A common opinion among qualified audio experts (restated here by me) is that…
All modern audio power amplifiers that present high input impedance (≥ 20 kΩ), very low output impedance (≤ 0.2 Ω), flat power response (20 Hz–20 kHz ± 0.5 dB at rated output), ultra-low distortion (≤ 0.1% THD at rated output), and a virtually inaudible noise floor will sound exactly the same when operated at matched (± 0.25 dB) levels, not clipped, and properly connected to the same load.
Do appreciate that amplifiers meeting the above specifications all sound the same because they share a single common attribute: They are highly accurate. A power amplifier that is accurate is one that faithfully replicates the original input source signal without making any change, loss, or addition. Given this measured assurance of accuracy, why do so many hi-end audiophiles conduct listening trials to detect the audible difference between power amplifiers? Well, here’s my opinion on…
Why listening trials persist, and grief to avoid when conducting a listening trial…
1—Irrelevance: To an overwhelming extent, today’s audiophiles are non-technical subjectivists*. They have no interest (and little understanding) of the measurements that define audio performance, and they don’t expressly seek accuracy. The audiophiles’ primary goal is to identify equipment that renders euphonic sound; i.e., sound that’s subjectively pleasing, and vacuum tube power amps provide more range of choice. This accrues because…
(a) THD at rated output runs > an order of magnitude higher for a tube-type power amp than on a solid-state equivalent, and that difference might be audible.
(b) And because transformer-coupled tube amps can’t approach the optimal low impedance drive of solid-state power amps, where Zout is typically ≤ 0.1Ω. The tube amp’s higher Zout will directly interact with the shifting Zin presented by the loudspeaker/crossover load because both values will then be of similar order. This will color the sound; the speaker’s sonic signature will differ slightly from its natural voice. The result could be deemed either pleasing or displeasing—a subjective option. Of course, the change will reflect degraded source accuracy, but accurate sound was never the audiophile objective.
2—Expectation: Do appreciate that it’s instinctive to assume that different amplifiers will sound different. Indeed, for a great many decades (from the 1949 origin of “hi-fi” until the early 1980s) most power amplifiers did sound different, and their sound was materially affected by load impedance. Of course, few of those power amplifiers met the stringent specifications cited in the opening paragraph, so they were not highly accurate amplifiers, although some might approach that goal when paired with a favorable load. When sampling amplifiers today, some listeners still expect to hear such differences, and some are certain to fulfill their expectation, regardless of the sound. Aural perception is a fickle and fleeting sensation, prone to uncertainty and subject to human frailty; be wary.
3—DC Offset Orphans: Since power amplifier comparisons imply the probability of high output operation, it’s helpful to know (for sure!) that your samples exhibit only nominal DC offset. So measure the amps’ initial DC offsets before running a listening trial. Don’t evaluate some random lemon with anomalous excessive offset. Be certain that your intended samples truly represent the breed. (Note: This need for offset screening applies only to direct-coupled power amps; not to transformer-coupled amps.)
4—Marginal Mavericks: If a comparison trial involves a power amplifier that is just marginally accurate, that margin might prove audible. This generally involves an amplifier with slightly out-of-spec power response, and the character of the aberration might be perceived by a sensitive listener. Lots of audiophiles express personal preference for a particular sort of biased sound (warm/detailed/liquid/analytic, et al.) that marginally differs from source reality. (In many cases they might do well to reassess their speakers’ crossover settings, rather than change their power amp.) Of course, a “maverick” is an outcast, it’s generally not representative of the breed, so another sample of the same model might not sound the same. Instrumented measurement is a useful way to isolate marginal mavericks, and testing power response (frequency response at high power output) is easily done, with basic test equipment.
5—Output Level Accuracy: All listening comparisons must be accomplished at precisely identical output levels. This is best assured by actually measuring the respective outputs at some appropriate (400-800 Hz) fixed test frequency, using a microphone or sound level meter, mounted on a stand, and fixed precisely at the intended listening position. Simply keeping the main volume dial at the same position will not assure identical output because different amplifiers exhibit different voltage gains. Most modern power amps have internal voltage gains that fall between +23 dB and +29 dB, and some amps have rear panel trim pots to independently adjust the left/right input balance. So, (a) review the technical aspects and check the gain spec; (b) inspect and adjust sample amplifiers as needed; then (c) listen as desired, but be aware that you’ll learn very little if the amplifiers are accurate (see opening).
6—Termination Variables: In the course of disconnecting and reconnecting the speaker load to the various amplifiers in a listening trial (and repeating that cycle numerous times—often as quickly as possible), it’s quite likely that there might be some variation in the uniformity of the termination. Indeed, sometimes even different connector types are used (see photo**), with different results. It’s really quite easy, in the course of this repetitive attach/detach process, to introduce some contact inequality, and that variable could cause aural disparity when conducting high power comparisons. To assure optimum signal transfer it’s vital to carefully inspect, clean, reseat, and tighten all terminations, and assure that all parts are in optimum visual alignment every time that they’re mated.
7—Room Acoustics & Comb Filtering: Aural comparison trials require that all acoustic environs be uniform. This includes a precisely fixed position within the listening space. It also means that the occupancy and every furnishing detail must be exactly the same. Even a minor change can materially affect the acoustic reflections and standing wave patterns, and these disruptions, however slight, can alter the sound. Despite every effort to comply, the small listening areas typical in most private homes can still yield appreciable (6 dB) variation over distances as small as four inches and frequencies as low as 200 Hz. This is the unfortunate consequence of comb filtering. (Refer p.101-104 of “The Audio Expert”, by Ethan Winer [Routledge, 2nd edition, 2018], ISBN 978-0-415-78884-7.)
8—Sight Before Sound: In a previous paper (refer “On Evaluating Audio Equipment…”), I cited a significant study (http://www.pnas.org/content/110/36/14580), conducted in 2013, describing the finding that visual cues convey far more impact than any audible evidence. In sum, your eyes will implant a more vivid and persistent impression than anything that you hear, and what you see will determine what you think you hear. As a consequence, any serious listening trial should be administered under blind test conditions. If you know which amplifier is playing, you will be unable to render valid aural judgement. The statistics in support of this finding are persuasive, and the evidence is undeniable. Listening trial choices that were formed prior to observing this guidance should be ignored.
9—Confirmation Bias: When you’re comparing your old power amplifier against some new and costly wonder that you’ve got on temporary loan, it’s very difficult to admit that there might be no audible difference. And in the event that you’ve already purchased that prized new model, I’d say that it’s not possible to declare that both amps sound the same. We’re not robots.
BG (November 16, 2019)
*Among audiophile organizations, the sole exception appears to be the Boston Audio Society, where most members comprehend and respect the value of technical analysis in defining product performance.
**The discrete banana plugs that are shown in the photo are a bit unusual. They have an internal free-floating slug that achieves uniform compression against the (side entry) wire without transferring any twisting strain; see…A TIP: If your amplifier’s rear clearance (to wall) is especially tight, you can shorten these banana plugs (by 5mm) by discarding the knurled end posts and substituting flat point metric set screws (use M8–1.0 x 8mm) to retain the wires.