II. EXPERIMENTAL TECHNIQUES
The arrangement used to measure the friction force on a bouncing ball is shown in Fig. 1. A wood block was supported on two cylindrical rollers so that the block could move freely in the horizontal direction with almost no frictional resistance. A ball incident obliquely on the block bounces off the block at an oblique angle, exerting a timevarying force on the block only during the brief period of the impact. For a tennis ball, the duration of the impact is typically about 5 ms. The force can be resolved into components perpendicular and parallel to the wood surface. The perpendicular component is equal in magnitude to the normal reaction force on the ball, and the horizontal component is equal in magnitude to the friction force on the ball. The horizontal component causes the block to accelerate in the horizontal direction. A 19-mm-diam, 0.3-mm-thick ceramic piezo disk was fastened to one end of the block as a simple and inex pensive ~$2! accelerometer. Piezo disks of this type can be extracted from piezo buzzers or musical greeting cards. The output voltage from the piezo disk is directly proportional to the acceleration of the block, thus providing a direct measurement of the time-varying friction force acting on the bottom of the ball. The only signal processing required was to connect an external 20-nF capacitor in parallel with the piezo disk to extend the time constant well beyond the duration of the impact. If the output signal is monitored with a high impedance probe ~for example, a standard 10-MV voltage probe!, then the output signal will provide a reliable measure of the friction force for times up to 20 ms or more. Ideally, we would like to measure the friction force acting on a block of effectively infinite mass, but such a measurement would be more difficult. In any case, there is no fundamental difference between the bounce off a moving or accelerating surface and the bounce off a surface that remains at rest. The horizontal speed and spin of the ball may be altered by the motion of the block as described in Sec. VI, but these effects can be minimized by using a block that is much heavier than the ball. Also, there is no effect at all if the ball slides on the block throughout the bounce. For the low speed bounces studied in this experiment, a block mass of 340 g ~about six times the mass of a tennis ball! was used to obtain an adequate output signal from the accelerometer. A block with a larger mass could be used if a more sensitive accelerometer were available or if one wished to study high speed collisions. A ball dropped vertically onto the wood block gave zero output from the piezo disk, demonstrating that the piezo disk responded only to acceleration of the block in the horizontal direction. However, to achieve this result it was necessary to provide some vibration isolation by fastening the piezo disk to a block of rubber attached to one end of the wood block. Two large area ceramic piezo blocks were mounted on top of the wood block so that both the vertical and horizontal components of the force on the wood block could be determined simultaneously. Each of the large piezo blocks was 51 mm square and 4 mm thick. They were connected electrically in parallel to act as a single, large surface area ~51 mm3102 mm! force plate. The upper surface of the plate was mechanically protected using a 0.3-mm-thick circuit board attached directly to the plate with double-sided adhesive tape. This surface is quite smooth and is referred to below as the low friction surface. For some experiments, fine grade ~P800! emery paper was taped firmly to the circuit board to study the bounce off a surface with a higher coefficient of friction. It is referred to in the following as the high friction surface. Bounces at several different angles on both surfaces were filmed at 100 frames/s with a digital video camera to determine the ball speed, spin, and angle before and after each bounce. Video clips were transferred in real time via a firewire connection to an iBook computer for analysis. The horizontal and vertical distance scales were calibrated simply by measuring the actual and image diameter of each ball because any other object or scale would necessarily lie in a different plane and introduce parallax errors.