ark (12/20/82)
An article in the March 1980 issue of High Fidelity Magazine discusses an experiment that was done to determine if people could really tell the difference between amplifiers by listening to them. Briefly, the experiment was as follows: Two amplifiers were chosen which would be expected to have considerably different characteristics (a Dynaco 400 power amp, rated at 200 watts per channel, and the power amp section of a 30 w/ch Lafayette receiver). Each amplifier was set up so that it could be driven by the output of the same preamp, but through an equalizer. Each amp had its own equalizer. The equalizers were set so as to give frequency responses that were, as far as it was possible to measure, identical (within 1/4 dB over the entire audio band). While doing this, it was discovered that different amplifiers of the same make and model had measurably different frequency response (1/2 dB or so). The purpose of the equalization was to eliminate, as far as possible, variations stemming from frequency response alone. The amplifiers were then hooked up to speakers through a pair of switches. The switches were designed so that one would switch from one amplifier to the other, and the other would do nothing at all (except interrupt the signal while the switch was being moved). The wires to the switches were scrambled so that not even the experimenter knew in advance which switch was which. Each member of a panel of five "golden-ear" listeners was confronted with this setup. They were told, in effect, the following: We are playing a signal back through one of two amplifiers. One of these switches lets you select which amplifier you are hearing. The other switch does nothing at all. Your problem is to figure out which switch is which. Each juror listened to the tests, and made choices, in isolation. Thus the opinions of one would not effect the others. One juror listened for an hour daily for three weeks, the others for about half that. Three of the jurors believed, before the experiment started, that they would be able to hear differences under the aforementioned conditions. The other two did not. Results: Two of the "believers" heard differences and wrote down thei identiidations throughout the tests. The other believer and the two nonbelieves wrote down identifications at first but later stopped as they decided that they could not be sure of differences. After the tests, the actual setup was decoded and compaed with the notes made by the two people who heard differences. One of the two was right 19 times out of 40 trials, the other was right 20 times out of 40. In other words, once the frequency responses of the two amplifiers were accurately matched, none of the listeners was able to tell them apart. In other words, don't hope to learn anything significant about an amplifier by listening to it. You will only learn something if the amplifier is broken.