How can a 40 watt amp outshine a 140 watt amp

02-20-11: Schipo
Richard Clark normally has CD source, amplifiers, high quality home audio speakers, and listening environment set up in advance. But if the listener requests, they can substitute whatever source, source material, amplifiers, speakers (even headphones), and listening environment they prefer, within stipulated practical limits.

Does this mean that Richard Clark will set up the test in my own listening room, with my own equipment? I presume the answer is no.

And that is what is wrong with the test.

I believe that the ABX nature of the test is an illusion. To call it an ABX test is to say, among other things, that only one variable changes. That is of course true in one respect, namely that the amp is the only component that is switched during the test. Hence the test appears to involve a single variable change.

But there is another respect in which hundreds of variables have changed the moment you sit down to take the test, namely the TESTING SYSTEM ITSELF is different from your own. That is the reason, I suspect, why no one can pass the test.

By conducting the test with a system that a participant is not extensively familiar with, the participant’s auditory frame of reference is eliminated. Without it, detecting the manipulation of a single variable change is hopeless.

An analogy: If you put a dish I’ve never eaten in front of me and ask me to tell you if it has ingredient X, I may not be able to do so with any reliability greater than chance. That may be true even if I know what ingredient X tastes like. But if the dish is one my wife has cooked once a week for three years, I can instantly tell you if it has a different ingredient. The reason: Familiarity with the dish.

That’s what missing from Richard Clark’s test: familiarity with the testing system (including the listening room).

Bringing one or two familiar components to the test isn’t enough to make the testing system truly familiar, since literally hundreds of variables are still new to the participant, so Richard Clark's accommodations to participants gives the appearance of scientific rigor without actually providing it.

Bryon

Swampwalker: your making sense when you state 12/12 correct, twice in a row is pretty darn remote.I would gather if Clark lowered the numbers 12/12 correct,there would be a better chance of passing the test. The more you play and listen the more confused audio memory/human perception becomes?

Phaelon: I don't know myself but they all seem to fail...I don't think the testing was constructed to mislead but to teach. I remember Arthur Salvatore writing somewhere when he opened a very expensive mono amp, asking why so much for so little in parts.Maybe the ride is more expensive when it comes to high-end? And only deep pockets need apply.

A 40 watt amp can out perform a 140 watt amp in two ways. First, a watt is current into a resistance, and the resistance is the speaker. But the speaker resistance varies with frequency. As the resistance offered by the speaker drops, the amp has to put out more current. If the 40 watt rating is based on 8-ohms, the amp would have to put out 80 watts into 4-ohms and 160 watts into 2-ohms. Only if the 40 watt amp has a capable power supply can that happen. That requires an expensive power supply (bigger transformer, higher quality and capacity of filtering caps). The Pioneer 140 watt may not have that kind of a power supply. It is rated at 140 watts at 8-ohms, but if the speaker load drops to 4-ohms, the amp may not put out 280 watts to drive the load, instead the power supply only puts out, say 200 watts, and at 2-ohms it says "I give up" and just clips.

The second factor is the quality of the output devices. The more expensive transistors are more linear and stable and distort the signal to a far lesser degree. High quality output devices in a smaller watt amp will definitely out perform a mass market higher powered amp.

When comparing amps, it is critical to note the rating at 4-ohms and, if able, 2-ohms in addition to the standard 8-ohms. The ability of an amp to put out current to a varying load maintains its sound quality throughout the audio band. But that is what drives up the price of an amp -- the power supply and output devices.

Don't forget, Gs5556 (catchy name, BTW!) that the amps under test are NOT run to clipping and are carefully level matched. In real world use, your objection is doubtless correct, but for purposes of this test? I'm less clear.

I think all the evidence needed is in front of us.

First, Carver pretty much proved amps sound different. He had to null his amp to the reference amp to make them indistinguishable. As far as I'm concerned that is 'game set and match' for the 'all amps sound alike' school.

Second, That null was valid ONLY for a particular speaker, though probably close on quite a few others.
The reason? Not only impedance but phase.
Try the same pair of 'nulled' amps on 3 speakers.... 1. A full range, single driver 2. Some Maggies 3. Some B&W from the '8' series.
I'll bet the null doesn't survive all speakers.

Now, keep in mind that the Clark test under discussion specifies NOT clipping the amp(s) under test. A 40 watt amp driving 83 db speakers with an impedance dip to 3 ohms at some wacky phase angle will almost surely clip. I'd be surprised if it didn't, if the level were above 'low'.

I would conclude that using a benign speaker load that it may very well BE impossible to distinguish 2 amps. Stereophile thought they could make such distinctions when challenging Carver. I'd be curious to know if Carver ALSO knew the speaker or had one of the test speakers with which to do his adjustments?

There are many subtle cues to telling gear under A/B tests apart. Even the best poker player can have a 'tell'. I'd suggest that the full test rules have a clue. Perhaps the exact level matching?