Multimode electromechanical transducers and optimum filters for resonant mass gravitational wave antennas

Resonant mass gravitational wave antennas are optimized to detect impulsive signals that result from supernova explosions, but so far no such events have been detected. Expanding the detectable range of signals to include swept-frequency signals (chirps), created in the final stages of the coalescence of a compact binary star system, would increase the probability of detecting gravitational radiation, since binary star coalescence events are predicted to be more frequent than supernova explosions. It is speculated that a few cycles of the chirp would sweep through the sensitive frequency band of the antenna in a few milliseconds, as opposed to the instantaneous antenna interaction time of the impulse. For maximum energy to be extracted from chirp signals, the antenna bandwidth must be increased. This may be achieved by optimizing the mechanical amplifier used in the electromechanical transducer and reducing the broadband noise in the electrical readout circuit. Furthermore, the search for chirps requires new data analysis techniques. The research presented here has focused on these issues.; First, we developed a two-mode torsional mechanical transducer for a 900 Hz antenna. The mechanical system had a bandwidth of 260 Hz and a mechanical amplification factor of 40. Secondly, we designed and tested a prototype superconducting thin-film inductance modulated readout circuit to assess the feasibility of a fully integrated thin-film transducer-SQUID amplifier system.; Finally, we developed optimum filters for chirp signals. We designed matched chirped filter templates for existing resonant mass antennas and tested them on data from the antenna operating at Louisiana State University. Future broadband resonant mass antennas will require an array of chirped match filters, and thus, it will become excessively computationally intensive to apply matched filter arrays to the data sets. To overcome this problem, we developed a fast filtering algorithm based on stochastic resonance that will be useful for an initial check of data for possible signals. For a Brownian motion limited antenna, the stochastic resonance filter gives the same S/N as the matched filter and is a factor of 5 worse for a broadband noise limited antenna.