Originally presented at the International Computer Music Conference, Glasgow, 1990.

A META-ACTION FOR THE GRAND PIANO

Alistair M. Riddell
Music Department
Princeton University
Princeton New Jersey 08544
amr@silvertone.princeton.edu

Abstract

This paper outlines a portable high performance computer controlled action for the grand piano. It was designed to be installed in a variety of instruments and, in effect, temporarily replaces the easily removed existing hammer and keyboard assembly. It can function from different computer systems in either a performer/machine interactive or computer only context. The principle features are a high degree of independence between the operation of the hammers and dampers, and the operational speed of these components. Performance sophistication lies in the ability to control these components. Such control has hitherto not been accessible in either traditional instruments or any player piano system.

Origins

The concept of a portable computer controlled action for the grand piano grew out of work with upright pianos which commenced in 1981. Through a series of customized instruments developed around the now defunct Pianocorder system, I was able to frame the idea of what a grand piano action, operating under computer control, should be capable of doing from both an aesthetic and performance perspective. Attention was specifically directed towards the use of the Grand Piano as superior instrument in both sound and functional design.

While inspiration initially came from the music of Conlon Nancarrow, the idea was essentially founded upon a desire to investigate computing technology as a means to achieve greater control and performance nuance. For example, allowing a greater sense of human performance interpretation in compositions that are otherwise impossible to play. Nancarrow's instruments had some obvious technical limitations but more importantly they were not conceived in the age of computer control and performer/machine interaction. But Nancarrow's music, perhaps, reflected a state of performance that was potentially accessible to a performer, in real-time, through a collaboration with the computing technology. Such complex musical structures as those found in his Study #25, for example, might give rise to an interesting piano technique, not to mention compositional direction.

The construction of the action was undertaken through a grant from the Australia Council's Special Projects Unit and the work carried out in the engineering workshop of the Physics Department at La Trobe University in Melbourne between June 1987 and August 1989. Considerable theoretical research into operational aspects of the traditional piano took place between June 1987 and March 1988. It was designed and constructed by Marshall Maclean and the author. Research is continuing in the areas of software development and composition in the Music Department at Princeton University.

Structural Organization and Functional Design

Two particular requirements strongly influenced the design of the action. Given that the keyboard and action chamber in most grand pianos is a complex space, the new action aimed to accommodate any inherent and critical geometric variations. Furthermore, since portability was an important factor, it also had to comprise of convenient subsections for both installation and transportation. Functionally, the action consists of approximately 160 solenoids, one per hammer and one per damper, mounted on 2 parallel but different shaped rails. Both rails are assembled from 3 sections into a seamless unit where all components can slide freely along a continuous surface. The rails are mounted on 5 support brackets which have adjustable feet. These brackets can be moved into any position along the unit rails themselves and the feet can tilt the action forward or backwards a few degrees off the vertical, even raise it slightly if necessary. The action frame was machined from aluminium which gives it an assembled weight of 35 Kg.

On the forward rail, which faces into the instrument, are mounted the hammer solenoids as two staggered rows. The length of the rail can be varied to suit different action widths and the height is adjustable from either the action feet or the individual hammers themselves. The actual hammer is a single moving part consisting of the solenoid `plug' or `core' at one end and the hammer head at the other end of a threaded rod. It is technically capable of operating very rapidly. The hammer tips have a base of aluminium and although vary in size are smaller than conventional hammers. They can be covered in a variety of materials to produce different timbres. The hammer travels approximately 18 mm and weighs slightly more than the corresponding conventional hammer. On both sides of the rail which appears at the front of the instrument are mounted the solenoids that lift the dampers. A damper is raised via a lever which passes under the hammer rail. At most, the hammer/damper assembly consists of 3 moving parts: the solenoid core, the lever and the damper itself. The damper solenoid moves approximately 5 mm and there is considerable adjustment possible at either end of the lever.
Hammer and damper interaction is intended to be either synchronous or asynchronous in operation. Each solenoid is connected to a driver board which is responsible for switching through the power to the solenoid upon receipt of the appropriate signal. The driver boards are in turn connected to a control processor. The hammer solenoids operate at approximately 170 volts while the damper solenoids operate at a significantly lower voltage since they have to remain on for indefinite periods of time.
The hammer solenoids receive power for very brief periods of time, ranging from approximately 10 to 20 milliseconds. This range determines the dynamics. It is approximately between the points where the hammer just reaches the string and where it is in contact with the string for too long, thus interfering with those waves returning along the string.

The requirement that the action should operate in a variety of instruments necessitated a design with considerable adjustment in many areas. While it is relatively easy to remove the traditional action, installation of the Meta-Action is expected to be more difficult. Most of the difficulty is in understanding what needs to be adjusted and fine tuning the operation of the action when in place. If it is to operate efficiently and sustain vigorous movement then fine adjustment to every moving part is critical.

Control and Performance

It was originally envisioned that the action should remain interactive, i.e. operate through performer/machine interaction. Which ever way this might be achieved - perhaps performer information from a commercially available controller is sent to a powerful computer, interpreted, redefined and then the output information passed to the action - performer/machine interaction is currently felt to be one of the most challenging directions in computer music.

In meeting the real-time demands of such an interactive system, the delegation of action control to a Secondary Processing System (SPS) appeared necessary. Two strategies are currently under consideration. The first, involves 2 (or more) stand alone processors, accessed by a serial communications link. The second, requires a memory mapped I/O coprocessor. Communication is via shared memory with the main processor. Either approach should permit complex combinations of hammer/damper operations, e.g. the ability to raise dampers on strings that are not currently being struck or leave them lowered while striking the strings. Also, the careful application of delaying strategies, where the dampers can be removed after the attack or raised and lowered during the natural decay of the sound, has powerful composition and performance implications.

Control of the action is divided on a function basis, since hammer and damper operations are not only quite distinct but vary in control complexity. The former is complicated by the receipt of asynchronous data while managing internal clock time critical events. The later, being less time critical, benefits from being in operation slightly earlier than the onset of the hammer. Thus the dampers can be manipulated freely without impeding any simultaneous hammer operations. If the dampers could not be raised early enough the system would have a severely restricted function. In the above discussion is was indicated that control maybe physically separate, in which case, hammer and damper operation would run independently on different machines.

Two communications protocols expedite data transfer through the system. One between the performer and the main computer (MIDI), and the other between the SPS and the action (proprietary). Since the SPS can be remotely reprogrammed, various protocols could be loaded for different compositions and performances. At this moment, the SPS outputs a 16 bit word to the driver boards. The 16 bits actually map to 16 signal control lines which select decoders, their specific output combinations, and in turn, the actual solenoids. One area of concern is response time through the system. This is easily appreciated as an unwarranted delay between initial action and response but what might be a result of the input should be taken into consideration. The system might, for example, have an acceptable response time for a `one-to-one' event mapping but a `one-to-many' mapping has more disturbing performance ramifications. The very nature of such complex temporal structures implies that they will require their own time in which to exist. Certainly, there comes a point when the performer no longer feels in control of the instrument. However, an interesting question is whether performers can learn to manage a sense of drifting in and out of control of their instrument and indeed, could that become part of the performer/machine interactive experience.

From the above discussion, it is obvious that the action was not designed to necessarily produce a traditional piano sound, nor in fact, to be necessarily played in a conventional manner. For the most part, the piano provides a convenient, abundant and aesthetically entrenched musical vehicle upon which to expedite ideas about humans, machines and art.

Meta-Action Photo Gallery

B/W, overview of piano with action installed - (jpeg 71872)
B/W, front view of installed action - (jpeg 72922)
B/W, close up of installed action - (jpeg 102302)
Early test installation using Chris Mann's Piano
Early test installation using Chris Mann's Piano
Early test installation using Chris Mann's Piano
Under constuction in the Physics department Workshop
Meta-Action near completion
Meta-Action near completion
Meta-Action near completion
Meta-Action near completion
Meta-Action near completion
Flight case
Flight case

References

Riddell, A. M. 1982. The Computer Controlled Piano - New Instrument, New Performer. NMA 1. NMA Publications. Melbourne, Australia.

Riddell, A. M. 1989. Towards a Virtual Piano Action. NMA 6. NMA Publications. Melbourne, Australia.

Riddell, A. M. 1989. A Perspective on the Acoustic Piano as a Performance Medium Under Machine Control. MA Thesis. La Trobe University. Melbourne Australia.