Selezione Radio – DIY Synthesizer (part 5)

In 1976/1977 a DIY synth project series was published in the Italian electronics magazine Selezione Radio. Of course, other magazines had their projects too in the same years, but this synth is little known.

We are looking for people that built or own(ed) this synth: do you have images and sounds, some experiences or user stories, please share them with us.

We will set up a series of posts with the separate articles from the original magazine, including the scans of the pages and an OCR’d text in Italian. We also translated the text to English (with help of Google, hopefully). And besides the scans of the pages you can also download the full article as pdf.

The fifth part is about the building of a sinus converter, pulse modulation and glissando module.

English, translated OCR version of article Part 5

We build an electronic synthesizer


fifth part. by Federico CANCARINI


If you have followed us, up to this episode and have tried to realize what we have presented to you, we can bet on the certainty that, as soon as the modules are finished, you are immediately throw yourself on tangled skeins of wires and cables, then plugged in and out of plugs and jacks, and turned knobs for hours and hours…. Nothing demerit:,! ‘experience is the one that teaches best; the modules, then, are proof against any strange connection: the only thing we do not guarantee is the integrity of you and of the whole if any jack is plugged into a power socket in your home. Anyway, with this article, let’s go back to ‘Talking about the synthesizer : this will be followed by a whole series of new modules that are very sophisticated and will add a huge versatility to your already versatile since. tetizer. The circuit in question concerns a module that is complementary to our VCO, but I don’t know: it is very useful for any VCO that provides a triangular wave with an amplitude between 0.4 and 0.9 V pep. This module is a “Converter in sine wave and also 1.m modulator in amplitude of the square wave”. So it is an ideal addition to: l V: CO that we have already presented. Once the components have been assembled and checked that there are no wiring errors, you can proceed with the calibration of the module. It would be better to have an oscilloscope, but if you want to be, you can use your ears. Connect the Veo ( triangular wave ) in J 1 and the sine output J2 to an amplifier. Then apply the voltage to the circuit and wait, with the circuit under voltage, about twenty minutes to allow the complex to stabilize (especially CIO and necessary for the electrolytic capacitors) so that later, using this converter, a sophisticated integrated circuit, there are no calibration or instability problems. This stabilization must be done only once : the first. Then connect to one of the three control inputs of the converter 04 or ] 5 or J6). At this point, using one of the provided Biases d, d modulo f> ower Supply, apply a voltage to the V ! C, O driving constant and such that the note predicted by the Veo is approximately in the middle C. Now turn the R32 trimmer so that it has the maximum resistance: · 1101 d 1 or you should now hear no sound at the output of the module. Now · gently rotate this trimmer on the opposite side, until you hear the typical sound of a pulse of a small width: at this point you: stop. If you check on the oscilloscope, you will instead have to adjust R23 until you see an image like figure 1. Now unplug the cable, which goes to the amplifier, from J2 and plug it into J3 that is the sinusoidal output. You will now have to hear some sound in the amp; ator, and, by turning the calibration trimmer R33 you should also notice a change in the timbre of the sound; and b well; you have to adjust the R33 until you find that point where the sound produced is the softest possible. This is the regular position from R33. If you have an oscilloscope, you can see that for positions of R33 before and beyond the calibration point, you will have irregular waveforms (see figure 2) represented: A) R33 c.: with resistance too great B) R33 with too little resistance C) R33 exactly calibrated. Now, this regulation of the sine wave may have caused slight changes in the regulation of the pulse R32. ALiora repeat the calibration operation for R32 by reconnecting the pLi: lse output to the amplifier and adjusting R32 until you get the sound (or oscillogram) c.: that the pulse had before. All this must be set up keeping one of the control inputs (J4-J5-J6) connected to ground. Now, and only now, will you be able to remove this mass, and, since the sum of the control voltages is carried out by means of resistances, the removal of this mass will cause an extension of the pubis, which you will have to be able to feel. You can now connect the “, pulse” output to an oscilloscope and start driving (using a Bias) with a voltage from 0 to..1… 5 V at the control inputs. As you can see, increasing the tension increases the width of the “Pulse”, while the sound you hear will become fuller. On the oscilloscope, check that for a driving voltage of – 5 V ( ± 10%) the “Duty Factor” of the pulse reaches 50 % : that is, a regular square wave is output. Similarly, also check the two other control inputs. Lastly, ground one of the three inputs with trolile to the converter, then listen to the pulse output in the amplifier, while the frequency of the VCO is progressively varied in all its range: if it happens that for some fre When the output sound becomes muffled, retouch the R32 slightly until the sound returns.


The only compromise that has been made in designing this module concerns the passive summation network, used in the control inputs u and the modulator. The reason for this choice derives from the fact that, like the other modules that use a passive summative network, the process of adding the tensions is not so critical as to make it necessary, and therefore right. add an integrated that works as an active adder. Instead, more can be said about how to forge the sine wave starting from a triangular one: this process is obtained by exploiting the results of a non – linear feedback around the fourth Nor-ton amplifier (pins 10, 11, 12). This non – linear counteraction is obtained by exploiting a circuit where diodes are used whose break- point has been previously calibrated. Thus, when observing an oscilloscope, the resulting wave can observe how the sin1soidc oH nuta deviates from an ideal sinusoid : in fact it is obtained by approximation by joining straight line segments one after the other. For all practical purposes, however, since the number of breaking points is more than sufficient: no, the construction of the wave does not introduce harmonics, of higher order (cl n are not present in a sinusoid p, ra) because they remain at such a low level of sound it is practically inaudible, working in with the dictions grandmother li.


Inputs: jack J 1 is a medium impedance input (4 7 k) which must be connected to the triangular output d, VCO To obtain the best results, the peak-to-peak amplitude of the input wave must be be between 0 and 0.9 V. Sinusoidal output: the jac.:k J2 is the output of the octnut sine wave forging the triangular. The output has a rather low impedance, circ lk and the output wave has an amplitude of 0.5 V pep. Pulse output : the J3 jack provides a low impedance output from which to pick up the pulse wave) and adjustable width (naturally gives a control voltage ). Control inputs: the three jacks (J4-J6) are the inputs for a control voltage that can vary between 0 + 10 V. The nominal limit would be + 5 V, but the module tolerates an overange very good. » By 100%. With a total driving voltage between 0 and + 5 V the «duiy faotor:» of the impulsive wave varies between 3% and 50%. With a voltage against: Jlo d + 10 V 100% of the «du ty factor» is reached. Negative control voltages turn off the impulse and do not damage the circuit. You must consider this model as an expansion of our <VCO: in fact, with the addition <l this PWM converter will be considerably more versatile, as far as the “pulse)> is concerned, and moreover you will have, a new form ù ‘on da is available : the sinusoidal. As you well know, a couple of natural instruments have the possibility of producing a wave that is very close to the inusoidal one : one is the flute and the other is the drum. The synthesizer can comfortably reproduce the flute by mixing two or more sine waves (e.g. ma and its 2nd octave) and passing the utto through the VCA to give the dynamic sound of a wind instrument : the ADSR must have zero release and have moderate attack and fall times. Even rancasse, congas, bongos, mutes and any other percussion that is not mixed with metallic noises (such as a snare drum, for example) can be simply reproduced by the sin : etizer using a mol or damped sine wave (with a quick decay). It is therefore necessary to use the same configuration of connections that is used with the car (but using one or two sinusoids jj with a lower frequency) by adjusting the ADSR (Gen. of envelopes) for a very rapid attack, a perceptible release and a moderate-to decay ( figure 3 ) Obviously, a constant 5 – step for the flute will have to be used as the trigger for the ADSR.


First of all, let ‘s remember what the duty factor of an impulsive wave is. Let us refer to figure 4. The percentages represent the values of the duty factor in the three figures. Take the square above the zero line as a reference and observe, with respect to the total period T, the duration (in percentage) of the peak itself. The zero line is such that it results: Let us now observe the first and last figures: they are different for “duty factor”, but, very importantly, the human ear cannot distinguish the difference between “duty ” values factor ” which, added together, give 100%. This fact will be exploited later_ The central figure instead shows the typical square wave, that is a “pulse” with a “duty factor” of 50%. Now when the “duty factor” of an impulse wave changes a lot of relatively complicated and musically very interesting things happen, to the harmonic content the pulse becomes shorter (that is, the “duty factor” decreases), the foundation decreases in amplitude, while the amplitude of the higher order harmonics increases considerably. At the same time there is a relative phase shift between the harmonic components themselves which gives the sound an almost incredibly artificial timbre. Just as a beginning experiment, mounted such a module, press with a variable voltage one of the ent-ratc with the roll. Obviously you must first connect the triangular output of the VC: O. to the J1 jack, you will then take the signal from the J3 jack directly connected to the amplifier. To drive the control inputs, the sine wave of the oscillator with control set between 1 and 3 Hz will be optimally used. It is obvious that while the control voltage goes up and down, at the same time the “dutv factor” of the wave increases and de -crescèe and you will perceive a significant effect due to the progressive shift. In practice this effect is what you would notice if you listened to the sound coming from a speaker that emits a square wave, but which in the meantime turns on itself. You wouldn’t even notice any difference between the two sounds produced in such a different way. It is also obvious that you can drive the controlfo inputs with any variable voltage : in practice, using any source of envelopes. Interesting arrangements are illustrated in the block diagrams of figures 5 and 6. With the second connection diagram, try above all the pcrcw sions, adjusting the “amount” of the control dryer and the frlter of the nois until reaching the most disconcerting effect (how do you want it or not?). Do you know that a mixer is used: if you want you can “slam” everything into the two entrances of the VC: A. (O dB and 3 dB). The actual mixer will be described in a future article. Now that you have d, ivertiti e eh, with these rum.oracci, you have made the whole condominium ferocious, it is time to deepen the topic on the exploitation of Du-ty : Pactor. From what has been said previously, you will have noticed that the Duty Factor is 50 ° ~ for a control voltage of + 5 ‘ but that the circuit tolerates a 100% ovcrange well: that is, nuHa ~ breaks when at J4-J5-J6 the sum of the control tc-nsions equals about 10 V. AnZJi : this is for the applications that we will explain later. Just keep in mind that when you have 9 -; – 10 V to the control inputs s, we are in a situation close to! The fig. 4-C which, again for what has been said, is not very different for the human ear, from the situations of fig. 4A. In a nutshell, a Duty Factor impulse will have the same timbn as one with Duty Factor = 90%. Or: suppose that for some arcan: reason you need a no of his that varies from a very small Duty factor to one close to 100%. St you have only one voltage available ( pilot from 0 to + 5 V it is very easy to set the purpose: just apply it simultaneously to two control inputs (fig. 7). In fact, the result of a connection is an apparent duplication. of the tern = pilot ion: in reality it is the (passive) adder circuit that detects twice ( and therefore adds ) 1, the same voltage. equivalent, it allows us to conveniently exploit the module for many other needs. Snppo · we want to create a particular timbrc, which requires an impulse wave whose Du.ty factor decreases while the control voltage, instead. is going down. The first thing that comes to mind would be that of t, we would be an inverter to, in fact, invert the pilot voltage before even applying it to the no – the “In verter / Buffer” module that we will present to you later. “Pat ching “(figure 8). All of this is great but it only complicates things. Ideal Real. In fact, since our hearing does not distinguish whether the Duty Factor, varying from 50%, is growing or de ‘crescendo, approaching 0% or 100%. all we have to do is add, in a second control input, a constant << ‘Ilias’ of..L 5 V, while we will leave, as it is, in another control input, our pilot voltage. The resulting effect is that the Duty Factor now starts at 50% (0V pilot voltage + 5 V of B-ias) and increases more and more towards 100 % with increasing pilot voltage (pilot voltage +5 V 13- ias). But this, as we said earlier, produces a sound identical to another whose Duty Factor varies from 50% to 0%. Here is the “patching” (figure 9). Remember that this argument, applied here in this case to this module, + 18V R: t ! = The is, however, generally valid for many (if not all) the other modules that use a circuit. active or passive adder. If you do n’t remember everything upstream, review the article on the VCA that you will understand even better now.


The whole circuit of this module is based on the exploitation of the! The integrated 1: M 3900 or MC 3401P whose casing contains four differential amplifiers working in current. Referring to the electrical diagram drawn (figure 12), we observe that when a triangular wave is applied to the input of the module it is cleaned of spurious pulses and therefore is amplified by the first amplifier (pins 1-5-6) of the DIP ( dual in – line packagc), before being applied to the pulse width modulator (two amps, pins 2-3-4 and 8-9-13) and a, l sine wave converter (last amplifier pins l 0-11-12). The triinmer in sories to the input network (R32) adjusts the level of the gain of the first stage, which can therefore be varied to compensate for the differences in live signal of the incoming signal. The PWM modulator is essentially a comparator-adder. The current that passes through R20, caused by the voltage applied to its ends (voltage which is nothing more than the triangle wave amplified by the first stage) is added to the sum of the three currents produced by the control voltages applied to the ends of R21-R22- R23, This total current flow applied to the inverting input (3) of the second stage of the DlP, is compared with a refueling current which, through R25, is applied to pin 2, that is to the non -inverse input. As long as the reference current going to the amplifier is greater than the total flux going to pin 3, the amplifier’s output 4 remains “high”. Note that the resistance values have been calculated so that, for a small portion of each cycle of the triangular wave, the current reaching pin 3 is greater than the reference current, so that output 4 remains low. and we have at the output a wave with very narrow pulses for a control voltage of 0 V. When the control voltage increases, with respect to ground, the total current arriving at the inverting input ” – ” of the amplifier, it also increases, so that the points of the triangular wave, in correspondence with which the comparator clicks <1, are lowered more and more: but in so doing they also move away, so that the period of time between the edges is always greater of ascent and descent of the i pulso. The result of all this is therefore a pulse whose width increases as the control voltage increases. When the comparator trips, at a high or low level, it consequently activates the third comparator, which in turn inverts the impulse and also squares perfectly the rising and falling edges. Diodes D6 and D7 have the function of cutting the peaks at about one volt, thus reducing the maximum height of the wave and consequently the effect of the canceling operation that the amplifier carries out on the last pulse. The diodes in (Yltre directly supply the inverting inputs of the two comparators when they would be inversely polarized : if there were no diodes, the presence of a single polarization (and also inverse!) Would have negative consequences for the circuit and the integrated sinusoidal converter is of the classic type with the non-linear counter – reaction circuit, which exploits, to obtain this, the diode system and related break points or “break points”. In figure 1 l illustrates a real sinusoid and that which, on the other hand, is actually obtained with the 4th amp, fig. joining, what is a real sinusoid: Let sri start from point A when comparing the two sinusoids. At this precise moment, then, it means that the Sine Con verter is observed when the tniangular wave entering it is rising and starting daMo zero The fourth amplifier (feet ni 10-11-12) is used in such a way that it functions as an inverter : when the input has a voltage (current) that rises, it will be raised at the output while it is decreasing. At mid-wave (point A), all diodes in the counter-reactive network are inversely polarized, so that the only element of the counter- reactive network, for which there is conduction, is the resistance R 17, which alone, for now, determines the gain of the stage, the amplifier and thus determines the manner in which the voltage at the output of the amplifier rises. This corresponds to segment one of FIG. 11. But then the output voltage reaches a 1 level such that it is greater than the voltage present at point 1 of the electrical diagram 01 divider is here formed by R6 and R7). Now Dl is directly polarized and puts R 7 in parallel with R 17, determining a lower gain in the amplifier stage, and therefore we will have a new segment (the 2 of fig. 11) with a lower inclination than, l1). nea of zero. Then the output voltage increases again until it reaches the point where D2 is also directly polarized, putting RS in parallel with R 7 and R 17, and again, decreased the gain, there is a new ~ egment with an even lower inolination (11.3 in fig. 11 ). Finally D3 also leads and so R3 adjoins the previous parallel. Result: segment 4. We are at the top of the wave and now that it descends everything is repeated in the opposite way: little by little each of the diodes is re-polarized, inversely polarized and thus a further sequence of segments arises, but all in descent. Here then we reach point B: now the diodes D4 and D5 enter the d, iscourse which, one after the other, are polarized directly, producing such gains that the segments 5 and 6 cl do have, fig. 11. It would take, in truth, a 7th segment (breaking point) to have perfect symmetry : but since the ounce produced by the Veo is already rounded in the b \ SSO, it is useless to further complicate the circuit. 13 / JI – Prototype of the converter after assembly. cuito. Evcntuailly, if you use a perfectly triangular wave, you can add another diode in addition to D4 and D5 by experimentally calculating the divider: which is not a cliff, it is easy to have an oscilloscope.


In simple terms this circuit is nothing more than an R / C circuit added to the output of the sampler circuit. See FIG. 14. The figure also shows approximately what the situation is at the control inputs of V: CO. Here the resistors labeled R1 are those placed at the input of the adder circuit. The value of these resistors directly determines the gain of the summing amplifier and consequently the way in which the control voltage is converted into a certain frequency. Note that in this simple circuit the pot that determines the glissando time constant (Rt) also appears in series with the resistance Ri at the input of the summing amplifier, so that when the glissando time is increased, the output of the summing amplifier inside the Veo decrements, resulting in a note that descends more or less slowly from one key to another. In such a circuit the effect <li this resistance Rt is removed from the input of the VCO “buffer” constituted by I-Cl. The output impedance of this integrated circuit is in fact so small that compared to that, the input impedance of the VCO is practically negligible. To inhibit the operation of the glissato, S1 disconnects the capacitor C3 from the circuit. The two circuits Rt / Cl and R2 / C2 are power decoupling filters.


Prepare as follows : – the synthesizer: in serite the sky output glided to the control input of the VCO and the use of the VCO to a BF ‘amplifier; since the sliding adjustment potentiometer is equipped with a switch, turn it counterclockwise until it clicks (, off position). Press a key, release it, and hit one that is lower in pitch than the first. The note must change without there being any: r.marchevole slip indicating the glissato. Check yourself and be yes, sure that nothing has changed now, with the previous situation (without glossing over ) : everything is fine if everything proceeds as before the modification. Now turn the potentiometer and bring it to about halfway. Press a key again, release it and press: keep one of a tone lower than the first: the note must pass smoothly from one tone to the other. This indicates the glissato. P, the more the potentiometer is rotated clockwise,  the longer it becomes the sliding from one note to another. The maximum time should be about two seconds per octave. The effect of glissato- does not, however, modify the ” step ” and “impulsive” outputs of the keyboard.


Pitch: nuIJa has changed; this potentiometer still allows the keyboard to be raised by one octave by turning it completely clockwise, while its rotation counterclockwise decreases the output temperature supplied by each single key. Step trigger: gives an output voltage which rises up to + 5 V as soon as a key is pressed and which remains at this value until the key is released. Impulsive trigger: gives an output voltage that has a rapid peak.a + 5V and then falls on’b, i, to zero. Output: gives an output voltage which is proportional to the key pressed… Glissato: it’s the only new control. It is turned off by turning the potentiometer comp:] anticlockwise until the switch trips. As you move it clockwise, however, the range of slippage between one note and the next increases.

The complete kit of this synthesizer (mobile excluded) can be requested from our editorial staff at the price of L. 210,000 including postage.

Italian (OCR’d version)