Blind Listening Tests & Amplifiers

[mefinnis]

Planet 10... two Statisticians proved beyond any doubt that i had that the only valid result that can be garnered from the ABX test is that 2 DUT are different. If you can't tell a difference the test does not prove or disprove anything.

Actually, this statement is both correct and incorrect (in it's inference). It is also a bit "cake and eat it too" .......... statistics can't prove anything, but they proved this beyond any doubt ???

Rob quite rightly states, "Statistical tests don't prove or disprove anything." ..... this is the correct bit

Statistical testing simply gives us the probability, by chance, of observing a particular set of observations. Literature (especially in my field - medicine) abounds with poorly designed studies and mis-used statistics. Mainly because people are trying to "produce" a result consistent with their preconceived ideas

The statement that a negative result "means nothing" is absolutely incorrect, with the proviso that the study was appropriately powered to answer the question asked. We are referring here to beta-error, or the probability of not finding a difference when one truly exists.

As Rob states in nice simple language, if
1. our question was, can "golden ears hear any difference between these devices"
2. we designed the trail well (technically)
3. we used sufficient numbers of golden ears
4. we repeated the study on different occasions and in different locations

This may not "prove" they cannot hear any difference, however, your "learned opinion" would be worth "jack-****" in comparison to anyone with any sort of scientific background.

Negative studies require a process called "power analysis" prior to starting the study to an agreed level of probability for a negative result to be acceptable. That is, we all agree to the ground rules prior to starting.

cheers
mark

PS: I am not a statistician, but it is a necessary evil of my work ..... else I might believe all sorts of nonsense studies and do horrible things to my patients.

PPS: A group of real statisticians went through an entire year of the leading dozen medical journals and over 70% of studies we underpowered to say anything about a negative result, used the wrong statistics, or had designed the study incorrectly. Reassuring ...... isn't it

[nw_avphile]

Originally posted by analog_sa

I was under the impression that nothing made a difference. Why would more output devices make a difference if the amp is not being overdriven? Or a bigger PS? Such excess can possibly be justified only by a headbanger, right?

$8 for a cap? you're surely kidding if you think you can get anything decent for that. Maybe in the seventies?

Well this probably doesn't need a reply, but I never said amps were all the same with respect to power output, low impedance drive capability, etc. These things are (obviously) determined by the size of the power supply, number of output devices, etc. For some people, too much is never enough while others have more modest power/current requirements. My point was this is an area where spending more money clearly gets you measurable and audible gains.

Given the choice of building a 50 watt amplifier with a limited power supply and exotic audiophile parts, or a 100 watt high current amplifier with "regular" parts, I'd choose the 100 watt high current design and would put it up against the 50 watt version with audiophile parts any day in a blind or null test (I'm assuming the same basic circuit design here).

As for capacitors, I was referring to *poly* signal capacitors like you would use to couple the input circuit. In blind and null tests, I've never seen a $0.30 poly film capacitor perform any differencly than an $8.00 "audiophile" one. If you're talking about the main *electrolytic* power supply capacitors, they can indeed cost more than $8 for large ones.

[mefinnis]

I should clear-up one other small point.

I have not at any point stated person 'X' cannot hear these differences. I have simply stated that DBRCT is the best means of ascertaining if this statement is true ........ it may be, it may not.

The whole point in doing a study well is not to have preconceived ideas as to what the result will be.

Most people arguing here have (I strongly suspect) never placed themselves in a well controlled blinded environment.

mark

[fdegrove]

Hi,

Given the choice of building a 50 watt amplifier with a limited power supply and exotic audiophile parts, or a 100 watt high current amplifier with "regular" parts, I'd choose the 100 watt high current design and would put it up against the 50 watt version with audiophile parts any day in a blind or null test (I'm assuming the same basic circuit design here).

Well, that will very much depend on the speakers you need to drive with the amp.

I'd drive them with a 50W well designed amp with HQ parts if possible and ample PSU.
Any single part used in an amp is important, even something as silly as a fuseholder.

As for capacitors, I was referring to *poly* signal capacitors like you would use to couple the input circuit. In blind and null tests, I've never seen a $0.30 poly film capacitor perform any differencly than an $8.00 "audiophile" one.

So to you all polyprops are the same sounding? Why do you bother to use polyprops in the first place?
You could save money using polyester there...
Obviously you don't need boutique parts at 25+ $ a piece but I'm sure you wouldn't know where to look for the better non-boutique parts either...mind you, they can even be more expensive than boutique caps.

Does it all make a difference? Yes it does, paying attention to detail and knowing what part is good at what really does pay off.
Than again, that kind of experience does not come overnight.

If you're talking about the main *electrolytic* power supply capacitors, they can indeed cost more than $8 for large ones.

Yes, and there are big sonic differences in electrolytics just the same, not all of them sound alike at all.

No wonder Fred is getting desperate.

Cheers,

[SY]

Originally posted by nw_avphile

On a more technical note, my experience with null testing is much of the error is from phase shift at high frequencies. Is this what you're referring to by "front-to-back time" SY? Amplifiers that have flat response for a few octaves beyond 20khz tend to do better with null testing unless you bandwidth limit the null signal.

Well, you get phase shift anytime you have a rolloff. And the rollof itself will indeed affect the null for a couple of octaves below the corner frequency (and the opposite for an LF rolloff). What I also meant to lump in with that vague term I used was the case where there's a constant delay independent of signal.

The argument of whether an amp that does better on a null test is "performing" better is a semantic issue. It is performing better if the design performance goal is a high degree of null on that test. It is irrelevant (within reason) to the question of audible differences; an amp with 60 dB of null across the audio band might well be audibly transparent and thus indistingishable from an amp with 80 dB of null across the audio band.

[fdegrove]

Hi,

Well, you get phase shift anytime you have a rolloff. And the rollof itself will indeed affect the null for a couple of octaves below the corner frequency (and the opposite for an LF rolloff).

Indeed, there is phase lag (running behind) and there is phase lead (running ahead)...

Which is one reason why a CDP, with it's limited bandwith and steep (brickwall like) filtering presents such a big problem...amongst other difficulties.

What I also meant to lump in with that vague term I used was the case where there's a constant delay independent of signal.

Which, in theory at least, should have less of an audible impact since it is delayed in it's entirity.

The argument of whether an amp that does better on a null test is "performing" better is a semantic issue.

There are a number of things the null test does not show, if it was such a perfect test we'd all use it and by now all amps at equal powerrating and passing the null test would be sounding the same as the source, unfortunately that is not the case.

While the test is not a bad one, it sure isn't the only valid one.

Cheers,

[SY]

Which, in theory at least, should have less of an audible impact since it is delayed in it's entirity.

Less? It should be zero. That's the basis of the entire concept of recording.

[fdegrove]

Hi,

constant delay independent of signal.

Sorry, SY...misread that important part.

Cheers,

[Tube_Dude]

Lets put some perspective in all of this...

As i have told i use this test for many years ago...so lets see some numbers....

With a 2 Volts input in all cases the null is:

Frequencys ----------------------1Khz--------------------10KHz

Sony TAF 3ES-------------------2,5mV--------------------20mV

Pioneer 705 R-------------------2mV---------------------18 mV

harman kardon 680------------12mV--------------------22mV

Onkyo 9511---------------------1mV----------------------5mV

Onkyo 9711 Integra-----------1mV----------------------3mV

Kora designe 30(tubes)-------70mV--------------------70mV

Counterpoint NPS 100---------42mV--------------------400mV

Rotel RB 980 BX-----------------2 mV----------------------6mV

Food for thougt!!

[Peter Daniel]

Originally posted by nw_avphile



As for capacitors, I was referring to *poly* signal capacitors like you would use to couple the input circuit. In blind and null tests, I've never seen a $0.30 poly film capacitor perform any differencly than an $8.00 "audiophile" one. If you're talking about the main *electrolytic* power supply capacitors, they can indeed cost more than $8 for large ones.

Are you kidding or you really don't know how much quality "audiophile" caps cost?