An inexpensive unit offering two independent channels of powerful percussion. Design by Mark Stuart

Currently, there exists a wide range of electronic percussion synthesizers, which provide almost all types of pulsed sound output. Sweeping tones, phased noise and voltage controlled filter (VCF) effects all have their place, but each additional feature adds to the cost and complexity of the drum machine.

The instrument presented here not only accurately simulates the sounds of acoustic percussion instruments, but also produces a wide range of artificial sounds — without being overcomplicated or expensive. If the drum machine is used with a good PA and some overdrive a very wide range of effects is available.

Outline

Two independent channels are provided, each with five sound controls and an input sensitivity control. Each channel has a piezo electric pick-up drum pad and an external trigger input socket for use with a sequencer. One channel is fitted with an additional circuit which provides a delayed trigger pulse to drive the other channel.

The delay is variable and can also be switched out completely. Then the channels can be used either independently, simultaneously, or 'B after A'. In the 'B after A' mode, channel A is triggered, and channel B automatically follows after a time set by using the delay control. Channel B may also be triggered independently when in this mode.

Sound controls allow adjustment of noise level, noise filter frequency, tone level, tone frequency, and envelope delay. The range of adjustment of the sound controls allows realistic bass drum, snare drum, cymbal, tom tom, and wood block simulation — it also leaves plenty of scope for experimentation.

Low power components are used throughout, resulting in very good battery life from two PP3 9V batteries.

Circuit Blocks

The block diagram in Fig. 1 shows the operating system for just one channel. The only difference between the two channels is that channel A has additional components which provide an optional 'delayed trigger pulse' for channel B.

Each channel can be triggered either by hitting the piezo-electric transducer or by positive pulses from a sequencer via the EXTERNAL TRIGGER socket, SK2. Control VR2 sets the trigger sensitivity.

The trigger amplifier output drives the trigger pulse generator, which has two outputs. One output feeds an envelope generator circuit, which produces the sharp attack/slow decay envelope voltage waveform, characteristic of percussion sounds. The rate of decay is varied by VR6 over a 100:1 range. The other trigger pulse generator output is used only on channel A to drive the delayed trigger pulse generator. VR8 is used to adjust the delay from zero to approximately one second. When not required the delayed trigger pulse is disconnected from channel B by SW2.

A control voltage waveform is fed from the envelope generator to the gain adjusting terminal of the voltage controlled amplifier (VCA). The input to the VCA is a mixture of tone and noise set by VR3 and VR2. VR4 controls the frequency of the triangular tone waveform generated by the oscillator circuit. Tone frequency can be set from below 50Hz up to 10kHz.

The output from a wideband noise generator circuit is passed through a simple 6dB per octave adjustable low pass filter. VR1 sets the turnover frequency of the noise filter. At the highest setting the noise is more suitable for cymbals, and the lower frequency settings for sounds like the 'rattle' from snare drums. The mixture of noise and tone is always present at the input to the VCA. It can only pass to the output when an envelope pulse is applied to the VCA gain adjusting terminal.

Trigger Circuit

The trigger amplifier circuit uses one section of the low power quad operational amplifier, IC2, in a standard inverting amplifier mode. An input signal from a sequencer, via the external trigger socket, or produced when the piezo electric transducer is tapped, passes to the inverting input of IC2c. The gain of IC2c is controlled by negative feedback set by VR5. When this is set to maximum only a small amount of signal is fed back, and the gain is high. Increasing the feedback by reducing the setting of VR5 reduces the gain. Negative half cycles from IC2c pass via diode D1, which acts as a half wave rectifier, producing a smooth negative voltage across R15/C7. As the input signal increases the voltage becomes more and more negative.

IC1 is a quad 2-input NAND Schmitt trigger, with IC1c wired as a simple inverting 'Schmitt trigger'. The operation of this circuit is such that the output (IC1 pin 3) is normally low — it sits at -9V until the input threshold voltage is reached. As soon as the input voltage reaches the threshold voltage the output immediately switches to its high state — in this case 0V.

Each time the input transducer is given a hard enough tap, the output voltage from IC1c rises sharply from —9V to 0V, and then falls sharply back as the tap ends. This pulse is fed to the envelope generator circuit.

Envelope Generator

The input to this part of the circuit is a short positive pulse from IC1c. This pulse is buffered by the emitter follower circuit around Q2. In the absence of trigger pulses Q2 is turned off, and C8 is fully discharged via R20 and VR6. When a trigger pulse arrives the emitter of Q2 'follows' its base voltage, producing a powerful positive pulse which rapidly charges C8. At the end of the trigger pulse the base of Q2 returns to -9V, but because its base-emitter junction also acts as a diode, C8 remains fully charged holding the emitter-base junction of Q2 reverse biased. The voltage on C8 falls steadily as it leaks away via R20 and VR6, which set the decay time of the envelope waveform.

The envelope waveform therefore consists of a fast rising positive edge as C8 is charged, followed by a controlled fall back to zero as the charge leaks away.

Noise Generator

The base-emitter junction of Q1 is used as a noise diode. R2 provides a current of about eight microamps, which seems about optimum for the transistor type used. Not all transistors are equally good when used as noise diodes — it may be necessary to try one or two to get the best effect. The performance is not critical, so it's a lot cheaper to buy several transistors than a noise diode. The noise output from Q1 is amplified by IC2a used as a non-inverting amplifier. This configuration gives a very high input impedance which best matches the low current output from R2/Q1.

VR1 and C4 form a simple R/C low pass filter network with a roll off rate of 6dB per octave, and a turnover frequency adjustable from 15kHz down to 1.5kHz. When VR1 is set to maximum, the shunting effect of C4 is greatest and only the lower frequencies can pass. When VR1 is set to minimum, the shunting effect of C4 is less and only the high frequencies are affected. After the filter, IC2b provides another stage of amplification before the noise signal passes to the noise level control VR2.

Tone Generator

IC1d is used to generate a triangular wave output which is amplified by non-inverting amplifier IC2d, and passed to the tone level control RV3. The operation of IC1d is as a simple inverting Schmitt trigger oscillator. The inverting Schmitt trigger circuit is most easily explained by reference to Fig. 2a. The output can only be at either the negative (low) or positive (high) supply voltage, the output switches vary quickly between these two levels depending on the voltage applied to the input. When the input exceeds the positive input threshold voltage (+V) the output switches from high to low — if the input then falls below +V the output remains low until the input falls to the negative input threshold voltage (-V). At this point the output switches to the high state. Adding a resistor and a capacitor to the circuit produces oscillation, because as soon as the capacitor C_t charges to +V via R_t, the output becomes low and C_t begins to discharge via R_t. C_t charges to +V via R_t, the output becomes low and C_t begins to discharge via R_t. When C_t discharges to -V the output switches to its high state and C_t begins to charge again up to +V. The output then switches low again and the cycle is repeated.

Frequency is determined by the values of C_t and R_t (C10 and VR4 in Fig. 2). A low value of R_t enables C_t to charge and discharge more quickly so producing a higher frequency output. The output voltage waveform from IC1d is a square wave, whilst the voltage on the input is an approximately triangular waveform. To obtain a better sounding tone the circuit uses the triangular waveform. This is coupled through C9 to the non-inverting buffer amplifier stage IC2d. The tone level is set by VR3.

Voltage Controlled Amplifier

The final part of the circuit takes as its input a mixture of noise and tone set by VR2 and VR3, and a gain control current from the envelope generator circuit. IC3 is an operational transconductance amplifier (OTA). It is similar to a normal operational amplifier with inverting and non-inverting inputs, but instead of a voltage output, it is a current output. The gain of the OTA is adjustable over a wide range by means of a control current applied to its bias terminal. In this circuit the control current is derived from the envelope generator output through R21, which converts the envelope generator voltage output into a current. Initially, when the envelope voltage is zero, the OTA does not receive any bias current and so its gain is zero. After the envelope generator is triggered the OTA gain rises sharply to maximum and then gently falls back to zero at a rate determined by the decay control setting. The percussive envelope shape is therefore applied to the noise and tone input, and the result appears at the OTA output. The output current from the OTA is converted to a voltage by load resistor R27, and coupled to the output socket by C12.

Delayed Trigger Generator

IC1a and b form a circuit which produces a second trigger pulse delayed from the initial one by an amount set by VR8. IC1a is connected as the now familiar inverting Schmitt trigger, and IC1b is used as a NAND gate. The NAND gate can be summarised as a circuit which produces a low output voltage only when all its inputs are held high. When any of the inputs is taken low the output will go high. In the absence of a trigger input, one input of IC1b is held high by R15. IC1a output is also held high because its input is held low via R14 and VR8. IC1b therefore has both inputs high, so its output is low. When a trigger pulse is applied the voltage across R15 falls and one input to IC1b is pulled low. IC1b output then switches high and C16 couples this to IC1a. IC1a output becomes low holding the other input of IC1b low. Now even when the trigger pulse is removed, and the first input to IC1b returns to the high state, the other input to IC1b is still held low by IC1a output. The output of IC1b and the positive end of C16 are therefore held high. At first the inputs of IC1a are held high via C16 but slowly C16 charges via R14 and VR8 and the voltage falls to the negative input voltage threshold of IC1a. At this point IC1a output switches back to the high state producing a positive output pulse via C15 to the delay trigger switch SW2, and returning the input of IC1b to the high state. Both inputs to IC1b are now positive and so its output becomes negative, C16 discharges via D3, and the circuit is back to the stable starting condition. The setting of VR8 determines how quickly C16 charges and so adjusts the delay time.

Construction

Each channel uses identical printed circuit boards which hold all of the components except for the jack sockets, delay trigger switch and controls (S2 and VR8), and a piezo electric transducer. The trigger delay components (see parts list) should be omitted from channel B PCB.

Construct the boards first, start with the veropins, wire links, IC sockets, resistors and the small capacitors. Add the large capacitors and the control potentiometers last.

The case specified has a thin aluminium top panel which is easy to drill to the pattern required. When the drilling has been done, mount the two boards to the panel using the potentiometer bushes. Fit the on/off switch SW1 and VR8/S2. The next stage is the wiring — see Fig. 4. The three power supply connections and the output connection are linked between the two boards using tinned copper wire fitted with insulating sleeving. All the other connections should be taken from the boards using connecting pins pushed through from the copper side of the printed circuit boards.

The two sockets are fitted one each side of the case, leaving room between them for the two PP3 batteries to be fitted end to end. Wire lengths are not critical, but try to be neat with the SK1 and SK2 connections.

The piezo electric transducers can be mounted one either side of the front panel on a thin disc of flexible foam, or they can be fitted into practice pads and wired to the external trigger socket via a suitable lead and plug. The foam was found to be essential to prevent vibration from one channel triggering the other channel.

Setting Up

The only present control is VR7. This is an offset null potentiometer for IC3. If this control is not correctly set there will be a DC voltage shift at the output of IC3, this will add an initial 'click' to each beat. To correctly set VR7 the tone level and noise level controls should be set to minimum. Connect the output to an amplifier and repeatedly trigger the channel being adjusted. Set VR7 for the minimum initial click.

In Use

The output level from the drum machine depends upon the setting of the noise and tone level controls. Up to 10 volts output is available — enough to overdrive any PA system. The output of IC3 is completely cut off between beats ensuring that the background noise level is extremely low. With the sensitivity controls set at minimum a good strong tap will trigger each channel. The available range of sounds can be explored by adjusting all of the controls. With VR8/S2 turned fully clockwise a beat should be produced from channel B about one second after channel A has been triggered. With VR8 set to minimum, channel A will trigger Channel B almost simultaneously. Switch off S2 and the channels will operate independently. External triggering, from a sequencer for example, can be applied via the trigger socket — positive pulses are required. Triggering can also be produced by using miniature loudspeakers or similar fitted inside practice pads. There is plenty of scope for experimentation — the sensitivity control should be used to find the ideal response.

Battery life should be very long due to the use of very low power integrated circuits and careful circuit design. Although both outputs are combined in a single mono socket it is quite acceptable to use each one independently, perhaps to achieve a 'stereo' effect.

Finally, for the more ambitious, it is possible to use several drum machines together. If the trigger delay components are fitted to channel B, a delayed trigger pulse can be produced to pass on to another unit simply by adding another delay control. In this way an elaborate sequence of widely different sounds could be set up.