Electro-Music Engineer (Part 2)
Tuning Up — A Review of VCO Calibration Methods
Tuning and Calibration, Part 2
This month we continue with some other practical methods for tuning and calibrating your VCO's.
Conventionally the vertical axis of an oscilloscope displays the amplitude of a signal applied to its input while the horizontal axis displays time. Timing is obtained through an internal sawtooth oscillator and if the oscilloscope has a calibrated scale then this may be used to obtain an approximate measure of frequency. For example, a waveform input having a frequency of 1kHz displayed with a sweep time of one millisecond per division of the graticule should show one complete cycle of the waveform per division.
Most oscilloscopes have facility for using an external time base on the horizontal input. Suppose that the same sinewave signal is applied to both inputs then due to the nature of the timing waveform and the equality of the two inputs a perfect circle should be displayed on the screen. Whether the circle is symmetrical will, however, depend on the quality of the amplifiers. If separate sinewave oscillators are now used for the two inputs then when their frequencies are equal a circle will again be generated. In this instance the circle could be asymmetrical due to variations in amplitude of the two inputs. Also as one of the oscillators drifts slightly the circle will turn into an ellipse, or even a straight line, at a tangent to the vertical axis. The latter is due to phase differences between the two signals. These displays are known as Lissajous Figures. Of greater interest from a calibration point of view is that if the frequencies of the two inputs are integer ratios then clearly defined figures are generated. For example, if the ratio is exactly 2:1 then two complete ellipses are formed resembling a bow tie. For calibrating it is best to keep to a 2:1 (or 1:2) relationship since at higher ratios the complex patterns formed as the oscillators drift slightly apart can be very confusing.
In references to the use of Lissajous Figures the additional requirement usually stated is a Signal Generator. The latter is just a name for an oscillator and so another VCO would serve the purpose and the essential factor is that the second oscillator should be stable - at least for the duration of the calibration. Similarly, the method usually refers to using sinewaves but many voltage controlled oscillators do not have a sine output. A triangle waveform will do just as well and while two triangle inputs are assumed below, a combination of sine and triangle waveforms will work equally well, but the shapes obtained will differ. The calibration arrangement is shown in Figure 3.
From a keyboard keep note A = 440Hz held and adjust the frequency of the stable oscillator to match the frequency of the VCO being calibrated. This will be shown by a stable single diamond shape on the oscilloscope screen (Figure 4a). At this stage you may need to adjust the amplitude control(s) to obtain the best shape. Now press next lowest A and if the frequency is not half that of the stable oscillator there will be a complex pattern on the screen. To find out whether the VCO being calibrated is sharp or flat adjust the frequency of the test oscillator first one way and then the other in order to obtain a stable pattern resembling Figure 4b. Do not worry about exact shape or whether in fact you end up with a figure resembling an eight the important feature is the figure now resembles a doubling of the original shape obtained at the first matching. If the frequency of the stable oscillator had to be decreased then the VCO is sharp, i .e., the frequency ratio is greater than 2, and the 'scale adjust' preset should be turned accordingly. Continue the first step (keyboard A = 440Hz; adjusting stable oscillator to match; pressing next lowest A) until pressing A = 220Hz gives a figure of the type shown in Figure 4b. In practice it will be difficult to maintain absolutely stable Lissajous Figures but so long as they are rotating at a speed of about one per second then this is adequate. This is not necessarily a reflection of the stability of the VCO but more the stability of the control voltage and the accuracy of its measurement.
The procedure is similar when using an external control voltage for calibrating. The initial slight difficulty that some may experience is obtaining a frequency in the region of 440Hz (within ±50Hz will do). This can be obtained by simply listening to the output of the VCO through an amplifier and getting in the region of A = 440Hz by ear. Alternatively if the oscilloscope has a calibrated time scale then use the method described earlier. Another approach relies on the fact that most constructors will be familiar with the sound of the dreaded mains hum and the control voltage can be set so that the output of the VCO into an amplifier resembles this. Alternatively, the output from the secondaries of a low voltage transformer may be input to one side of the oscilloscope and the VCO frequency matched to a single figure. With both the latter methods the control voltage is then increased by three volts which should bring the VCO into the correct frequency range. Note the voltage reading, get the single Lissajous Figure as above and then decrease applied voltage by exactly 1.00 volts and proceed as already described.
The Lissajous Figure technique is as accurate as the first method using a digital frequency meter but the remaining problem is having scaled the VCO one does not have the means for tuning it to an exact standard. The technique may also be used for the high frequency adjustment (when fitted) although the figures will be less stable. Simply aim to get the double figure rotating as slowly as the initial single setting up figure - or at least as slowly as patience will allow!
This is a popular method since it does not require additional test equipment and many constructors will know someone who has a tuned keyboard instrument. In fact, it does not matter for calibration purposes whether the musical instrument is exactly in tune so long as it is a polyphonic instrument which uses a frequency divider. If the instrument is another synthesiser then the calibration can only equal its scaling and in these circumstances it would be preferable to use Method 4. The test set-up is shown in Figure 5.
The arrangement shows the outputs going to a stereo amplifier but they could equally well go to separate amplifiers/speakers which may be built into the instruments. The method will work best if the waveforms are of low harmonic content, for example, a triangle from the VCO and an equally mellow sound from the musical instrument being used. Also observe that in this and other methods we show the voltage controlled oscillator being calibrated going direct to the amplifier, or other equipment. In some synthesisers it may not be practical to have a direct link from the VCO but this can usually be circumvented with jump leads. The alternative is to connect the amplifier, or other test equipment, to the output of the synthesiser. Normally there will be a voltage controlled filter and amplifier in the signal path and these should be set to their fully 'open' position. If the VCA does not have an 'open' facility then set its envelope generator to maximum sustain and the note on the keyboard will have to be kept held down during the calibration step - a matchstick wedged between keys will keep the hands free!
From the keyboard attached to the oscillator being calibrated press A = 440Hz and also press the same note on the musical instrument. In this instance the two sounds will have to be matched using another adjustment attached to the VCO, for example, the initial frequency adjust (RV17 on VCO 1 of the 'Spectrum'). As the pitch of the two instruments become closer one should hear a secondary beat frequency and the time interval between beats decreases as matching approaches. When zero beat is obtained press next lowest A for both VCO and musical instrument. In this method the best approach is to turn the 'scale adjust' pre-set until zero beat is obtained once more. The direction of turning as well as the number of turns to achieve matching should be noted. Afterwards turn the pre-set half the number of turns back in the opposite direction. Remember, as in all methods, to keep an eye on the voltmeter and press the key again if the voltage changes during a calibration step. Press A = 440Hz on both instruments and repeat the above procedure of matching with the scale adjust preset and going back halfway. The number of turns of the preset will gradually decrease until only a part rotation is required for perfect matching.
When an external control voltage is used a different approach is required. Press A = 440Hz on the musical instrument and adjust the calibrating voltage for zero beat. Note the voltage reading. Press next lowest A and reduce calibrating voltage until zero beat is obtained once more. If the voltage adjustment required is less than 1.00 volts then the VCO is sharp and the scale adjust preset is rotated in the direction to achieve the correct scale. If this is not known then turn it in one direction and repeat the previous step and note whether the voltage required is lower or higher than before. If lower then it is being rotated in the correct direction. Simply continue the procedure: A = 440Hz; match with applied voltage; press next lowest A and alter scale adjust preset if the voltage change required to achieve zero beat varies from 1.00 volts.
At high frequencies beats are much more difficult to discern and thus the method is not really practical for a high frequency trim adjustment. If this trim is fitted then normally its wiper will be grounded during the low frequency scaling and the best approach is to subsequently set the wiper in the mid position and tweak the pre-set if there is an audible tendency for the oscillator to go sharp or flat at high frequencies.
The advantage of using a correctly tuned instrument in Method 3 is that both calibration - at least at the important lower end - and tuning can be achieved. This 'beat frequency' technique may, however, be used for calibration when only a stable oscillator is available, such as, another VCO.
The arrangement is the same as Figure 5 but with the stable oscillator in place of the musical instrument. With a keyboard/VCO calibration set-up, note A = 440Hz is pressed and the stable oscillator adjusted until zero beat is obtained. Now press next lowest A and use the same procedure as Method 3, namely, alter the 'scale adjust' for zero beat. This time, however, the beats are on notes an octave apart and some will find them more difficult to hear. Turn the trimmer half way back, as before, and repeat the procedure until zero beat is obtained on both notes. Often some benefit will be obtained by adjusting the volume from the amplifiers so as to more clearly discern the beats.
The procedure using an external control voltage is virtually the same as that described in Method 3, except that the stable oscillator is not touched and one relies on hearing beats between notes an octave apart.
This method is applicable to most situations. For example, with the 'Spectrum', VCO 2 could be used as a separate stable oscillator by disconnecting it from the keyboard by omitting R163. Similarly, in a modular system, such as the Digisound 80, a second initially uncalibrated oscillator may be used as a stable VCO when calibrating the first and then the latter used to calibrate the second. Furthermore, the 'beat frequency' technique can be very accurate especially if the person has what is known as a 'musical ear' which is not the same as an ear for music. In contrast, however, there is a surprisingly high proportion of people who cannot hear or have difficulties with beat frequencies. These problems are increased when the beats are an octave apart. The next method overcomes this problem.
The set up is shown in Figure 6 and the techniques used are exactly the same as for Methods 3 or 4 according to whether a tuned musical instrument or stable oscillator is used. The outputs of both the VCO and the calibrating aid should preferably be waveforms of low harmonic content and in this instance they are taken to the two inputs of a ring modulator. One will recall that the action of a true ring modulator is to produce the sum and difference of the frequencies of the two inputs. If the inputs were two pure sinewaves then when the inputs become matched the output from the ring modulator will be a pure tone an octave higher and with an amplitude half that of the signal inputs. Thus there is a quieter pleasing tone when the two inputs are matched. If, however, we put in sinewaves of, say, 438Hz and 440Hz the output will contain frequencies of 2Hz and 878Hz. The effect of the 2Hz component is to cause variations in amplitude, rather like 'beats' but much more clearly audible.