Technique for reducing variance in transducer array pulse-echo response

There is widespread interest in finding low cost approaches for circumventing the problems caused by variations in transducer performance due to device material and geometry variations. In this work, deficiencies in the response were estimated and used for the design of transmitter waveforms or receive filters that are programmed to compensate, at least in part, for the deficiencies. Fortunately, many modern 'high end' systems have programmable waveform generators in the transmit signal path and/or a versatile digital filter in the receive signal path. In the approach used here, power absorbed (Re(VI*)) is measured and used as a surrogate for a measure of pulse-echo response since it is arguably easier to measure. This procedure was investigated using a 1D thickness mode transducer model that included two matching layers that were allowed to vary randomly over a range /spl plusmn/15% (approximately equivalent to approximately /spl plusmn/20 microns). In a simulation run involving 100 random designs, the spectral performance of the corrected transducer was observed to be much closer to desired spectral performance than that for the original uncorrected case. The mean deviation of the pulse-echo response measured over the 20 dB bandwidth was reduced to approximately one third of the value obtained without using the compensation approach. The -20 dB time domain pulse duration was also reduced by approximately 10%.