The Harmoniser must be one of the most fascinating and versatile electro-music effects units available today. Unfortunately, harmonisers are traditionally expensive; prohibitively so to all but the more wealthy musicians and groups. The E&MM TRANSPOZER breaks the price barrier, bringing harmonising power within the reach of the average electro-musician or group. No facilities have been sacrificed to achieve a reasonable cost, and the specifications are by no means modest.

The prime function of the TRANSPOZER is to shift the pitch of the input signal by an adjustable musical interval, allowing for instance a vocalist to have his or her own automatic harmony accompaniment. The harmony voice will faithfully reproduce all the tonal and dynamic characteristics of the singers voice, and word perfect!

The basic principle of harmonising is to write digital representations of the analogue input signal at a constant rate into a store, then to read this digital data from the store at a variable rate, converting it back to its original analogue form to produce a pitch shifted output. Thus if the data is read at twice the write rate, then all the frequencies present in the input signal will be doubled, or in other words raised in pitch by one octave. This process is analogous to playing a tape on a tape recorder at twice the speed that the tape was recorded at. The big difference is that the tape recorder cannot perform this process in real time. Also the harmoniser, unlike the tape recorder, maintains the original tempo. The E&MM TRANSPOZER provides pitch shifts of up to one octave up or down, and any musical (or non musical) interval in between, all in real time. Setting the interval is made easy since apart from the main shift control, a fine shift control provides a trim capability of half a semitone up or down for very precise setting.

The TRANSPOZER is capable of changing pitch in real time since it can effectively "record" and "play" at the same time, but at different speeds. It only works on short sections of the input signal at a time, storing a section, using it for reading for a while, then storing the next section in its place and so on. Obviously if the data is read at twice the rate that it is being written, then sooner or later the data is going to run out. In this case each section of data is read twice. Similarly during downward shifts not all the data is read, so excess data is simply discarded.

The storage capability is used to full advantage by the incorporation of a selectable delay mode, where the data is read at the same speed as it is written, but after a delay. This mode allows many delay related effects such as echo, reverb, ADT (automatic double tracking) etc to be achieved. The delay time is switchable from 6 milliseconds to 200 milliseconds. The delay time switch also controls the section length used in the shift mode, which can be optimised for the sound source being used.

The TRANSPOZER is by no means limited to just guitars and vocals. It produces excellent results when used with any musical instrument or sound source, mono or polyphonic.

Architecture

As in any digitally based audio processing equipment, the heart of the unit is the analogue-to-digital converter (ADC) and complementary digital-to-analogue converter (DAC). The number of digital bits necessary to represent the audio signal is usually considered to be at least 12 to achieve a useful dynamic range. However, a special DAC has been used in the E&MM TRANSPOZER to squeeze a very useful 72dB dynamic range into just 8 bits. The DAC has companding characteristics providing much finer resolution at low signal levels than at high signal levels, where the resolution is less critical. The companding characteristics are achieved by the use of a piecewise approximation to a logarithmic response. One of two modes can be selected; encode or the complementary anti-log decode. Having only 8 bits means that an 8 bit data bus and more importantly an 8 bit wide memory can be used, which reduces the cost of the project.

The block schematic diagram, Figure 1, reveals that there is no ADC as such. The DAC is used in conjunction with a few other devices to perform analogue-to-digital conversion using a technique called 'Successive Approximation'. This involves building up the digital word, bit by bit in a Successive Approximation Register (SAR) and after each bit, comparing the resulting DAC output with the signal input voltage using a fast comparator to decide whether the next bit should be set or not. The binary word is thus assembled in 8 cycles. The data is clocked through the register at a cycle rate of 600kHz, each cycle making the binary word more closely resemble the input voltage. When an analogue-to-digital conversion is complete, the data assembled in the SAR is placed on the data bus ready to be written into the store.

The write sample and hold freezes the input voltage, presenting a steady voltage to the ADC during the conversion time under the control of the write clock. The write clock is responsible for sequencing many actions within the TRANSPOZER, such as starting conversion, store control and read control. The sample and hold captures snatches of voltage from the DAC, converted from data in the store in between write conversions, holding the voltage steady until the next snatch is received. The read filter then reconstitutes this stepped analogue signal to a smooth clean form, free from clock frequency. The input filter is an anti-aliasing filter which is required to attenuate any high frequency harmonics on the input which could otherwise cause intermodulation with the clock frequency.

An overload detection circuit is provided giving LED indication of signal peaks so that the input sensitivity can be adjusted for maximum dynamic range. The treated signal can be re-circulated and hence retreated by mixing it with the amplified input signal, the composite signal being passed to the anti-aliasing filter. The treated signal is also separately mixed with the amplified input signal to produce a variable dry/treated output mix.

The memory address in which the write data is placed is determined by the write address counter, which is incremented by the write clock once for every write conversion. The write clock runs at a little over 20kHz in the delay mode. This is derived from sampling theory which tells us that the audio signal must be re-evaluated at a rate which is at least twice the signal bandwidth. To allow for filter roll-offs and component tolerances, the signal bandwidth is in fact somewhat less than 10kHz. In the shift mode however, the write clock is changed to about 40kHz to allow for a one octave down shift which, during reading effectively halves the sample rate. Similarly, the read address is determined by the read address counter, which is incremented by the read clock, the read clock can be varied between 20kHz and 80kHz either using the pitch controls or by applying an external CV.

Reading and writing are performed alternately, the address source being selected by the address multiplexer. In the delay mode however, the write address is permanently selected so that reading is done at the same location as writing. However, reading is done before writing in each new location so that the oldest data is read, resulting in a delay equal to the time taken to access every store location. It is possible to "freeze" the contents of the store at any time by using an external repeat footswitch. This locks the store in the read mode, preventing any of the data from being changed. Short passages can thus be played and repeated indefinitely until the footswitch is released.

Construction

Construction is very straightforward since almost all components are mounted on a single, double sided PCB, which virtually eliminates wiring errors. The component overlay is shown in Figure 2.

The main requirement for construction is an ability for fine soldering, particularly in the area of digital IC's on the PCB. We recommend a soldering iron with a bit of about 1mm and the use of high grade 22swg solder (the latter being supplied in the kit). Remember when using a fine bit to leave it in contact with the component lead a fraction longer than normal to ensure proper melting when the solder is applied to the lead (not the bit!). This is particularly important where there is a large area of foil which acts as a heatsink.

Construction should proceed as follows — Step 1. To reduce cost the double sided PCB utilises 'track pins' to connect tracks on the two sides. This method is very simple and saves soldering components on both sides as well as eliminating wire links. Place the PCB component side uppermost (E&MM TRANSPOZER in top right hand corner) onto the smooth side of some corrugated cardboard somewhat larger than the PCB. The track pins are in lengths of about 50 pins and the procedure is simply to push the next exposed pin into a hole and gently bend the stick of pins which will cause the inserted pin to snap apart. Now press the pin firmly into the hole with a convenient implement (e.g., the tip of a small pair of pliers) so that is will not jump out when other pins are inserted. It is best to insert about twenty pins before soldering them on the exposed top side. When soldering, ensure that the pin is soldered to the track. Track pins are inserted into every hole in the tinned tracks which are on the component side. When all are soldered into place then turn the PCB over and solder the pins protruding through the board. Next carefully inspect, preferably with a magnifying glass, the top side of the PCB to ensure that joins have been properly made and that no solder bridges have been formed between tracks. If you are in the habit of removing excess solder flux from PCB's, which is good practice for all projects as well as greatly aiding inspection, then the top side of the PCB should be cleaned at this time.

Step 2. Now proceed with insertion of components in the normal manner, that is, mount components in order of increasing height which will more easily allow them to be held in place while being soldered on the reverse side. In the few instances where axial components cross over the foil tracks on the top side ensure that their leads do not touch the tracks. The only real risk is with diode, D1, but this is easily avoided by bending the leads such that the body of the diode is over the tracks. Double check the placement of every component prior to soldering and be particularly careful with the orientation of the bridge rectifier, transistors, regulators, diodes and electrolytic capacitors. Note that there are five 'Veropins' to be installed at positions marked PR1 to PR5 on the component overlay. These pins are primarily intended for connecting the 'Pitch Ratio' display, designed for this project, which is to be published next month.

Step 3. When installing the +5V regulator, IC31, bend its leads carefully to conform with the holes for the leads as well as the mounting hole. During this step slide the heatsink under the IC to ensure that holes are lined up correctly. The regulator need not be insulated from the heatsink but a little heatsink compound would be worthwhile, if available. Bolt the heatsink and regulator to the PCB (nut to be applied from the component side) prior to soldering the IC since this will avoid undue stress on the leads.

Step 4. A PCB mounting rotary switch is used for SW2. Its rotating shaft and wafer securing rods should preferably be shortened prior to attaching the wafer — check length before cutting! Do this carefully and leave the nuts on the rods below the cutting point since they will clean up the thread after sawing. The PCB of the switch faces towards the rear of the main PCB and the order of assembly is: brass spacers, wafer, plastic spacers against the switch PCB, and then the securing nuts. Position the end stop such that the shaft will only rotate through the six positions required and secure with the extra nut provided. Now install it on the PCB and having made sure that it is properly seated solder in place. The latter check also applies to the potentiometers and jack sockets since if they are not properly seated they will not align with the panel holes.

Step 5. All components, except IC's which are later installed into DIL sockets, should now be in place. Close crop (max length 2mms) all leads on the underside of the PCB, remove excess solder flux if possible, and then rigorously check for solder bridges, poor joins and solder splashes. It is usually better to use a desoldering aid to clean up solder bridges and splashes and then re-make the join. Next assemble the case in accordance with the instructions provided and then mount the following components on the front panel: switches SW1 and 3; two 3.5mm jack sockets, J3 and 4; and the two LED's. Tighten the extra nut provided for each of the five potentiometers and offer up the PCB assembly to the panel and lightly secure. Judge the length of wire required to connect up the non-PCB mounted panel components - refer to Figure 3. It will be found easier to remove the panel in order to solder the wires to the PCB although it is not essential. The 0V wires to the centre tap of RV4 and RV5 may be made at this stage. Reassemble and secure panel to PCB with the nuts and shakeproof washers and wire up the panel components by reference to Figure 5. Keep the wires neat and as short as practical.

Step 6. Offer the panel to the case and rest the PCB on the chassis plate. When properly aligned mark the position of the three support holes. Also locate the position of the transformer at the rear of the power supply area and also mark the position for its mounting holes. Drill the chassis plate — 3.5mm for the PCB mounting holes which allows for slight error and 4mms for the transformer. Note that there are two holes near IC20 and IC21 which are for the pitch ratio indicator option. If it is to be installed later then it is advisable to mark these holes on the chassis at the same time as the main PCB mounting holes are marked. If holes of between 9 and 13mms are drilled into the chassis plate at these points it will allow the pitch ratio option to be installed without dismantling the unit. After drilling ensure that all swarf is removed. To mount the PCB insert the 3mm bolts provided from the underside of the chassis plate and secure with two nuts which also act as spacers. Place the PCB, combined with the panel, over the bolts and adjust the spacing nuts, if necessary, in order to make the PCB parallel with the chassis plate while at the same time the panel holes are lined up with the holes in the side frames. Bolt down the PCB and check that no component leads are touching the chassis plate. Now fit the bezel and mains switch to the panel.

Step 7. The mains wiring is the last step. Ensure that the insulators provided (terminal cover for transformer; insulating boot for fuseholder; and terminal covers for mains switch) are used and remember that wires have to be threaded through these protectors prior to soldering. No mains leads must be left exposed. Keep wiring as short as practical. Connect the secondaries of the transformer as shown in Figure 3 and leave just enough wire to reach the screw connector at the top left hand corner of the PCB. Apply solder to the stripped ends of these wires to keep the individual strands together. Now mount the transformer and connect mains earth to the solder tag fitted to one of the securing bolts. Connect the primaries to the PCB connector taking particular note of the 0V input. The base of the box may now be secured and the unit readily accessed from the top ready for checking and final setting up.

Testing and Adjustment

The only equipment needed for testing and setting up is a multimeter and a pair of musical ears! There are only three presets to adjust, none of which are critical to the performance of the unit. When you are sure that the power supply wiring is correct and safe, connect the unit to the mains and switch on. Note that no IC's should be in the DIL sockets at this stage. The +5V LED should come on but check all voltages; a convenient ground is available at the pin PR1 while +5V and -12V are available at PR2 and PR3 respectively. The +12V may be checked at diode D5 or else the top side track adjacent to C36. If there is failure at this stage then check whether regulators are hot which may point to a short on the appropriate power line and also whether there is a voltage into the regulator.

If power supplies are correct then switch off and install the IC's into the DIL sockets. Take the usual precautions with the MOS IC's 15, 16 and 22 to 29. Double check the IC's for location; for pins bent under or outwards and, not least, orientation. As regards the latter all IC's operating from +5V have pin 1 (notch, band, dot, etc.) facing towards the front of the PCB while those powered from +/-12V have pin 1 facing to the right (with the panel towards you). Switch on and if the +5V LED does not illuminate then switch off immediately. As soon as possible check all the voltages again and if they are not close to specified then switch off. Should, you need to switch off then quickly run a finger over the IC's (after switching off!) to determine whether any are unduly hot which may pinpoint the problem area. If there is a failure then you will have to re-check what should have been done thoroughly before, namely, orientation of IC's or a short circuit between power pins and adjacent pins.

If the above is satisfactory then plug in your musical instrument, or other sound source, to the input of the E&MM TRANSPOZER and its output to your amplifier. Set feedback and mix controls fully anticlockwise; pitch controls to centre off; rotary switch to '1' and put SW1 in the delay mode. Adjust the level control such that the loudest input just causes the peak LED to flicker slightly. The sound from your instrument should now be heard from your amplifier as it is passing along the 'dry' signal path. Now turn the mix control fully clockwise and the signal passes through the ADC/DAC filter system and should be reproduced albeit with slightly reduced treble due to filtering.

Next put the rotary switch in position 6 and mix control anticlockwise. Advancing the mix control to its central position should produce a signal comprising the original sound followed by a similar but delayed signal. If the feedback control is advanced an echo effect should be produced. Oscillation may occur as the feedback is rotated but do not worry about this and set the control to a point free from such an effect. Changing the delay setting will affect the repeat speed, ie, the amount of delay. The feedback adjustment may now be made. Put rotary switch in position 4, mix central and feedback fully clockwise and then adjust RV6 (located at rear of feedback pot) until the feedback starts to reduce in level. Continue turning back very slowly until the feedback decays to zero without any adjustment. This ensures that 'runaway feedback' will not occur at any control setting. If, however, you feel that runaway feedback may be a useful effect then adjust RV6 so that it just starts with the feedback control fully clockwise.

Now switch to the harmony (shift) mode with the feedback control anti-clockwise, mix central and delay set at 5. Your instrument should now be accompanied by a signal similar in nature but with a different pitch, the latter being varied by the pitch controls. Set the pitch controls to their mechanical/electrical centre off position and adjust RV8 such that the input pitch is unchanged at the output. With the Pitch Ratio Indicator fitted this would be a reading of 1.000. Next turn the wide (coarse) pitch control fully clockwise and adjust RV7 until the input pitch is doubled, or one octave up, at the output (pitch ratio reading 2.000). Return to centre off and if necessary repeat these latter steps. Due to non-linearity of the control it may be found that when the wide pitch control is fully anti-clockwise the pitch input may not be quite halved, or one octave down, (0.500 pitch ratio reading). Turning the narrow (fine) control anticlockwise should bring the unit into range but if not then repeat the setting up steps but set RV7 to give halving of input pitch when the coarse control is fully anti-clockwise.

During the above adjustments switch, SW3, above the external control input should be put off. With the pitch controls at centre off the effect of putting the switch on is to set the initial pitch to about -1 octave such that a 0 to +5V external control voltage will sweep the pitch up two octaves. With the switch off and pitch controls central then a +/-2V5 input voltage will swing the pitch approximately +/-1 octave. The latter is ideal for some synthesisers with keyboards operating from a centre zero and also for use with simple low frequency oscillators. The pitch controls are still effective when an external control voltage is being used and may, therefore, be used to set the sweep into different ranges. Irrespective of the combined level of control voltages the pitch change will not go significantly outside the range of +/-1 octave and excessive voltages to IC14 are avoided by saturation of IC13.

The final check is the freeze (repeat) footswitch. Connect the footswitch (latching push to make type) to the freeze input socket, J4, and select delay mode with delay '6'. It should now be possible to 'freeze' a short passage from your instrument by operating the footswitch immediately after the note is played. The frozen signal will repeat over and over indefinitely until the footswitch is released.

If you are unable to reproduce any of the above effects then check that there are no signs of overheating — the memory IC's will run quite warm. If all seems in order then proceed with any remaining steps and make a careful record of any steps which do not appear to work. The design is such that failure is unlikely so long as the soldering has been carefully done. If you are not familiar with the operation of such circuits then limit your troubleshooting to inspection of component placement and orientation as well as further thorough examination of soldering. Do not poke about in the hope of finding a fault since one can do more harm than good. The most likely outcome is that the above procedure will check out satisfactorily and so the lid may now be fitted to the case.

Next month we will continue with the circuit diagram, circuit description, details of the Pitch Ratio option and practical uses for the Transpozer.

A complete kit for parts for the Transpozer including the case, front panel, PCB and all components is available from E&MM, (Contact Details) at £159.95 including postage, packing and VAT. Please order as Transpozer kit.

The Pitch Ratio Display board kit is also available at £32.95 including postage, packing and VAT. Please order as Transpozer Display Kit.

These units are supplied with the kind co-operation of Digisound Ltd.