Got an empty Washing Up Bottle handy? Then sling it out and get yourself a couple of rubber car mats and a lump of hardboard. Andy Honeybone is going to show you how to build a set of MIDI drum pads. No, seriously.

Drumming was how I started. Equipped with a red plastic 'Beatles' snare, Olympic hi-hat and the arrogance of ignorance I bashed my way through school bands and Status Quo covers. Now, only a few years later (ahem) I find myself full circle with a need to add drum parts to sequenced compositions in a more intuitive way than step-time programming or keyboard tapping will allow. Herewith, and by popular demand, the results of nights huddled over a soldering iron, namely the Making Music Pad to MIDI converter.

The thought behind the project was presented back in issue 10 and I'm pleased that my finger in the air hypothesising has worked out. The spec for the system is expandable according to your own abilities, but the software to be presented will handle two pads and allow a footswitch input to assign alternate note values to the pads. Velocity can be both monitored and a minimum level preset.

You will need: a BBC 'B' micro, Making Music or E&MM MIDI interface, Acornsoft FORTH, a MIDI drum machine, lots of wires and the two pads and pad interfaces to be described here. Got that? Good, then we'll begin.

First, the pads. The transducers are piezo elements originally intended for buzzers. The Maplin FM59P types were used because I was too lazy to solder wires onto the cheaper unmounted variety. I used a mastic compound intended to stick plastic angle around the sides of a bath to afix the transducers to the underside of the pads.

I cut two 8x5½in rectangles of hardboard and two similarly sized pieces of rubber car mat to form the playing surfaces. A clearance hole was drilled at each corner of the rectangles and, sandwiching in a peripheral layer of foam rubber, the 'batters' were screwed onto a chipboard base plate out of which had been cut holes to allow the transducers to be attached from underneath. This 'multi-layer floating head' design was chosen to minimise cross-triggering but the final answer to this particular menace was a software 'voting' system as will be revealed next month. The rubber matting was stuck to the hardboard with further lengths of mastic strip.

The electronics has to be designed within the constraints of a low part-count and five-volt supply. Two LM324 quad op-amps, an LM339 quad comparator and a 4016 quad analogue gate provide enough building blocks to make four trigger interfaces. Each trigger has a digital reset input, a digital 'read me' output and an analogue velocity level. Four such interfaces would therefore completely use up both the joystick and user ports of the Beeb. The footswitch is connected via one of the joystick fire buttons and a normally closed unit such as a Yamaha sustain pedal is required.

The transducer exhibits a high capacitance and a high input impedance buffer is required. The two input clamping diodes offer some protection. No gain was found to be necessary as the transducers were quite 'hot' enough without — especially when activated by a real drummer and I thank my good friend Author Barnett for his suggestions in the development of this project.

The trigger pulse passes through a sensitivity preset and into a modified peak detector. The output of the transducer is an almost classical ADSR shape with the initial spike being very fast. I chose to suppress this spike and concentrate on the bulk of the output which appeared better related to the velocity of the hit. The one kilohm resistor is the spike killer and you may care to experiment with its value to give different 'feels'.

The capacitor holds the slugged peak maximum and a one megohm bleed resistor ensures that leakage currents don't spuriously trigger the interface. An analogue gate is used to reset the capacitor once its value has been read. I used an ordinary bipolar transistor instead of the gate in the prototype and it worked just as well.

In order to signal that a pad has been hit, a comparator is used to detect a velocity voltage above a preset value. An LM339 'proper' comparator was used but, again, the 324 used in the prototype worked just as well so feel free to adapt. The reference voltage to the comparator is provided by the minimum trigger level preset. Don't ever connect the interface to the Beeb unless the power is off. The A/D chip is very prone to static damage.

That should give you enough to get on with and rather than waffle on, I'll use the remaining space to answer a few letters from my postbag. A 'decent' sampling program for a Commodore 64 is needed by L J Haywood of Cannock. Sorry to have to disabuse you but the C64 is best left to sequencing. You only need look at the two grand price tag of the ADAD 16-bit add-on for the Atari to realise that although a computer is an integral part of a sampler, there's a lot more required under the bonnet than a C64.

Electronic music college projects seem popular from your letters although lack of funds/equipment is also a common theme. A guitar chord illustrator was one such project written by R D Jones of Llanelli who asks if I've got any ideas for his HND. Given that a keyboard is available I'd suggest using the computer as a wave table oscillator sound source. The D/A could be a simple resistor network hung on the user port. A useful reference book would be Hal Chamberlin's 'Musical Applications of Microprocessors' (Hayden).

Finally, Matthew O'Keefe needs some sequencing software for his Beeb and E&MM Beeb MIDI interface. I only know of the UMI and EMR packages and I gather that they both require custom interfaces. I have written a real-time sequencer (in FORTH of course) but abandoned the project in favour of a QX7. Having very quickly reached the limitations of the QX, I am thinking of resurrecting the code in the form of routines which will capture merged track data from the QX and allow proper song construction with tempo changes, jamming facilities, etc. I'm also planning to write a simple FB-01 voice editor. You should see both before too long. A wind synth controller would be nice too — no wonder I get headaches.