Structure of almost all synthesizers is basically the same. It starts with the tone generator whose function is to create the sound. On analog synthesizers, this generator is a simple oscillator circuit, which can generate few basic waveforms like pulse, saw or sine wave. Therefore on analog synthesizers it is more common to use term oscillator than tone generator. On digital synths used terms are: wave gen, wave generator, tone generator, and sometimes, even an oscillator (though it is not a real oscillator inside).
Next comes the filter which defines the timbre of the sound and adds / removes harmonics from the original sound created in oscillator. Filter is followed by an amplifier, in which you set up volume change of the sound. Envelopes and LFO’s are used to manipulate various settings. For example, in oscillator they can control it’s pitch. In filter they can define filter changes over the time.
This was the basic description, however each synth can have it’s own and more complicated structure. Image above shows us the structure of Roland’s XV synthesizer. Each patch can contain up to four tones each with it’s own settings (sounds). Patch also contains common data, which consists of parameters that apply to all four tones like: patch name, overall level, octave shift, key mode (mono/poly), portamento settings, bender range and more. Patch also contains modulation control (matrix control) in which you specify which controller will change which parameter – for example mod wheel on the keyboard to change amount of cutoff and resonance of the filter or the pitch of the wave generator (WG).
The purpose of oscillator is to produce a sound that you will later process with a filter and amp. Once you press the key on the keyboard, you “activate” the oscillator (in analogue synthesizer it is actually always on). Oscillator (OSC) is the starting point of any synthesizer. It is a place where the waveform is being created. In analog synthesizer, oscilator can be digitally (DCO) or voltage controlled (VCO) and it usually produces pulse or saw wave of which pulse’s width can be controlled and even modulated (PWM). All oscillators, no matter for what application, work in the same way. If we look at their heart will find it looks something like this.
To be described in the simplest way, oscillator is an amplifier and a filter that operate in a loop. The basis of operation is the tuned resonant circuit – for example LC circuit that is made of inductor (L) and capacitor (C). In this circuit voltage and current vary sinusoidally with time and are 90 degree out of phase. There are instants when the current is zero, so the energy stored in inductor is zero, but at the same time the voltage across the capacitor is at it’s peak, while all of the circuit’s energy is stored in the electric field between capacitor’s plates. There are also instants when the voltage is zero and the current is at a peak, with no energy in the capacitor. Then, all of the circuit’s energy is stored in the inductor’s magnetic field. As you can see, the energy stored in this electrical system is swinging between two forms. Unfortunately this ”swinging” won’t go forever due to circuit losses. Conductors have some resistance, as well as capacitor and inductor. That is the reason why we need amplification. The goal of this amplification is not to add some high gain, but just to compensate the losses we have in LC circuit. You can imagine oscillator like pendulum, which due to drag and friction loose it’s movement energy, so you need to kick it from time to time.
At the output of oscillator we can have various waveforms, depending on the type of oscillator that we built. For example if our oscillator is producing a sinusoidal wave, we call it: sinewave oscillator. There are many types of oscillators, most known are shown on image above, and those are (from left to right): sine, triangle, square and saw. Square is actually a pulse wave, with 50% width.
Digital synthesizers have a different kind of “oscillator” in their heart. It is not a classic oscillator that we just described, but a device that uses PCM waveforms. Pulse code modulation (PCM) is a digital representation of an analog signal where the magnitude of the signal is sampled regularly at uniform intervals, then quantized to a series of symbols in a binary code. If you have some wav files on your computer, those are the PCM waveforms, the same that are in your digital synthesizer. Advantage of digital wave generator is that it is not limited to basic waveforms (sine, saw, square, etc.) but can contain any kind of data that was previously digitized. You can have real piano samples, guitars, drums, etc. Of course if you want you can have basic waveforms too. Most digital synths have at least a sine and a saw wave sample inside their waveform memory – so you can create some analog sounding tones too.
Now you might ask, why one needs analog synth, when the same type of waves can be created with digital PCM wave generator. The answer was actually just said is in the previous sentience, it is a word PCM generator. Each time you trigger this type of generator, it will produce exactly the same sounding waveform. Result is uniform tone that stays the same. In contrast, analog synthesizers have imperfect oscillators, which result in various types of fluctuations that are different every time you hit the key. In a sense they sound unpredicted and different on each other key. But that is just the beginning of the story. Once you engage pulse width modulation, there is no digital PCM synth that can enter this area. Then comes the analog filter, which again adds its own character. In short, this is one of the reason why 30 year old synthesizers usually cost more than the latest “state of the art” PCM digital synthesizer with 9000 patches.
It is important to understand that we talk about two different worlds here. It would be pretty naive to blame some digital synth and call it poor because it can’t do some good analog sounds. If someone asks why a PCM synth can’t make a super thundering Moog bass sound, the simplest answer is – it was never designed to so.
The most dramatic change of your sound is taking place in the filter circuit. The richer the harmonic content on the waveform, the more the change. Example of rich harmonic waveforms would be square and saw wave. If your synthesizer does not happen to have resonant filter, you are actually missing one really important and charming aspect of a synthesizer.
If we look at frequency domain of a sound, for example square wave which is rich in harmonics and play a low note at say 20 Hz, we can see that these harmonic components are spread in the whole audio range (20 Hz to 20 kHz). What filter will do is to isolate some parts of this range in order to accentuate it’s other frequencies.
Types of filters
The most know filter is the low pass filter – LPF. It reduces the volume of all frequencies above the cutoff frequency. You can specify this cutoff frequency in filter settings. Once you cut out high frequency range, the sound will become more mellow.
Next type of filter widely known is a high pass filter – HPF. It is doing exactly opposite thing from LPF. It cuts the part of the spectrum which is below the cutoff frequency. It can be useful for percussive sounds (nice analog sounding hi-hats can be made with it). High pass filter can be resonant too, but actually there are not so many synthesizers that feature it. Korg MS-20 one of the rare analogue synths that features analog resonant high pass filter.
An one of the most specific sounding filters is probably the band pass filter – BPF. This filter leaves only the region in the vicinity of the cutoff frequency, and cuts the rest. With resonance setting you are actually shaping the width of this filter. The more resonance, the more narrow this filter will be. Image below shows frequency response of three basic filter types we just described.
Once you activate resonance on a low pass filter, new harmonics will pop up in the lower range, creating new sonic components that didn’t exist in original sound. High levels of resonance can produce self oscillation. Roland’s Juno manual has a very interesting and specific explanation of resonance (i find it really fun to read it today, but actually it is a good explanation):
“This control emphasizes the cutoff point set by cutoff frequency knob. As you raise the knob, certain harmonics are emphasized and the created sound will become more unusual, more electronic in the nature. If you alter the cutoff frequency while the resonance knob is set to a high level, you can create a type of sound that is attainable only from a synthesizer.” – Roland Juno 106 Manual
Filter poles / slopes
Sometimes you will read that a filter is 4-pole type. This is just another term for 24 dB filter slope, which is the most common filter type in the world of analogue synthesizers. Number of poles defines the sharpness of the filter. The more poles filter has, the sharper it’s frequency response will be. This also affects the resonance, since sharper filter results in more powerful sounding resonance. Roland TB-303 uses the not so common, 18dB filter slope. Now you might wonder, what this 18 dB means anyway. It is a unit which tells you how much a filter will block per octave. Which in this case is 18dB. If you put a filter cutoff point to 440 Hz, one octave above at 880 Hz signal will be attenuated for 18 dB (which is about 63 times).
If a filter is 4 pole, the signal will be attenuated 24 dB, which for 440 Hz cutoff point means that the signal on 880 Hz will be about 255 times weaker. Image above shows attenuation curves of four filter types: 6dB, 12 dB, 18dB and 24 dB.
Which filter should you use? The choice is up to you. If you prefer nice smooth filter sweep sounds, you should use 12 dB filter with a good resonance value. For thundering bass sounds or high resonance zaps you should use 24 dB filter (full resonance for zaps). While 6 dB filter is generally not used, it comes hand for sample playback, to gently remove high end from the harsh sounding samples, while not totally distorting the phase of the sample.
Amplifier and envelope
Last stage of sound manipulation that is taking place in the synthesizer is happening at the amplifier section. Its purpose is to control volume changes of the sound. On analog synthesizers amplifier is usually called VCA (if it is voltage controlled) or DCA (if it is digitally controlled). On digital synthesizers amplifier is usually called AMP or in case of Roland it is called TVA, which stands for time variant amplifier. Main part of the amplifier is the ADSR envelope.
This stands for: Attack, Decay, Sustain, Release and it represents four points. Once you hit the key, you are at the attack point. With Attack you are setting amount of time that it will take for sound to evolve from its starting level, to the point where you press the key. This is followed by Decay in which sound can evolve to another level that you set. This is followed by Sustain point. As long as you are holding the key pressed, you are at sustain point, and the level you set at sustain point will be the level of the sound during the time the key is held. Once you release the key, the sound will go off. To prevent the sound from going away to soon you can set the Release point. Release sets the amount of time it will take for sound to get to zero level, after you released the key.
There are envelopes with more than four points that we described, but they all work in the same way. Roland’s typical envelope consists of two decay levels. Some Yamaha’s SY series synths such as SY-77 and SY-99 have loop points that you can set to the envelope which is pretty cool feature.
LFO and control
The purpose of LFO is to alter various sound settings in back/forth cyclic manner. Usually LFO can apply change to oscillator’s pitch, filter cutoff frequency and amp level. If you apply LFO to the pitch, you will get vibrato, if you apply it to filter, you get sweeping sound (such as wah-wah), if you apply it to amp level, you get tremolo.
As it’s name implies, LFO is an Low Frequency Oscillator. Usually it consists of a few basic oscillator wave types like sine, saw or square. You can imagine LFO like a helping device, which is helping in way that you don’t have to manually change the pitch or level of the sound. Instead you program LFO to do that. However if you do wish to change some setting manually, then you need to specify them in controller section.
The two images above show amplitude as a function of time (waveform display). As you can see original sound had constant level (amplitude). Once we applied LFO, level started to modulate from minimum to maximum value. Shape of the sine wave that was modulating original tone can be clearly seen on second image.
Audio example 2: Tone’s frequency modulation. Click here to hear original non-modulated sound. LFO was set to modulate oscillator’s pitch (frequency). LFO’s waveform was sine wave. Result can be heard here.
Image above shows frequency as a function of time (spectral display). It can be clearly seen how applied LFO modulates the pitch (frequency) of the oscillator. Shape of the sine wave that was modulating original tone can be clearly seen. Original tone was fixed frequency at 440 Hz. Once we applied the LFO, tone started to vary the pitch for about 50 Hz above and 50 Hz below original frequency.
Most of digital synths offer you to apply various controllers like mod wheel, aftertouch or velocity to some of the sound’s parameters like filter’s cutoff point or resonance, sound level, etc. Some synths call this feature modulation matrix. It goes something like this. First you specify the source of the controller – for example modulation wheel on the keyboard. Then you specify its destination, for example filter cutoff frequency. Now you set the amount, and you are ready to modulate the filter with modulation wheel. If filter settings on the sound are on the maximum open position, then you need to apply negative value to the controller, so that when you start to move the wheel, filter gets closed, for the amount you specified. Some better modulation matrix systems will allow you to apply almost any synth feature as a source to modulate it’s destination – for example LFO1 to modulate speed of LFO2 which can result in very complex and unpredicted results (this is in case the synth has two LFO’s). This is the area that requires a lot of experimenting, but results are always rewarding.
Abbreviations and common terms
Circuit bending is changing, removing or adding new electronic components within synthesizer to achieve different performance that is unavailable in original version. Usually cheap synths are being circuit bent to sound more wild, unpredicted, strange, or all together. Circuit bending results with warranty void, and can damage your synth permanently.
Legato is a function that should only work in monophonic mode. When Legato is on, pressing one key when another is already pressed causes the currently playing note’s pitch to change to that of the newly pressed key while continuing to sound. This can be effective when you wish to simulate performance techniques such as a guitarist’s hammering on and pulling off strings.
Modulation wheel affects the sound as specified by the control parameters (control matrix). On many synths it is set to vibrato by default.
Portamento is a function that causes the sound’s pitch to change smoothly from one note to the next note played. Portamento is common on guitar, violin and other string instruments. However, portamento is not possible on a fixed pitch instrument like the piano. On a synthesizer, parameter called ”portamento time” or ”portamento speed” defines the speed at which an oscillator moves to a new note you pressed on the keyboard. When the Key Assign Mode is mono, this can be effective in simulating performance techniques such as a violinist’s glissando.
Pitch bender (pitch wheel) bends pitch of the played note up or down, and is spring-loaded to return to center position.