A complete electronic organ to build at low cost

PART 2: Construction of the keyboards and of all the upper manual circuits described

To successfully make the Matinee organ you do not need any knowledge of electronics, though you will undoubtedly learn some as you go along. The only vital requirement is the ability to be able to solder correctly and neatly. Good soldering is a skill that is learnt by practice and it is most important that you learn this skill before starting this project.

How to solder

The main point to remember is that both parts of the joint to be made must be at the same high temperature. The solder will flow evenly and make a good electrical and mechanical joint only if both parts are at an equal high temperature. In this organ almost all the solder joints to be made are between a component wire and a printed circuit track, so in this case both the wire and the track must be at the same high temperature before the solder is applied. Before attempting this project practise soldering using a piece of Veroboard and some resistors or scrap components.

The first thing you'll need is a good soldering iron. There are plenty to choose from such as the Antex X25 or Litesold LC18 which both have a fairly high wattage to help the heat flow quickly into the joint so that the heat need be applied for only a short time. When the iron is hot apply some solder to the flattened working area at the end of the bit and wipe it with a piece of cardboard or damp cloth so that the solder forms a thin film on the bit. Always use a good quality solder such as Ersin multicore. A standard 60% tin, 40% lead alloy solder with cores of non-corrosive flux will be found the easiest to use.

If the wires of the component to be fitted are parallel to one another on the same side of the component (radial) they should fit directly into the PCB without bending, but if the leads come out of opposite sides of the component (axial) they should first be bent at right angles close to the body of the component. Push the leads through the PCB from the non-coppered side and splay out the ends of the leads a little on the coppered side so that the component does not fall out when the board is turned over and so that the wire touches the edge of the copper track.

Now melt a little more solder on to the tip of the soldering iron and put the tip so that it contacts both the copper track and the component wire. It is the molten solder on the tip of the iron that allows the heat to flow quickly from the iron into both parts of the joint. If the iron has the right amount of solder on it and is positioned correctly, the two parts to be joined will reach the solder's melting temperature in about half a second. Now apply the solder to the point at which the copper track, component wire and soldering iron are all touching one another. The solder will melt immediately and flow around all the parts that are at or over the melting point temperature. It should be possible to take the iron away within one and a half seconds of first touching the iron to the joint. Make sure that the lead and PCB do not move after the soldering iron is removed until the solder is completely hard. This should happen within five to ten seconds. If one of the components moves during this cooling period the joint may be seriously weakened.

The hard cold solder on a properly made joint will have a smooth shiny appearance, and if the component is cut off and the wire from the joint pulled as hard as possible the wire should not pull out of the joint. It should be possible to pull the copper track up off the Veroboard by pulling on the wire. In a properly made joint the solder will bond the wire to the copper track very strongly indeed. It is also important to use the right amount of solder, both on the iron and on the joint. Too little solder on the iron will result in poor heat transfer to the joint, too much and you will suffer from the solder forming strings as the iron is removed, causing splashes and bridges to other tracks. Too little solder applied to the joint will give the joints a half-finished appearance: a good bond where the soldering iron was, and no solder at all on the other side of the joint. Too much solder applied to the joint will cause bridging when adjacent points are soldered.

Now, with all this in mind, make and test solder joints over and over again on a piece of Veroboard until you are making good solid joints every time. Remember it is much, much more difficult to correct a poorly made joint without damaging something, than it is to make the joint properly in the first place. Anyone can learn to solder properly, it just takes practice.

Keyboard Construction

When you are quite sure that you have mastered the art of soldering you are ready to start the construction of the Matinee. The first parts to be put together are the two keyboards. They are supplied already fixed into a frame so you do not have to worry about getting them set at the right height and offset with respect to one another; the manufacturers have done this for you. Another advantage with the frame is that each keyboard can be hinged up individually on the frame to make construction and servicing really easy.

There are two identical printed circuit boards; one for each manual. Forty-nine diodes, one for each key, are mounted on each PCB. Place the diodes in the positions marked on the PCB taking the wires through the holes, but do not cut off any of the leads yet. Long white lines are marked close to one edge of the PCB and this is the bus-bar edge. The bodies of the diodes should be positioned close to this edge of the PCB and not centrally between the holes for the diode's leads.

One end of each diode has a ring printed around it. This end is the cathode and corresponds to the thickened end in the drawing on the PCB. All the diodes must be placed the right way round, that is with the cathode farthest from the bus-bar edge of the PCB. Now take the wire from the other end of the diode (the anode) and push it back up through the next nearest hole on the PCB as shown in Figure 3. Make sure there is at least 1cm now sticking up above the PCB. Now solder all the cathodes and cut off the excess wire. Pull the lead from the anode fairly tight so that the loop under the PCB is as small as possible and then cut each wire so that it protrudes above the surface by about 1cm.

Take one of the contact springs and place it on one of the diode wires. Slide the wire in from the belled out end of the spring. Bend the wire and spring forward a little, towards the bus-bar side of the PCB and solder the wire to the spring. Place the iron on the bell of the spring for two or three seconds then touch the iron to the diode wire as well and apply the solder at the same time. The solder should run up the wire and fill the bell in the spring. Repeat this for all forty-nine diodes on both boards then bend them all forwards across the bus-bar lines on the PCB so that they lay flat on the PCB. Finally solder the loops on the coppered side of the PCB on each of the diodes.

Nine bus-bars are required, each about 18cm long. On eight of them slide two bus-bar support blocks. The bars should pass through the hole farthest from the mounting lug on the block. Now bend the bars as shown in Figure 4. Push the ends of the bars into the PCB at the same time pushing the lug on the bus-bar support blocks through the hole drilled out for them on the PCB. Melt over the ends of the lugs with a hot clean soldering iron, then clean the iron and solder the bus-bars ends to the PCB. Cut the ninth bus-bar to make the contact loop required for the forty-ninth note on each keyboard and bend it as shown in Figure 4 and solder to the PCB, keeping it at the same height from the PCB as the others. Note that the springs should all now be under the bus-bars.

There are two seventeen-way PCB-mounting plugs and one must be soldered to each PCB. The PCB destined to be fitted under the upper manual should have its plug fitted in the position marked 'solo' on the PCB and the PCB to be fitted under the lower manual should have its plug fitted in the position marked 'accompaniment' on the PCB. Make sure that you fit the plug the right way round so that the locking lugs are on the same side as the lugs printed on the PCB. This will ensure that when the ready-made connecting cable is plugged in, it can only be the right way round. The other end of this cable will be plugged onto the main PCB. We shall describe construction of the main PCB in Part 3.

Press fit the five spacers to each keyboard PCB and mount the PCB under the appropriate keyboard using self-tappers (No. 4 x ½in.) through the spacers into the pre-drilled holes in the plastic frame of the keyboard. Carefully take the loose end of each spring and push it through the hole in each plunger under the keys. Finally check that when the keys are at rest none of the springs are touching the bus-bars and that each contact spring touches the bus-bar as each key is pressed. This completes the construction of the keyboards.

Keyboard Operation

The seventeen wires to the keyboard are connected in a 5 x 12 matrix. All five C's on the keyboard are linked together, all four D's are connected together and so on for the twelve notes. Each of the five separate bus-bars are connected with the other twelve wires to the M108 matrix inputs. Thus the M108 detects which keys are pressed because each key makes a unique connection between one of the twelve key wires and one of the five octave wires. However, if two C's for example were pressed simultaneously the two octave bars involved would be shorted together and if any other key in these two octaves were played the M108 would not be able to tell which of the two octaves the key was in. Therefore a diode is connected in series with each key contact so that the octave bars cannot be shorted together. In addition the inputs to the M108 for the octave bars require pull-up resistors and these are found on the main circuit board.

The M108 is capable of being switched to operate in one of two modes called the 'single' mode and the 'split' mode. In order to produce a fairly sophisticated instrument, in the Matinee both keyboards have their own M108 although the IC was designed originally as a single-chip organ. In the Matinee, on the lower manual it performs its dual function, but on the upper manual, which we shall look at first, it is locked permanently into the 'single' mode.

The M108 sends a signal to the twelve key lines in turn (pins 21 to 32) and looks on the five octave lines (pins 33 to 37) to see if any of the signals are returned. If one or more keys are played on the keyboard then a signal on that key line will be detected on the appropriate octave line and the internal keyboard decoder will allow the appropriate set of three frequencies to be switched to the three octave related outputs (pins 4, 5 and 6 for the lowest octave and pins 16, 17 and 18 for any other note).

The matrix scanning frequency is set by the frequency connected to pin 40 on the IC whilst the output frequencies are determined by the frequency connected to pin 39 on the IC. In the Matinee these pins are connected together and the same master frequency used for both. This is divided down to provide the output frequencies and the chip can generate 85 different notes (all at once as well if you could stand the noise!). When any key is pressed three output frequencies appear, one on pin 6 or 16, one on pin 5 or 17 exactly one octave higher and one on pin 4 or 18 exactly one octave higher than that.

Since the octave bar input to pin 33 on the M108 only ever has top C (from matrix output pin 32) connected to it, eight of the other eleven possibilities (pins 21 to 31) are used to allow various options. The options on pins 24, 27, 28, 30 and 31 are inoperative if the IC is permanently locked in the 'single' mode as is the case with the upper manual, and pins 21 to 23 have no function when linked to pin 33. Each of the eight possibilities has one function when not connected and another function when linked. There must be a diode in the link as they work in the same way as the key contacts.

On the upper manual pin 25 is not connected and this enables an anti-bounce circuit that stops noise and switching transients from the key contacts triggering the IC for a few milliseconds after any key is pressed. Pin 26 is not connected and this locks the IC into the 'single' mode. Finally pin 29 is connected via D143 to pin 33. This connection switches a sustain function on. If it were not connected then after playing, when the last key was released, the frequency outputs from the M108 would cease immediately. Since we are generating our own decay envelopes, we require the frequency outputs from the M108 to remain on. If this pin is connected, the frequency outputs present just prior to all keys being released, remain on until a new key or keys are pressed.

Most of the remainder of the functions of the M108 (on pins 3, 7, 8, 9, 10, 11, 12 and 14) are concerned with the 'split' mode of operation and will be described when we look at the lower manual in Part 4. This leaves six pins. Pin 1 is connected to -6V and pin 20 is connected to +6V. Pin 19 is used by the manufacturer during testing, but in use it has no function and must be linked to pin 20.

Pin 15 provides a continuous low (-6V) DC level all the time any key or keys are played and a continuous high (+6V) DC level when no keys are played. This signal is called KPS (key pressed solo). Pin 13 (TDS) provides a trigger pulse whenever a key is pressed, but this facility is not used in the Matinee. Finally pin 2 is used to reset the logic in the IC when power is first applied. A short positive going pulse is required here and it is provided by C208 and R379.

Organ Voice Circuits

The two 4 foot outputs are connected to each of the voicing filters. On the flute filter the lower octave output passes through a different part of the filter from the rest of the keyboard in order to maintain the quality of the flute sound, but on the other voices the two outputs are simply mixed together and fed through the one filter. The same applies to the 8 foot and 16 foot pairs of outputs. At the inputs to the filters the square wave from the M108 is present, though of course this will be a complex waveform if more than one key is pressed. The square waves are perfectly symmetrical about 0V referenced by 1k resistors.

The flute filters convert the square waves to sine waves which are very much like the waves produced by a real flute. The string filters, after the original square waves have first been amplified by an op-amp, produce short pulses that are rich in harmonics, like the waves produced by a bowed string. The characteristic clarinet sound is produced by a resonant circuit in the clarinet filter. Finally, the cello sound is produced by mixing and filtering the 16 foot output from the M108 with the outputs of the op-amps in the 4 foot and 8 foot strings before filtering.

The output of each filter is connected to its drawbar which acts as a volume control. The sliders of the flutes and clarinet are resistively mixed, pass through L4 for final voice improvement and then go to IC42a; the flute mixer. The sliders of the strings and cello are resistively mixed and go to IC42b; the string mixer. The output of IC42a goes to IC43b whilst the output of IC42b goes to IC43a. IC43 is a dual transconductance op-amp connected here to operate as two independent voltage controlled amplifiers.

The percussion is simply a 4 foot flute voice and the banjo envelope (TR44) is used again as it produces a very fast attack and decay to produce a 'plink'.

The harpsichord voice is obtained by mixing and filtering 4 foot string and 8 foot string. The envelope (TR47) has a fast attack and a very fast initial decay, but when the voltage on the capacitor C171 falls below the point at which D82 conducts (set by R523 and R524) the decay continues at a much slower rate under the control of the normal discharge path R521 and D80. If KPS returns high due to all the keys being released during the normal discharge time, a very fast decay is initiated via R539 when TR50 turns on as described in the piano below.

The piano voice is obtained by mixing and filtering 4 foot string, 8 foot string and cello. The envelope (TR49) has a fast attack and like the harpsichord has an initial fast decay until D93 ceases to conduct (at the point set by R543 and R544), then the decay continues via the normal discharge path R536 and D91. If KPS returns high due to all the keys being released during the normal discharge time, a very fast decay is initiated via R538, due to TR50 turning on when KPS went high. The action of TR50 is inhibited by the 'loud pedal' (S36) — this is the glide switch except when the piano stop is pressed — such that the note sustains for a long time when the switch is operated whether the keys are released or not.

The harpsichord, banjo, piano and percussion voices are capacitively coupled to KPS so that only the negative transitions on KPS are detected and the continuous level on KPS does not override the envelope. On all other stops including accordion, KPS is DC coupled and the note sounds continuously as long as the key is pressed.

A negative voltage occurs on pin 15 (KPS) of the M108 when any key or keys are pressed, and TR46, an emitter follower prevents loading of the KPS signal. However, one output is taken directly from KPS and applied to TR45. When KPS goes negative, TR45 switches on and C166 charges via D149 and R505. The value of R505 sets the attack time and the voltage across C166 is the control voltage applied to the control inputs of the two upper manual voltage controlled amplifiers (VCA). When the last key is released TR45 ceases to conduct and C166 discharges via the sustain drawbar RV44, and R504, D148 and R507 which shuts the VCA down slowly depending on the setting of RV44.

Preset Voice Circuits

When any preset voice switch is selected, the output from the envelope shaper is inhibited by connecting an earth to the control voltage line. But if the 'drawbar add' switch is operated, the earth is removed and the organ voices sound again.

The five preset voices (piano, harpsichord, accordion, banjo and percussion) have their own voltage controlled amplifier and each voice has its own individual envelope shaper to control the VCA. The voicing of the preset stops is achieved by simply mixing and further filtering the basic string and flute voices already described.

The envelope shaper with each preset voice is triggered by the KPS signal from the same emitter follower as the other voices and operates in the same way as the flutes and strings envelope shaper. However, there are slight variations in each one to produce the characteristic attack and decay of the instrument concerned.

The accordion voice is achieved by mixing and filtering cello, clarinet and 8 foot string. The envelope (TR48) is designed to give a very slow attack and a fast decay. The slow attack results from having a high value charging resistor: R530.

The banjo voice is produced by mixing and filtering 4 foot flute and 8 foot string and the envelope (TR44) gives a very fast attack. The decay characteristic, however, has two stages; an initial decay that is very fast and a later decay that is much slower. Because there are two series diodes in the discharge path, when the charge on C139 falls to the voltage at which the diodes cease to conduct (about 1.5V for two series diodes), the discharge is considerably slowed giving a final low-level ring to the voice.

The outputs of each of the five preset voices are taken through double-pole double-throw (DPDT) latch switches. One pole switches the tone to the signal input of the effects VCA (IC45a) and one pole switches the output of the envelope shaper to the control input of the effects VCA. The switches are interlocked so that only one preset voice can sound at any one time.

In part 3 we shall describe the remainder of the electronic construction: making the main PCB and making the power unit. We shall also look at the interwiring and describe how the power unit and the pedalboard works.

[Please check the corrections listed in part 6 of this series - where possible, corrections have been applied to the text but in some cases there are circuit diagram amendments.]