How to optimize VLSI ROTOR processors

The principle of operation of ROTOR, a novel processor of nuclear signals, has been recently demonstrated. Nevertheless the prototype versions realised so far are affected by limiting factors which do not permit to fully exploit the potentiality of the ROTOR principle. For example presently the shape of the filter weight function is forced to be trapezoidal, which is not optimal for low-rate operation and/or when the 1/f noise of the preamplifier is not negligible. Furthermore a very precise timing must be supplied for proper operation. In this paper we suggest that these limitations can be overcome by (i) shrinking the analog pulses provided by the preamplifier, (ii) supplying them to the reference input of a DAC used as a multiplier, whose digital-input tracks the wanted weight function shape, (iii) integrating the signal at the DAC output. In this way on the one hand the weight function may be shaped so as to match the optimal profile, on the other hand precision is no more required in the gating intervals where the flat top and the shoulders of the weight function lay. Eventually the proposed solution is shown to be less sensitive to the parameter-matching accurateness of the used technology.