Making sounds with analogue electronics - Part 8: Early versus modern implementations

Temperature drift of the octave interval is the problem that most people mean when they say that analogue synthesizers go out of tune. Trying to match two exponential curves means that two interdependent parameters need to be changed: the offset and the scaling.

The offset sets the lowest frequency that the VCO will produce, whilst the scaling sets the octave interval to get the doubling of frequency for each successive octave. On a monophonic instrument this is not so hard, and any slight errors only help to make it sound lively and interesting. For polyphonic analogue synthesizers, this process can be very time consuming and very tedious. With lots of VCOs to try and adjust, the problem can begin to approach piano tuning in its complexity.

One method used to provide an 'automatic' tuning facility for polyphonic analogue synthesizers was introduced in the late 1970s. A microprocessor was used to measure the frequencies generated by each VCO at several points in its range and then work out the offset and scaling correction CVs. Because of the complexity of this type of tuning correction, and its dependence on a closed system, it has never been successfully applied to a modular synthesizer. (Autotuning is covered in more detail in Section 4.3.)

High-frequency tracking is the tendency of analogue VCOs to go 'flat' in pitch at the upper end of their range. This is normally most noticeable when two or more VCOs are tuned several octaves apart, and although often present in a single VCO synthesizers, is only apparent when they are used in conjunction with other instruments.

Most VCOs use a constant current source and an integrator circuit to generate a rising voltage and resetting the integrator when the output reaches a given voltage. This produces a 'sawtooth' waveform. The higher the current, the faster the voltage rises, and the sooner it will be reset, which produces a higher-frequency sawtooth waveform.

At low frequencies, the time it takes to reset the integrator is not significant in comparison with the time for the voltage to rise. But at high frequencies the reset time becomes more significant until eventually the waveform can become triangular in shape, which means that only one part of the waveform is actually controlled by the current source, and so the oscillator is not producing enough high frequency (Figure 3.7.1). Some VCO designs generate a triangular waveform as the basic waveform and so do not suffer from this problem.

FIGURE 3.7.1 At low frequencies, the rising part of the sawtooth waveform is much longer than the fixed reset time. But at higher frequencies, the reset time becomes a significant proportion of the cycle time of the waveform and so the frequency is lower than it should be. This high-frequency tracking problem needs to be compensated for in the CV circuitry of the VCO.

Controllers are another source of tuning instability. The stability of the pitch produced by a VCO is dependent on the CVs that it receives. So anything mechanical that produces a CV can be a source of problem.

Slider controls are one example of a mechanical control, which can be prone to movement with vibration, whilst pitch-bend devices with poor detents can cause similar 'mechanical' tuning problems. The detent mechanism varies. One popular method involves using the pitch-bend wheel itself – it has two of the finger notches opposite to each other. One is used to help the user's fingers grip the wheel, whilst the other is used to provide the detent – a spring steel cam follower clicks into place when it is in the detent and pops out again when the wheel is moved. This can wear, and produce wheels which do not click into position very reliably, which can mean that the whole instrument is then put out of tune.

3.7.2 Voltage control
As has already been mentioned several times, despite the name, most of the electronic circuitry used in synthesizers is actually controlled by currents, not voltage! The voltages that are visible in the patch-cords in the outside world are converted into currents inside the synthesizers and the control is achieved using these currents. Two 'standards' are in common usage:

1. 1 volt/octave
2. Exponential.

1 volt/octave
The 1-volt/octave system uses a linear relationship between the CV and pitch, which in practice means that there is a logarithmic relationship between voltage and frequency. This means that small changes in voltage become more significant at higher frequencies – just where small changes in pitch might become significant and audible as tuning problems. A 0- to 15-volt control signal can be used to control a pitch change of 15 octaves.

Exponential
The exponential system uses a linear relationship between the CV and the frequency. Because this method provides more resolution at high frequencies, it can be argued that it is a superior method to the 1-volt/octave system, since minor tuning errors at low frequencies are less objectionable. If the highest CV is 15 volts, then one octave down is 7.5, then 3.45, 1.875, 0.9375, 0.468,75, 0.234,375, 0.117,187,5, 0.058,593,75, and so on …, halving each time. Note that just eight octaves down a voltage change of 58 millivolts is equivalent to an octave of pitch change.

Despite the apparent advantage of the exponential system, the most popular method was the 1-volt/octave system. Conversion boxes that enabled interworking between these two systems were available in the 1970s and 1980s, but they are very rare now.