom of the diagram) cause the sample-and-hold to change its output so that it matches the current level of the input.
In an analog synth, a sample-and-hold module (often abbreviated S&H or S/H) is a type of control voltage processor. It usually has an input for the voltage to be processed; the input can come from any voltage source in the synth. Within the sample-and-hold is a clock source. A clock is similar to an LFO except that its output is a trigger rather than a waveform. (To learn more about trigger signals, see Chapter Seven.) Each time the clock sends out a trigger, the sample-and-hold measures (samples) the level of the incoming voltage. It then sends to its output a signal whose voltage level is the same as the level just measured. The crucial point is this: In between clock pulses, the output voltage doesn’t change, no matter what happens to the input voltage. Even if the input fluctuates in a continuous manner, the output is always stepped. One possible result is shown in Figure 6-3.
Sample-and-holds are configured by manufacturers in various ways. Some have an external clock input, so that they can be triggered from some other module — even from a keyboard, so that each note played causes a new sample to be taken. Because white noise (a randomly varying signal) is the most commonly used input for a sample-and-hold, some S&H modules have built-in noise generators. Some have an input for the signal to be sampled, which is a very desirable feature, while others always sample their internal noise source.
IDEA FILE: If your sample-and-hold has an input for the signal to be sampled, send it a sawtooth wave from an LFO. For this to work, the S&H itself has to be getting its clock triggers from some other source, not the same LFO. While modulating the pitch of the oscillator (s) from the output of the S&H, change the frequency of the sawtooth wave. As the sawtooth speeds up, the rhythm of the steps will stay constant but the number of steps in each staircase will get smaller, while the individual steps get farther apart.
When you select an LFO’s sample-and-hold “waveform,” the usual result is that you’ll get a stepped random output. In other words, the LFO’s sample-and-hold process will always sample a noise signal or the equivalent. Applying the S&H signal to pitch, filter cutoff, or both produces the somewhat hackneyed but still useful “electronic gizmo” effect heard in many science fiction films of the 1960s and ’70s.
A few LFO-based sample-and-holds can be set up to sample other inputs, such as another LFO. If the other LFO is producing a sawtooth wave, the sample-and-hold will produce staircase effects, as shown in Figure 6-4.
Figure 6-4. When a sample-and-hold is sampling a sawtooth wave rather than noise, the output is a
stepped control signal.
For some reason, LFOs are sometimes designed so that when you choose the S&H “waveform,” the clock source used by the sample-and-hold will be faster than the LFO rate you’ll hear if you choose another waveform. I have no idea why this type of design is used, but I’ve seen it in more than one instrument.
Smooth Random. Related to the stepped random output of the sample-and-hold is the smoothed random “waveform.” Found on only a few synths, this selection produces a smoothly varying but unpredictable signal. A very slight amount of smooth random modulation can add an attractive sense of
“liveness” to the sound. Larger amounts are useful for special effects.