Features

In the September 2000 issue I wrote an article about tube amplifiers sounding louder than transistor amps, given the same nominal RMS power output. The measurements showed that the greater perceived loudness was not a subjective phenomenon but objective fact.

However, the evidence produced was based on experiments with a single recorded tambourine stroke, and the measurement was a pure electrical one. Reflecting on the project, I thought it would be interesting to see if it also worked with, for instance, a soprano voice and with massed instruments and singers. I decided not to take a voltage measurement on the loudspeaker input this time, but a sound-pressure measurement of the acoustic speaker output; that is, I would measure what a listener would actually hear.

I set up an omni microphone half a meter in front of my usual listening position. The output of the mike amp was connected to a storage oscilloscope, an Agilent 54622A this time, kindly lent (as before) by Agilent Technologies Netherlands.

The two fragments I used for my measurements were selected from Riccardo Muti's powerful performance of Verdi's La Traviata (EMI CDCB 47538). Near the end of the second part (Introduzione) of Act I, the Marchese sings "Dunque attendi" ("So, listen"), whereupon the entire Philharmonia Orchestra, plus the whole of the Ambrosian Opera Chorus, burst out in "Si, attendi al canTOR!" ("Yes, listen to the poet!"). The second syllable of "canTOR" is one of the loudest passages on the CD, a true fff. A bit later on the CD, still in Act I, is a duet with Alfredo and Violetta (track 5); 1:24 into this track, Renata Scotto (Violetta) hits a high note pretty loudly.

All the graphs shown have a time scale of 100ms/div., so the span of one screen is exactly one second. I used the same three amplifiers as before—a 25W transistor amplifier and two tube models, one a 4W push-pull triode design using EL84s, the other a 9W 300B-based design—into a pair of Audio Note E loudspeakers, and cranked up the volume as high as it could go before Scotto's voice began to distort. (To be sure it wasn't my ears distorting, I wore Etymotic Research earplugs, which give a 12dB noise reduction linearly across the audioband.)

Let's take a look at the soprano first. Fig.1 is the 25W transistor amp (now restored to its original 25W). It reaches almost 42V peak-peak at the output of the mike amp (vertical scale 10V/div.).

Fig.1 25W transistor amplifier plus Audio Note E speakers, soprano from CD 1, track 5 of La Traviata.

Next, in fig.2, is the 300B SE amp, nominally some 10W but reaching over 46V p-p, leaving its 2.5-times-more-powerful solid-state cousin behind.

Fig.2 10W 300B single-ended amplifier plus Audio Note E speakers, soprano from CD 1, track 5 of La Traviata.

The tiny 4W EL84/triode push-pull amplifier managed almost 30V p-p (fig.3), meaning it came within throwing distance of the 25W solid-state amp at only 3dB less loud, against a nominal power difference of 8dB. Both tube amps showed some compression but didn't sound distorted. The transistor amp could not be pushed a dB further, as it immediately started screaming!

Fig.3 4W EL84/triode push-pull amplifier plus Audio Note E speakers, soprano from CD 1, track 5 of La Traviata.