Ok. What about making your own ABX comparator, and testing your equipment in your appartment when you are the only one there? Your audiphile friends will never have to know you can't hear the difference. The only thing that might be a bit stressing is the thought that your conclusion may be that you've paid a few bucks too many for your hifi
I mentioned before that I invited forum-members on a dutch diy audio site to do a DBT at my place. At first there were almost no replies to the thread, but now it seems there ARE a few people interested.
We'll probably test only a few components. I would seem easy to switch between different speakercables and interconnects with the use of a speaker A/B switch and the inputselector on the pre-amp. Is it possible to do instant switching between cd-players? Is there a way to let two identical cd's start at the same moment in two different cd-players? Perhaps press the pausebutton at the same time?
HOw do I do it with amps? Can I get them levelmatched with a simple multi-meter?
We'll probably test only a few components. I would seem easy to switch between different speakercables and interconnects with the use of a speaker A/B switch and the inputselector on the pre-amp. Is it possible to do instant switching between cd-players? Is there a way to let two identical cd's start at the same moment in two different cd-players? Perhaps press the pausebutton at the same time?
HOw do I do it with amps? Can I get them levelmatched with a simple multi-meter?
Konnichiwa,
That is one methode I have been using previously. But it does not provide a double blind scenario.
We have done that too, using a suitable remote. It really only works well if the players have the same sort of platform (Drive, Control uP etc.)
You would need to build a suitable switch.
HOWEVER, and I repeat this, prior to actually attempting to test anything else you should proove your test with "known audible" stimulae under double blind conditions. Just like a measured result is useless if the measurement equipment was not calibrated so any other test device (including a DBT Setup) is useless unless verified. And if your test would happen to fail to show reliable audibility for items that are generally considered audible, what use would there be to apply it to items of questionable audibility?
Sayonara
keyser said:
That is one methode I have been using previously. But it does not provide a double blind scenario.
keyser said:
We have done that too, using a suitable remote. It really only works well if the players have the same sort of platform (Drive, Control uP etc.)
keyser said:
You would need to build a suitable switch.
HOWEVER, and I repeat this, prior to actually attempting to test anything else you should proove your test with "known audible" stimulae under double blind conditions. Just like a measured result is useless if the measurement equipment was not calibrated so any other test device (including a DBT Setup) is useless unless verified. And if your test would happen to fail to show reliable audibility for items that are generally considered audible, what use would there be to apply it to items of questionable audibility?
Sayonara
Actually, in this, I agree with Thorsten. Make sure that your test setup and listeners can distinguish well-established things like a 0.2dB level difference or a 1dB shelf at 1K before running tests on, um, less well-established phenomena. And avoid the major error source of most amateur approaches- level matching MUST be done precisely, and that is tedious, fiddly, and requires a precision attenuator (which is why most casual tests botch it).
Nice! to see these two knowledgable men at the same frequencie...
I am willing to invest some time and trouble (but hopefully not too much money) on the test. Would a ABX comparator make things a lot easier? I've seen netpages about the use of it, but can't find a page with a schematic or building guide. You know of one? Is a regular stereo pot a "precision attenuator"?
(why do I believe you are not on one line on this one
)btw. last few days I've been reading in an old thread that goes about subjects similar to the ones on this thread. Stuart took part in it too. Seems this kind of thread always shows a similar patern...
http://www.diyaudio.com/forums/show...d=12752&highlight=amplifier+sound+quality+dbt
Konnichiwa,
Probably.
Nope, what you need are solutions that allow you to switch between two line level sources and two speaker sources click-free and (obviously) without material impact on transparency.
Correctly timed mercury relais for speakers and mercury wetted reed-relais for line sources should be usable. As each relay takes a certain time to switch on/off you need to account for these times so that as one releases the other engages accuratly.
I have in preference used 12-Pole rotary switches with a semi-random arrangement of connections and no endstop (and no markings on the knob) and used a spare deck to actually register the absolute position. Such switches can often make for transparent line level switching. With speakers this is more difficult.
BTW, rapid siwtching is neither a requirement nor indeed (IMHO) desirable for DB Testing, therefore you can make, break and remake connections as you require, as long as it is done in a fashion that complies with the requirements of the "double blind" principle.
No, you need to be able to set levels reliably and repeatebly to better than 0.1db difference across a wide range of levels. You could build one from tree sets of 10 Pole switches and resistors. However the transparency of the attenuator itself may be questionable as well as the possibility of interactions with the equipement under test, which requires an attenuator with constant input and output impedance which brings us to t-network types. And the impedance should be high enough to not present problems loading the source. All in all a solvable, but non-trivial problem.
Unscrupulous experimenters bent on "prooving their point" (eg the ABX Mafia) have been observed of using digital domian level adjustments, which of course can only be done transparently if the system after the attenuation has enough extra resolution (in the digital domain) to offset the difference. Therefor using cooledit to change the volume oin identical copies of a CD is not permissabel, neither is A2D conversion with D2A conversion back (all these where found among the methodes employed by said group).
Due to the subject remaining the same, the issues remaining the same and the positions remaining the same that is unavoidable.
In my view the positions are too solid and irrconcilable and sadly both sides have commited enough follies and unscientific behaviour that what is on record so far is not useful to answer us the question:
"Is there a potentially audible difference (which may still be inaudible in some systems and to some listeners) between amplifiers that measure identical in conventional terms."
Also, there is no answer yet to:
"What measurements correlate well with good subjective sound?"
Worse, most of the tests so far published on the subject tend to come from people with an agenda and further are subject to many drastic and severe criticisms of methode and analysis, which in effect prejudice strongly against considering their results as having any significance.
And unless you wish to add to that large number of tests that tell us nothing you need to clearly define your question and then consider how you need to arrange tests to allow it to provide sufficient in quantity and quality and free from prejudice on grounds of procedure and/or sensitivity of the test (eg, first test the test against known quantities). That is a non-trivial process, if done to actually answer the question, as opposed to just provide pseudo-scientific justificiation for persuing certain crusades and other Audio McCarthyism.
Something that gives an idication of how you can arrange tests to give data relevant to a given question and on the approximate processes, procedures and challenges are found again in this red cloth to all Technocrate Bull's, Stereophile, in Marcus Sauers Article "G*d is in the Nuances".
SayonaraStereophile - God is in the Nuances
In this article he describes a double blind test carried out by a German Psychologist, Juergen Ackerman. Do a search for "Expert testimony" in the above linked document and read from there.
Note also that the test cited does not answer "do amplifiers et al sound different?", nor did it attempt to.
Sayonara
keyser said:
Probably.
keyser said:
Nope, what you need are solutions that allow you to switch between two line level sources and two speaker sources click-free and (obviously) without material impact on transparency.
Correctly timed mercury relais for speakers and mercury wetted reed-relais for line sources should be usable. As each relay takes a certain time to switch on/off you need to account for these times so that as one releases the other engages accuratly.
I have in preference used 12-Pole rotary switches with a semi-random arrangement of connections and no endstop (and no markings on the knob) and used a spare deck to actually register the absolute position. Such switches can often make for transparent line level switching. With speakers this is more difficult.
BTW, rapid siwtching is neither a requirement nor indeed (IMHO) desirable for DB Testing, therefore you can make, break and remake connections as you require, as long as it is done in a fashion that complies with the requirements of the "double blind" principle.
keyser said:
No, you need to be able to set levels reliably and repeatebly to better than 0.1db difference across a wide range of levels. You could build one from tree sets of 10 Pole switches and resistors. However the transparency of the attenuator itself may be questionable as well as the possibility of interactions with the equipement under test, which requires an attenuator with constant input and output impedance which brings us to t-network types. And the impedance should be high enough to not present problems loading the source. All in all a solvable, but non-trivial problem.
Unscrupulous experimenters bent on "prooving their point" (eg the ABX Mafia) have been observed of using digital domian level adjustments, which of course can only be done transparently if the system after the attenuation has enough extra resolution (in the digital domain) to offset the difference. Therefor using cooledit to change the volume oin identical copies of a CD is not permissabel, neither is A2D conversion with D2A conversion back (all these where found among the methodes employed by said group).
keyser said:
Due to the subject remaining the same, the issues remaining the same and the positions remaining the same that is unavoidable.
In my view the positions are too solid and irrconcilable and sadly both sides have commited enough follies and unscientific behaviour that what is on record so far is not useful to answer us the question:
"Is there a potentially audible difference (which may still be inaudible in some systems and to some listeners) between amplifiers that measure identical in conventional terms."
Also, there is no answer yet to:
"What measurements correlate well with good subjective sound?"
Worse, most of the tests so far published on the subject tend to come from people with an agenda and further are subject to many drastic and severe criticisms of methode and analysis, which in effect prejudice strongly against considering their results as having any significance.
And unless you wish to add to that large number of tests that tell us nothing you need to clearly define your question and then consider how you need to arrange tests to allow it to provide sufficient in quantity and quality and free from prejudice on grounds of procedure and/or sensitivity of the test (eg, first test the test against known quantities). That is a non-trivial process, if done to actually answer the question, as opposed to just provide pseudo-scientific justificiation for persuing certain crusades and other Audio McCarthyism.
Something that gives an idication of how you can arrange tests to give data relevant to a given question and on the approximate processes, procedures and challenges are found again in this red cloth to all Technocrate Bull's, Stereophile, in Marcus Sauers Article "G*d is in the Nuances".
SayonaraStereophile - God is in the Nuances
In this article he describes a double blind test carried out by a German Psychologist, Juergen Ackerman. Do a search for "Expert testimony" in the above linked document and read from there.
Note also that the test cited does not answer "do amplifiers et al sound different?", nor did it attempt to.
Sayonara
RHosch--"Challenge"?
This has been an interesting forum.
RHosch,
I am somewhat new to the forum, forgive me if I ask the question, what is the "challenge" that you are referring to? I would like to try the challenge.
Personally, I like to find out if I am in fact can tell the different sonic subtleties in blind tests.
I in fact did search for different amps, speakers, sources, etc over the years when I can, financially that is. I did hear differences in the music when changing pre amps, amps, speakers, sources.
The last time I did was with 4 different pre amps, 3 speakers, 5 amps, same source (bunch of CD and some Vinyl), same speaker cables (just 10GA copper), etc:
Change Only one component at a time, other components remain the same; all possible combinations, 60 combinations, I believe. The only thing remain the same was the source and recorded material. The trial was over couple of weeks.
I did hear differences in the music. better? I can not say. Measurable data, I can not provide due to lack of instruments, except my ear.
But I did settle on one that I prefer.
This has been an interesting forum.
RHosch,
I am somewhat new to the forum, forgive me if I ask the question, what is the "challenge" that you are referring to? I would like to try the challenge.
Personally, I like to find out if I am in fact can tell the different sonic subtleties in blind tests.
I in fact did search for different amps, speakers, sources, etc over the years when I can, financially that is. I did hear differences in the music when changing pre amps, amps, speakers, sources.
The last time I did was with 4 different pre amps, 3 speakers, 5 amps, same source (bunch of CD and some Vinyl), same speaker cables (just 10GA copper), etc:
Change Only one component at a time, other components remain the same; all possible combinations, 60 combinations, I believe. The only thing remain the same was the source and recorded material. The trial was over couple of weeks.
I did hear differences in the music. better? I can not say. Measurable data, I can not provide due to lack of instruments, except my ear.
But I did settle on one that I prefer.
The most commonly mentioned "challenge" is the years old but still live and well challenge offered by Richard Clark. Google on his name or search this and other audio related forums and I'm sure you will find plenty of information. $10,000 is yours with little "ante" money required from you (perhaps none... can't remember all the the possible alternatives and details).
Another possibility is the James Randi $1 million prize, which might be available for some audio related feats of magic. Some have suggested (and Randi himself has made mention of) the possibility of audio snake oil falling into his baliwick. Not sure though... again do a search for more info.
Another possibility is the James Randi $1 million prize, which might be available for some audio related feats of magic. Some have suggested (and Randi himself has made mention of) the possibility of audio snake oil falling into his baliwick. Not sure though... again do a search for more info.
The DBT will probably be held in June. We decided we need a list with equipment and knowledge we need. I have a testtone cd, and a multimeter. This should be enough to match level, right? I do the organisation together with someone else, but neither of us is very convident with a soldering iron. Building an ABX comparator seems too difficult.
cables shouldn't be so hard to compare: for interlinks I could use a Y-connector (RCA) from the source to two interlinks, and lead those to different source inputs on my pre-amp. For speakerleads I could use a similar setup, with one pair of leads from speaker out A on my poweramp and another pair on speaker out B of my poweramp. Both wires to the same loudspeakerterminals. The remote of my preamp enables me to instantly switch between speaker out A and B.
For the DOUBLE blind part we could use some kind of curtain before the equipment, so there can be no contact between the test subject and the person that does the switching.
What more do we need to know/do?
Why does is need to be click free? For the test subjects are to know when the switch will be. What is so bad about a >click< ?
cables shouldn't be so hard to compare: for interlinks I could use a Y-connector (RCA) from the source to two interlinks, and lead those to different source inputs on my pre-amp. For speakerleads I could use a similar setup, with one pair of leads from speaker out A on my poweramp and another pair on speaker out B of my poweramp. Both wires to the same loudspeakerterminals. The remote of my preamp enables me to instantly switch between speaker out A and B.
For the DOUBLE blind part we could use some kind of curtain before the equipment, so there can be no contact between the test subject and the person that does the switching.
What more do we need to know/do?
Konnichiwa,
Well, if your test protocol contains any "same" pairs there will be no switching and thus no click.
Equally, if there are clicks they tend to be non-consistent, meaning they depend on the exact signal level when switching and they may thus generate (unintentionally) false cues.
I would have thought so much would be obvious.
You think so? I would have thought that the pitfall from this is painfully obvious? What if one of the cables tested (or both) create interactions with the output stage driving them? If so you may find that these swamp out other differences.... Consider this post over at AA:
http://www.audioasylum.com/audio/general/messages/7083.html
That might work okay, but again you have both cables in circuit at least in terms of their shunt behaviour all the time.
That is NOT double blind, btw, it is single blind.
Also, what sort of provisions are you making for validating first the setup against known differences? You realise that faling to do so leaves you open to a large amount of potential criticsm?
Sayonoara
keyser said:
Well, if your test protocol contains any "same" pairs there will be no switching and thus no click.
Equally, if there are clicks they tend to be non-consistent, meaning they depend on the exact signal level when switching and they may thus generate (unintentionally) false cues.
I would have thought so much would be obvious.
keyser said:
You think so? I would have thought that the pitfall from this is painfully obvious? What if one of the cables tested (or both) create interactions with the output stage driving them? If so you may find that these swamp out other differences.... Consider this post over at AA:
http://www.audioasylum.com/audio/general/messages/7083.html
keyser said:
That might work okay, but again you have both cables in circuit at least in terms of their shunt behaviour all the time.
keyser said:
That is NOT double blind, btw, it is single blind.
Also, what sort of provisions are you making for validating first the setup against known differences? You realise that faling to do so leaves you open to a large amount of potential criticsm?
Sayonoara
An awful lot i'm afraid. Lack of proficiency with a soldering iron is not a good beginning. Complete lack of understanding of what can screw up the sound is even worse.
As comparing cables through a 50c Y-adaptor and listening to one set of speaker cables while another is connected at the same time?!
Why did so many people waste time giving you advice?
As KYW mentioned, cueing by differences in the sound of clicks is fatal. Checking (and compensating for) differences in level and frequency response between DUTs is vital.
The order of the variable (same or different, or whether X is A or B) should be predetermined using a valid random number generator. Or a coin flip. The key sheet should be held by the experimenter (out of sight of the participants) and a copy left with a third party.
Take a look at some of the protocols that Randi has used for his testing of psychics, dowsers, and the like. It's amazing how easy it is to overlook an unintended cue (cf, the Clever Hans effect, the SRI "testing" of Uri Geller...).
The order of the variable (same or different, or whether X is A or B) should be predetermined using a valid random number generator. Or a coin flip. The key sheet should be held by the experimenter (out of sight of the participants) and a copy left with a third party.
Take a look at some of the protocols that Randi has used for his testing of psychics, dowsers, and the like. It's amazing how easy it is to overlook an unintended cue (cf, the Clever Hans effect, the SRI "testing" of Uri Geller...).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.