SY, i understand that; as stated the preference test approach does has its problem too, but normally more to the point that it could _mask_ a difference (produce a false negative) than that it could favour a false positive.
At least as long if you´re doing group tests, because the preference might be not consistent among the group.
As long as the presentation order isn´t known and changed i just don´t see that the preference of the group for one answer could favour one cable overall (Answer "A" means sometimes cable "X" and sometimes cable "Y").
As shown by the results, the negative control trial is indeed the only one in which answer "A" and "B" were nearly balanced, while the answer "same" as usual is quite rare, which enlightens again the well known fact, that the detection of "sameness" is quite difficult if .....
Wishes
P.S. I´ve read that you have just little spare time at the moment, so don´t feel under pressure, there is no hurry. 🙂
They will tend to hear a difference if they hear the switching. If they don't know whether it was switched or not (A, B or X) you remove one more non-auditory cue whether there is a difference or not.
jd
Cables have microphony effects, they must also have some mechanical resonance. If you test a cable for microphony with a sharp transient while it is unplugged from source, you will hear it resonate for a long time in relation to the forcing transient. All I can tell you is everyone who's heard it swears by it. Please contain yourself BupD, we all share a common love. Regretfully I cannot pass such comportment along to friends. Why don't you just listen?
That might be the case, but the idea behind the forced choice test is that it doesn´t matter wrt to the result.
If you would do a test like this with on participant you normally would even omit the choice "no difference" and instead force the participant to choose between "A is better than B" or "B is better than A" .
Sturm included the "no difference" answer because the groups seemed to be disturbed by single participants who need that answer; he find that out in preliminary tests.
I a single participant test it wouldn´t matter if a single person doesn´t like to be forced that way, but if the whole group will be disturbed it is another case.
But in anyway as long as the DUT behind "A" and "B" is changed it doesn´t matter, if participants will tend to hear a difference as the effect will be random.
BTW, this "tend to hear a difference" seems to be exactly the effect that led to the difficulty to detect "sameness" . 🙂
Wishes
Jakob, during the negative control, did the presenter know that there was no change? If not, how was that set up?
The paper is not specific in this point, but i´m sure he must have known according to the switching scheme he had to follow for every trial (otherwise the controller inside the preamplifier must had been changed, to prevent the VL from knowing and i doubt that); from a discussion a couple of years ago, i think that he pretended to do a switch, but that isn´t clear.
Wishes
Old Business
The question asked was why some people think speaker cable makes more of a difference than interconnects. When I answered this you seemed to differ.
8m may seem long in some locations, but you probably waste 1m from the amplifier to the floor and another 1 to 1.5 to the loudspeaker mounted at ear height.
Now many folks place their equipment rack between a pair of loudspeakers. This is turns out to be a bad place. The normal base line of vibration in a quiet location is about .05G this rises to around .2G measured on a 6 x 6 x 4" case at 1m from a single loudspeaker at 98db. An electrolytic coupling capacitor inside such a case would create a voltage from this level of vibration of about 20uv. If this were the output coupling capacitor of a CD player that vibration by itself would limit the CD players resolution to 16 bits. Of course most CD cases are larger, there are often capacitors at lower signal levels where the amplification that follows increase the error and many people replace the electrolytic capacitors with film types that have higher vibration induced noise output. Of course you can play your CD's so that the peak level is less than 98db (78db average listening level with headroom allowance) but then the ambient noise level will reduce your dynamic range to around 9 bits of resolution. Of course if you use any kind of processing these numbers will be reduced again. (Both analog and digital processors put more capacitors in the signal path)
A better approach is to take advantage of the room's natural null nodes and to place your equipment there. That is why even in a small room 8m is a typical cable length.
The next issue to affect cable length is the ability of the room to support low frequencies. Even if we oversimplify things by way too much we want the longest dimension to be at least 1/2 of a wavelength of the lowest frequency so a 4m room is limited to about 45hz. The room will really be "muddy" at around double that. Of course the ratio of the length to width to the height should be considered. Most rooms have a fixed ceiling height that is based on drywall dimensions these days. When rooms were done with wet plaster the dimensions varied more and the walls had less absorption and flexing losses. The length of a room can often be twice the width but 1.5x or 1.414 even 1.732 are better ratios. So again we want a bit more cable.
As to my tale about being able to discern .5db frequency adjustments. That was at a football stadium 100m or more from the loudspeakers. You may sit closer in your listening room. Dr. Toole as I recall in his '83 papers mentioned that this is detectable around .1 to .2 db but I am pretty sure he was refering to prior papers for that. Across the pond Martin Colum (SP?) has mentioned that .25db is his window for a loudspeakers on axis curve. (Again from memory) Now I haven't done a double blind test, but you certainly can look to others who have. Of course anyone can verify this with their own tone controls. Dr Toole of course has a lot more of insight to offer on not just loudspeakers but also rooms, but here that is OT.
Now as to the value of cable resistance. I don't know if you have ever measured this yourself. Every time I measure a given gauge product "Big Surprise" it always measures at the bottom of the standard for the rated thickness! So your 10 Gauge solid wire may actually measure .0965" instead of .1019" Of course you quote stranded. This is usually made of 26 gauge wires. They also come in under size, but often worse they are not truly round but oval shaped. So the resistance of 10 gauge instead of .9989/1000' is more often around 1.125/1000'. If you have an 8 ohm loudspeaker that drops to 4 ohms as most do. (Specifications usually allow the minimum impedance to be 1/2 of the rating) Any length of 10 gauge greater than 6.27m of loudspeaker cable may create differences that can be heard. Of course you may wish to actually connect your cables and when you allow for the .025 ohms of each connection this drops to 4.92m!
As to soakage most wire uses soft insulation such as PVC this means that not only does it increase the capacitance of the cable but it is a "bad" dielectric. Because most audio amplifiers use feedback to correct the output voltage level the soakage adds an unwanted time dimension to the correction. Similarly dispersion can be seen in the change to what a unit impulse response looks like going out from the amplifier and what comes back to be corrected. In a 10 gauge cable the skin effect limits the complete utilization of the conductor to a frequency around 2600hz. The damping factor of an amplifier rises with frequency so these effects become more noticable. I will not even go into cable contamination cause by the drawing process.
So we actually agree, if you are listening to a CD player positioned between your loudspeakers, have cables less than 4m, using an equalizer, a reasonable quality amplifier and loudspeakers, you most likely are right, you will not hear any potential differences between cables.
The question asked was why some people think speaker cable makes more of a difference than interconnects. When I answered this you seemed to differ.
8m may seem long in some locations, but you probably waste 1m from the amplifier to the floor and another 1 to 1.5 to the loudspeaker mounted at ear height.
Now many folks place their equipment rack between a pair of loudspeakers. This is turns out to be a bad place. The normal base line of vibration in a quiet location is about .05G this rises to around .2G measured on a 6 x 6 x 4" case at 1m from a single loudspeaker at 98db. An electrolytic coupling capacitor inside such a case would create a voltage from this level of vibration of about 20uv. If this were the output coupling capacitor of a CD player that vibration by itself would limit the CD players resolution to 16 bits. Of course most CD cases are larger, there are often capacitors at lower signal levels where the amplification that follows increase the error and many people replace the electrolytic capacitors with film types that have higher vibration induced noise output. Of course you can play your CD's so that the peak level is less than 98db (78db average listening level with headroom allowance) but then the ambient noise level will reduce your dynamic range to around 9 bits of resolution. Of course if you use any kind of processing these numbers will be reduced again. (Both analog and digital processors put more capacitors in the signal path)
A better approach is to take advantage of the room's natural null nodes and to place your equipment there. That is why even in a small room 8m is a typical cable length.
The next issue to affect cable length is the ability of the room to support low frequencies. Even if we oversimplify things by way too much we want the longest dimension to be at least 1/2 of a wavelength of the lowest frequency so a 4m room is limited to about 45hz. The room will really be "muddy" at around double that. Of course the ratio of the length to width to the height should be considered. Most rooms have a fixed ceiling height that is based on drywall dimensions these days. When rooms were done with wet plaster the dimensions varied more and the walls had less absorption and flexing losses. The length of a room can often be twice the width but 1.5x or 1.414 even 1.732 are better ratios. So again we want a bit more cable.
As to my tale about being able to discern .5db frequency adjustments. That was at a football stadium 100m or more from the loudspeakers. You may sit closer in your listening room. Dr. Toole as I recall in his '83 papers mentioned that this is detectable around .1 to .2 db but I am pretty sure he was refering to prior papers for that. Across the pond Martin Colum (SP?) has mentioned that .25db is his window for a loudspeakers on axis curve. (Again from memory) Now I haven't done a double blind test, but you certainly can look to others who have. Of course anyone can verify this with their own tone controls. Dr Toole of course has a lot more of insight to offer on not just loudspeakers but also rooms, but here that is OT.
Now as to the value of cable resistance. I don't know if you have ever measured this yourself. Every time I measure a given gauge product "Big Surprise" it always measures at the bottom of the standard for the rated thickness! So your 10 Gauge solid wire may actually measure .0965" instead of .1019" Of course you quote stranded. This is usually made of 26 gauge wires. They also come in under size, but often worse they are not truly round but oval shaped. So the resistance of 10 gauge instead of .9989/1000' is more often around 1.125/1000'. If you have an 8 ohm loudspeaker that drops to 4 ohms as most do. (Specifications usually allow the minimum impedance to be 1/2 of the rating) Any length of 10 gauge greater than 6.27m of loudspeaker cable may create differences that can be heard. Of course you may wish to actually connect your cables and when you allow for the .025 ohms of each connection this drops to 4.92m!
As to soakage most wire uses soft insulation such as PVC this means that not only does it increase the capacitance of the cable but it is a "bad" dielectric. Because most audio amplifiers use feedback to correct the output voltage level the soakage adds an unwanted time dimension to the correction. Similarly dispersion can be seen in the change to what a unit impulse response looks like going out from the amplifier and what comes back to be corrected. In a 10 gauge cable the skin effect limits the complete utilization of the conductor to a frequency around 2600hz. The damping factor of an amplifier rises with frequency so these effects become more noticable. I will not even go into cable contamination cause by the drawing process.
So we actually agree, if you are listening to a CD player positioned between your loudspeakers, have cables less than 4m, using an equalizer, a reasonable quality amplifier and loudspeakers, you most likely are right, you will not hear any potential differences between cables.
Trial no.3 in each run were indeed not double blind, but look at the results of this trial; a not blinded experimentator that favours one possible answer is thinkable, but an experimentator that keeps track of the answers already given and favours each answer so, that both "A" and "B" will be nearly balanced at the end of the fair?
Wishes
Your post reminded me I owe boconner a graph. Below, a Spice sim of the frequency response at the speaker terminals of 15 feet of 18 gauge - below the maximum recommended by conservative sources like Roger Russell or Audioholics - driving the DR MTM project from PartsExpress. The amplifier has infinite damping factor and no output protection coil. The cable is modeled in 1 foot pseudo-lumped sum segments from values in a Nelson Pass paper. This is why probably why speaker cables are considered more audible.
Attachments
This observation is extremely interesting and merits further experiments at home. Many thanks. To me, post is probably the most valuable contribution in this thread.
(Der er langt mellem snapsene).
Don't laugh, I had a professor for QM, an Italian fellow, who would walk around the room during an exam looking over the students' shoulders as they'd work through the problems. Talk about nervous-making. I remember doing a problem on a pop quiz when suddenly, WHAM!, he had hit me on the back of my head. "What-a the hell-a you doing-a?!?! I no teach-a you that!!!!"
Member
Joined 2002
I'll take it! One of my professors assigned a textbook he'ld authored, then instructed his students to derive from first principles the correct replacement equations for his many errors. The student assessments of his performance were so appallingly bad the university had little choice but to promote him to dean of the department.
- Status
- Not open for further replies.