Blind Listening Tests & Amplifiers

[SY]

I suspect that Sig Linkwitz isn't spending too much time painting goop on knobs or freezing his CDs!

In any case, Fred, the factors you name are well-known in sensory research. Yes, people sometimes do design tests badly- that's why it's silly to take the null results of one test and generalize from it. But when test after test after test, done by various people using various control methodologies, keep piling up one null result after another, at a certain point, one has to say that when it comes to boxes of gain, there don't seem to be any factors lying outside of conventional engineering which can be audibly distinguished.

A question for you; you mentioned the possibility of piezoelectric effects from SiO2 passivations in chips. I'm curious if there's any experimental evidence of that; as far as I'm aware, the lack of symmetry in the amorphous form (which is what results from passivation and CVD) would exclude this as a possibility.

Could that SiO2 somehow crystallize in normal processing and use? Quoting from one of your links:

"Unlike e.g. amorphous silicon, amorphous silicon dioxide will not crystallize upon annealing at normal temperatures. ("Devitrification" -- that is, crystallization -- of quartz furnace tubes used for high-temperature oxidation is sometimes observed after thousands of hours of use at temperatures exceeding 1200 C.) "

[Fred Dieckmann]

"A question for you; you mentioned the possibility of piezoelectric effects from SiO2 passivations in chips. I'm curious if there's any experimental evidence of that; as far as I'm aware, the lack of symmetry in the amorphous form (which is what results from passivation and CVD) would exclude this as a possibility.

From:
http://www.azom.com/details.asp?Arti...Piezoelectrics

Quartz has this property as it has a complicated crystal structure with a low degree of symmetry.

I don't know. I would think it would certainly be much less piezoelectric than crystalline form which exhibits a high degree of piezoelectric effect evidenced by it's use crystal oscillators.

Fred


Dieckmann /not a semiconductor physicist and doesn't even play one on the web

[nw_avphile]

Originally posted by Fred Dieckmann
Yes unless they effected the differential pair that can cause gain variations outside the influence of the negative feedback loop. Before everybody decides that vibration effects in ICs are impossible, consider that silicon dioxide is commonly used as a passivation layer. The piezoelectric properties are known

You bring up a good point. As I said, I haven't ruled it out and both SY and I suggested Peter test it.

Originally posted by Fred Dieckmann
As the voltage signal changes across a resistor the power dissipated (V squared divided by R) changes the temperature and by proportion to the temperature coefficient; the resistor value. That is the resistor value is changing as a function of signal. This is also related to the thermal mass and thermal time constant of the resistor. A similar effect exist for the voltage coefficient.

It seems that two amplifiers identical expect for the resistors tempco will differ in output as a function of these resistor values dynamically changing with signal. Steady state sine waves are probably poor test signals to demonstrate this with and most of the test discussed may have insufficient resolution to test this effect. The mechanism for distortion is there and who is to say that we cannot hear it in a system of sufficient resolution and listeners with sufficient experience.

And who's to say a pilot of a 747 cannot detect some bird droppings on the wing from the way the plane handles? I'm sorry, someone needs to do the math to demonstrate this is anything but a red herring. It's really hard to imagine that the tempco effects (which are really tiny to begin with) combined with any dynamic heating effects (which are going to be small to non-existant depending on how the resistor is sized and its location in the circuit), combined with the effect a change in that resistor's value actually has on the amplifier after feedback, won't be anything but a percentage so small that gamma rays from another galaxy probably have a greater effect on the amplifier. But, if you can come up with some real-world math to show me otherwise, I'll try to keep an open mind.

[Steve Eddy]

Originally posted by mikek
In analogue signal processing, it is more usefull to consider such excitations as a complex voltage, Ve^(jwt)=(Vcoswt+jVsinwt). This aids the determination of steady state response of a network to a sinusoid, because the complex exponential time waveform, e^(jwt) remains in the same form after differentiation or integration. This vastly simplfies analysis...as the real part of the applied voltage, which is cosinusoidal, (Vcoswt), is then used to solve of other quantities such as phase, fourier components, currents, etc..without the inelegant tedium of resorting to trignometric identities...

So? All you're saying here is that an impulse can be described mathematically. Which has never been in dispute as far as I'm aware.

The question, again, is how do you evaluate an amplifier's impulse response using a single, steady state sinusoid as the stimulus?

se

[nw_avphile]

Originally posted by SY
I invite the participation of anyone who wants to help out. Specifically, I'm looking for the donation of some "approved" fashion plugs, resistors, and caps (they can even be painted with Tube-O-Later!); besides the switch and the box, I'll happily provide the Radio Shack grade comparison parts.

Can someone tell me if there's a thread on this topic somewhere? It sounds interesting!

[mikek]

Originally posted by Fred Dieckmann



Hmmmmmmm....... I believe that is why they invented Spice modeling,which is far from tedious and is easy..... easy..... easy.

Nobody's written about it in the AES journal?

....using pspice without understanding its limitations, the underlaying circuit theory, how it arrives at results, and real world ramifications of its short comings amounts to cooking with a blindfold....

[mikek]

Originally posted by Fred Dieckmann



Hmmmmmmm....... I believe that is why they invented Spice modeling,which is far from tedious and is easy..... easy..... easy.

Nobody's written about it in the AES journal?

..... we are yet to have arbitrary waveform generation in pspice......?

[nw_avphile]

Fred, thanks for providing some constructive points that can be addressed.

Originally posted by Fred Dieckmann
I absolutely accept that psychological factors do alter perception!

Well that's good to hear at least.

Originally posted by Fred Dieckmann
1. Reliance on the viewpoints of someone whose experience, motivation, listening skills and test methods I have incomplete knowledge. Some of these factors can influence testing measurements even outside situations with subjective factors.

Who's asking you (or anyone else) to do that? If you're saying you don't trust someone else to set up the test properly, that's a pretty weak argument--especially when you're welcome to set up the test yourself in your own system. You only need another person to assist with the blind switching and record keeping. If you don't trust them, have someone supervise them if you'd like? If I misunderstood the question, please let me know.

Originally posted by Fred Dieckmann
2. Design of a test who's particular outcome is desired to prove an existing exception This should be the biggest red flag for any objective test.

I'm not proposing that. I'm proposing that people design the test to determine if differences exist between two pieces of gear (or capacitors, or whatever). The outcome can go either way. I'm suggesting people use their own systems, their own music, etc. The number of variables changed from how you normally listen now are minimal.

Originally posted by Fred Dieckmann
3. Anxiety, time pressure, social pressure, and unfamiliarity with system under test for the test precipitants.

Well see above relative to the system, as to anxiety, etc, it needn't be much greater than any other non-blind critical listening test. And again, I have to say if the differences or so subtle that having someone else in the room causes you not to hear them, how significant are they? For more on the above, and the issue of time, I quote from post #15 in this thread by Steve Eddy:

Tom Nousaine has set up a number of self-professed golden-eared audiophiles with an ABX comparator in their own home using their own system and allowed to run tests at their own leisure over periods of months or more.

So far, none have been able to statistically discern any differences once basic issues such as distortion, frequency response and level matching have been addressed.

Originally posted by Fred Dieckmann
4. Design and resolution of the test system.

See the answer to point number 2 above. I'm suggesting people do these tests in their own systems. If they can or cannot hear a difference in their own system, with their own ears, with their own music, isn't that what matters most?

Originally posted by Fred Dieckmann
It sound worse, it sounds different, it sounds the same. Many of these depend on break-in time, power line conditions and RFI background (very time of day dependent by the way and discussed in non audio literature).

Well here's where the null test can be really useful. If you can show different null test results at different times of the day, then I'll conceed some of the above MIGHT have an audible effect. But I've run lots of comparison null tests and they never seem to vary by time of day. Our EARS certainly change by time of day and that is a much more likely explanation.

Originally posted by Fred Dieckmann
My biggest problem with a lot of these discussions are the cynical presumptions made about the methodology, motivations, techical abilities, and experience of those who claim to hear differences based on factors outside of mainsteam audio engineering.

I can't speak for everyone, but all I'm suggesting is this: If things people hear cannot continue to be heard when they don't know what they're listening to, they're VERY LIKELY caused by those psychological factors you agreed exist at the top of the post.

If you agree psychological factors are an issue, how do you suggest they be removed for accurate comparisons? If you tell me you have superhuman control of your mind and can ignore them, it will be my turn to laugh.

You appear frustrated by what you call "cynical presumptions" by folks who want objective evidence. I'm frustrated by subjective folks, who against all the objective evidence, against even common sense, insist they can hear things that have been shown time and time again, in a variety of ways, don't exist. Or, in a more general sense, they insist they're somehow exempt from well documented psychological behavior. Worse, they often refuse to participate in a blind test--at least one that can be verified by an unbiased third party. I wonder why?

I'm NOT criticizing your listening skills--I have no idea how skilled you are in that area. I've particpated in blind tests where one listener can pick out a piece of gear with 100% accuracy and someone else in the exact same session cannot manage better than random. Interestingly, this seems especially true with perceptual encoding (i.e. MP3, MD, etc.). Some people are sensitive to throwing away part of the audio signal, and some aren't. But none of this invalidates blind testing. It only matters what YOU can hear (assuming you're building/buying for yourself).

If you're designing a piece of gear to be sold or enjoyed by many, then it makes sense to have many listen to it during the development process. This is true for blind or non-blind listening tests. The blind ones are just much more likely to produce accurate results.

[nw_avphile]

Originally posted by Peter Daniel
Let's assume that somebody takes part in one of those blind tests, yet for some reason it is not performed properly and the individual under test cannot identify the amps. This carries out a psychological effect on him and he starts to believe that all amps should sound the same (and indeed sound the same to him) and spreads the news to everybody he has a chance to.

Well this isn't any different than all the people running around claiming to hear differences that have no objective basis in fact. And the way blind testing works, if you really can't hear anything, for that person at least in that system, it probably means the things being compared really ARE similar enough to be considered to sound identical (in other words a "false null result" would be rare).

If you want to talk about people running around sharing their knowledge, what about all the people who claim to evaluate an amplifier having only heard it in an unknown system? I run into that all the time. For example:

"Yeah, I stopped into Snake Oil Audio for the first time yesterday and listened to the new XA3000 Pass amp and MAN did it sound good!"

How rational is that? Yeah, the SYSTEM they heard may have sounded really good, but HOW ON EARTH did they attribute the good sound to the amplifier??? What about the speakers, do they know exactly how those sound (and when was the last time they heard them)? What about the room? What about the source material? Even most audiophiles will agree all those things are likely to make a much bigger difference than the amplifier.

The above can seemingly only be explained by audiophiles believing they're listening skills are SO superhuman, they're able to somehow exclude the speakers, room, source material, etc. and magically evaluate the sound of the amplifier in the above situation. I'm not saying they all do this, but it happens so often it amazes me.

[SY]

I don't know. I would think it would certainly be much less piezoelectric than crystalline form which exhibits a high degree of piezoelectric effect evidenced by it's use crystal oscillators.

OK, I'll put you out of your misery- it's not piezoelectric at all. That's a good thing, too, since my cats are often perched near my window while I'm playing music and I'd hate to see them electrocuted by the rattling.


SY/ Formerly a semiconductor physicist/chemist