At first, I naively set up the top two chambers to supply atoms to the central 
chamber at a rate of 1 per 6 cycles. However, the size constraints of the assembly chamber proved to be very challenging indeed. Finally got a working layout using two bonders and worked up to an output rate of 1 per 41 cycles, but still more than twice as slow as the 1 per 18 I was envisioning. Works well enough for a portable machine, but I shall leave it to greater alchemists to illuminate more efficient methods.