To Doug Schneider,

I’m writing you because I’m very interested in having your opinion on the [Amphion Argon3L] speakers. Do you think that this anechoic measurement justifies the fact that you only measured 85dB sensitivity in your test versus the “official” 88dB from Amphion?

Thank you very much in advance for your insight.

Kind regards,
Bill K.

Sensitivity tells you the output level of a speaker in decibels given a certain input (usually 2.83V input, which translates to 1W if the speaker is an 8-ohm load). In turn, that’s an indicator of how much amplifier power you will need to drive the speaker, but it’s not an indicator of sound quality. The higher the sensitivity, the less power you’ll need to attain a certain SPL.

When I see a 3dB difference between our measurements and a company’s claims, I don’t think too much of it because there can be reasons for the discrepancy, which I’ll explain. Anything more than 3dB, though, can be a cause for alarm because it can be very misleading for the consumer. The most egregious example we ever saw was with the Coincident Speaker Technology Total Eclipse. At the time, the company was saying the sensitivity of that speaker was upwards of 93dB, and they were claiming it to be a “high-sensitivity” design. Some reviewers actually bought into it not realizing it would be next to impossible to get that kind of output with the driver complement they were using. At 86dB, which is what we measured, it’s certainly not that sensitive at all. In fact, it’s about average, which is precisely what you'd expect from that kind of design. Pity the person who bought the high-sensitivity claim and powered those speakers with too small an amp.

A huge discrepancy such as that one can’t be explained away, but small ones can be. One reason for the discrepancy might be that the company’s method for measuring sensitivity is different from ours. We inject 2.83V and then take an average of the output from 300Hz to 3kHz. Most companies will also use 2.83V, but they might choose a discrete frequency point in the audio band and use that for the spec -- 1kHz is common, but some might pick the frequency that’s loudest and go with that. In the case of the Argon3L, that would be about 750Hz, which, if they chose that point, would make it 88dB. Another reason might be the variation you get between a real-room measurement versus one done in an anechoic chamber. The anechoic chamber presents the worst-case scenario for sensitivity because the room adds no gain to the speaker’s output -- it’s the equivalent of putting the speaker in free space, meaning without walls, a floor, or a ceiling. Put that speaker in a real room and you get about 2-3dB of room gain from the boundaries, which, if you add that to our 85dB rating, gives you the 88dB that Amphion claims. If you look closely at manufacturers’ specs, few specify whether the sensitivity was rated in an anechoic chamber or in a live listening room. Frankly, I think they should say which one. The benchmark, though, is the anechoic environment, which is what our measurements provide.

Speaker sensitivity seems like a straightforward topic, but it’s not always as simple as it appears. I hope the information I gave helps. . . . Doug Schneider