Usually, the impact interaction between the components in the pin--bit--rock system is only considered for the phase of bit introduction in the rock, since the effectiveness of rock breakdown is of primary interest. However, in impact drilling machines, the end of the introduction phase is the beginning of the preparation for the next impact, and the postimpact behavior of the striking pin and the bit plays a significant role in this preparation. In [I], the recoil rate of the striking pin was estimated in a model in which neither the pin nor the bit consisted of stages. In actual systems, the striking pin and bit axe more complex in structure; furthermore, the motive force of the drive acts on the pin during impact. These factors may markedly influence the process, and therefore k is expedient to develop a more complete theoretical model and to examine the recoil not only of the pin but also of the bit. The inverse impact of the bit on the frame of the drilling machine has a significant influence on its vibrational characteristic [2] and also affects its working life. The basic assumptions and computer programs from [3] will be adopted in the discussion. First, consider the results obtained for three impact systems of pneumatic hammer drills with different types of striking pin. Their basic characteristics are given in Figs. 1 and 2. In Fig. 3, displacement curves for the ends of the pin Xp(t) and the bit Xb(t) are shown for the idealized case when the action of the motive force on the pin is disregarded (F = 0). The energy of the impact and the type of rock is the same for all three systems: A = 63 J; the reaction of the rock is taken from [4] for a strength value f = 12. As is evident, these systems behave in markedly different ways. Major differences are the time at which the inverse pin motion begins and the influence of the rock reaction on this motion. The thin pin (type 3) is driven back by the wave reflected from the impact end of the bit. The pin of type 2 begins to move away from the bit considerably later. In this case, the recoil occurs under the action of the wave reflected from the stage of the bit formed by the cutting edge. The pin recoil coefficient is 0.081 in system 3 and 0.12 in system 2. In both cases, the reaction of the rock has no influence on the inverse pin motion. For pin 1, the situation is different. Initially, as in system 2, pin 1 is driven back by the wave reflected from the cutting edge of the bit; the recoil coefficient is 0.11. Then, after 1.1 msec, a series of interactions between the pin and bit begins, and the rock reaction reaches the pin through the bit. As a result, the rate of recoil is almost tripled -- from 0.86 to 2.4 m/sec; the recoil coefficient increases to 0.32. Of course, repeated interaction between the pin and bit does not occur in the case of low rock strength and elasticity; the bit cannot overtake the pin, and the recoil coefficient remains at 0.11. This situation is observed if the angle a 1 in the force--insertion diagram (Fig. 2) is simply reduced to a level corresponding to a rock strength f= 10. One feature of the series of pin--bit interactions is noteworthy. In 0.66 msec, there are five contacts between the impact ends, in which the potential energy of the elastic deformations remaining in the bit after its interaction with the rock is partially transferred to the pin and increases the kinetic energy of its inverse motion. As is clearly evident in Fig. 3, after the recoil at t = 0.63 msec and prior to the next interaction, the pin is in an excited state. At the next interaction with the bit, the wave processes in the pin are damped, and its potential energy, which, of course, is an order of magnitude less than that in the bit, is consumed in increasing its speed. Thus, we are dealing with the collision of excited rods. Within the framework of the stereomechanical model of impact, this interaction may be represented as a collision with the instantaneous injection of energy into the system from outside; in this case, the velocity recovery coefficient may be greater than 1. In fact, it is 1.45 for the pin and bit. It is expedient to use the simplest model of impact in investigating the dynamics of hammer drills, which may be described by a system of nonlinear