Capacitor Sound Blind Test Results

[Swerd]

Last Saturday (Oct 9), I had the great experience of attending an annual meeting of DIY speaker builders mainly from the Washington, DC area, the DC2004 DIY meeting. It was organized by Dennis Murphy, Peter Smith, Neil Davis, Bob Cardell, Ken Ahern and others who I can't remember. It was my first time at such a meeting and I really enjoyed it. At least 20 speakers were presented by the builders. I can't say I heard any that sounded bad, although many were different and some were outstanding. My reason for posting this is not to describe the DIY speakers, but someting different.

As a side show to the speakers, the organizers performed blind listening tests of various expensive (Solen metalized polypropylene), moderate (mylar), and cheap (electrolytic) capacitors mounted in the crossovers of several speakers. They built crossovers that contained two different capacitors that could be switched by the listener via a remote controlled relay. The bottom line was that there was no evidence of an ability to hear a difference or to prefer one capacitor type over another. I had never been part of a blind listening test before and it was actually fun and illuminating. Most people there were very interested in the findings, and we talked about it for several hours afterwards. I'm afraid it did overshadow the speakers themselves.

The details of these test are presented at the Madisound discussion board http://www.madisound.com/cgi-bin/discuss.cgi

Pay attention to at least these threads:

• DCDIY2004 is a wrap Dennis Murphy (pool-151-200-49-42.res.east.verizon.net) -- 10/9 8:06p.m. http://www.madisound.com/cgi-bin/discuss.cgi?read=320118
• More Cap Test Results Dennis Murphy (pool-151-200-38-6.res.east.verizon.net) -- 10/10 11:12a.m. http://www.madisound.com/cgi-bin/discuss.cgi?read=320147
• Capacitor Listening Test Methodology Bob (pcp09279223pcs.eatntn01.nj.comcast.net) -- 10/10 11:29a.m. http://www.madisound.com/cgi-bin/discuss.cgi?read=320152
• Cap Listening Test Results Bob (pcp09279223pcs.eatntn01.nj.comcast.net) -- 10/10 12:05p.m. http://www.madisound.com/cgi-bin/discuss.cgi?read=320154
• Measurements for Cap Test Dennis Murphy (pool-151-200-139-152.res.east.verizon.net) -- 10/11 6:57p.m. http://www.madisound.com/cgi-bin/discuss.cgi?read=320426

If you are interested enough to wade through the discussions that followed the above posts you will hear all the usual tired arguments speculating about why blind listening tests are invalid or should be ignored. Some of you will recognize one of the participants in particular (whose intitals are JR). I think all the objectors missed the point. No single listening test is perfect from all scientific or statistical viewpoints. This was a simple and limited (no more than 30 people participated) test that clearly showed there is little reason to do more elaborate testing concerning capacitors. I understand a similar test is planned for the upcoming DIY meeting in Dayton, Ohio to see if this can be repeated.

The organizers worked very hard to produce a fair test of the peoples' ability to hear potential differences in sound due to capacitors. The important take home lesson is quite simple. If you object to what was done, get busy and do your own test.

 

[WmAx]

Silly reply from me on Risch...

The test certainly had faults. If the results had been positive I would have disected the methodology from that perspective. I discussed the testing procedures with Bob a couple of months ago when the test was being put together; I made him aware of several issues that should be addressed for the most reliable results -- but the contraints of a general meet environment and the amount of time to inverst into pre-testing/development prevented some of these things from being employed. But I don't intend to address every one of those issues in this discussion, I just wanted to pick on JR after you made mention of him....

John Risch does have some valid points on this comparision, but some of his *points* vary from what I believe to be deceptive to rediculous.

Valid point: test was performed with continuing music, not repeated samples -- this would cause significant reduction of sensitivity, but in retrospect the increased sensitivty possible with repeated short samples may not be represenative of casual listening conditions. In support of those that are critical of the test: I found no published information on the specific room acoustics, detailed speaker measurements(not just basic FR graphs), ambient noise level measurements or detailed description of the environment the test subject had to endure. These are all very important issues from the perspective of test sensativity. But again, I'm not really replying to critique the test so much as Risch's specific excuses....

-Deceptive point: he urges people to remember that null results don't prove a negative -- TRUE -- but fails to mention the significance of this in perspective; the test failed to support the claims of capacitor audibility -- claims that were based on nothing more then subjective sighted listening. Their was never a good reason to believe the claims to begin with -- failure to support those claims certainly does not lend credibility to those who make the claims. But this is not the 1st capacitor test. Indirectly, refer to amp or preamp blind tests -- the different amplifiers contain different brand/model capacitors in the signal paths -- but this has not demostrated to be important to audibility since in controlled scenarios the tests have failed to produce positive results for audibility.

-Rediculous point: he claims that relays may be masking the signals -- pardon me for stating my feelings here; but I find this a bit like blaming the failure to videotape Casper on the Lochness monster's breeding cycle. Essentially making unfounded claims upon unfounded claims! Geez. Where is his evidence that the relays were audibly masking the signal? This is a positive claim -- one with no credible data presented to back it self up. But continue reading his posts and eventually he makes the extraordinary claim to be able to identify the 'type' of capacitor in a speaker crossover by sound. Uhm. Yeah. Sure. Proof? This sounds alot like those times he used to claim he could indentify the type of insulation on wire by sound. Wonder if he still makes that claim?

It is unfortunate that you believe the capacitor test was the primary focus(this is what I gather from your statement). Unfortunate that an empty claim such as this is prioritized above the important parts of the event(the actual speakers).

-Chris

Last Saturday (Oct 9), I had the great experience of attending an annual meeting of DIY speaker builders mainly from the Washington, DC area, the DC2004 DIY meeting. It was organized by Dennis Murphy, Peter Smith, Neil Davis, Bob Cardell, Ken Ahern and others who I can't remember. It was my first time at such a meeting and I really enjoyed it. At least 20 speakers were presented by the builders. I can't say I heard any that sounded bad, although many were different and some were outstanding. My reason for posting this is not to describe the DIY speakers, but someting different.



As a side show to the speakers, the organizers performed blind listening tests of various expensive (Solen metalized polypropylene), moderate (mylar), and cheap (electrolytic) capacitors mounted in the crossovers of several speakers. They built crossovers that contained two different capacitors that could be switched by the listener via a remote controlled relay. The bottom line was that there was no evidence of an ability to hear a difference or to prefer one capacitor type over another. I had never been part of a blind listening test before and it was actually fun and illuminating. Most people there were very interested in the findings, and we talked about it for several hours afterwards. I'm afraid it did overshadow the speakers themselves.

The details of these test are presented at the Madisound discussion board http://www.madisound.com/cgi-bin/discuss.cgi

Pay attention to at least these threads:

• DCDIY2004 is a wrap Dennis Murphy (pool-151-200-49-42.res.east.verizon.net) -- 10/9 8:06p.m. http://www.madisound.com/cgi-bin/discuss.cgi?read=320118
• More Cap Test Results Dennis Murphy (pool-151-200-38-6.res.east.verizon.net) -- 10/10 11:12a.m. http://www.madisound.com/cgi-bin/discuss.cgi?read=320147
• Capacitor Listening Test Methodology Bob (pcp09279223pcs.eatntn01.nj.comcast.net) -- 10/10 11:29a.m. http://www.madisound.com/cgi-bin/discuss.cgi?read=320152
• Cap Listening Test Results Bob (pcp09279223pcs.eatntn01.nj.comcast.net) -- 10/10 12:05p.m. http://www.madisound.com/cgi-bin/discuss.cgi?read=320154
• Measurements for Cap Test Dennis Murphy (pool-151-200-139-152.res.east.verizon.net) -- 10/11 6:57p.m. http://www.madisound.com/cgi-bin/discuss.cgi?read=320426

If you are interested enough to wade through the discussions that followed the above posts you will hear all the usual tired arguments speculating about why blind listening tests are invalid or should be ignored. Some of you will recognize one of the participants in particular (whose intitals are JR). I think all the objectors missed the point. No single listening test is perfect from all scientific or statistical viewpoints. This was a simple and limited (no more than 30 people participated) test that clearly showed there is little reason to do more elaborate testing concerning capacitors. I understand a similar test is planned for the upcoming DIY meeting in Dayton, Ohio to see if this can be repeated.

The organizers worked very hard to produce a fair test of the peoples' ability to hear potential differences in sound due to capacitors. The important take home lesson is quite simple. If you object to what was done, get busy and do your own test.

 

[Swerd]

John Risch does have some valid points on this comparision, but many of his *points* vary from deceptive to rediculous.

I agree with you on all your points here. It was interesting to me that JR's comments, although lengthy as usual for him, were much more coherent and civil than his usual rants that I've read or tried to read in the past. Perhaps his new medication is working well .

Valid point: test was performed with continuing music, not repeated samples -- this would cause significant reduction of sensitivity but in retrospect, the increased sensitivty possible with repeated short samples may not be represenative of casual listening conditions.

Perhaps this was valid and perhaps not. About 3-6 people at a time took the tests. Some groups did back up and listen to certain passages over and over, and others did not. People also often decided to skip certain types of music as being poor for detecting the very subtle differences said to come from different caps. I'm not sure there was any agreement as to which music passages were the most useful to hear. Each was on his own to examine the music for features that could illuminate any differences between the caps. Because no one knew what exactly these subtle differences were, I thought it was better to allow listeners to try various musical passages for themselves, rather than for the test organizers to preselect music for the listeners. Afterwards, both Bob Cardell and Dennis Murphy, who administered the tests, admitted that they could not hear any differences at all between capacitors, and that is why they couldn't select music passages for standard use during the tests.

I was actually one of the 5 people whose score approached significance. I selected correctly 8 out of 12 times. (Note that I said "approached significance". That is the same as saying "not statistically significant".) This is 2 hits better than random guessing leading to 6 out of 12, and 2 hits worse than a convincing score of 10 out of 12. The one listener who scored 9 out of 12 really preferred the cheap caps. So his score was actually 3 out of 12, but odds of randomly getting 9 of 12 or 3 of 12 are really the same.

During the test, I thought that I had learned to discern very small and subtle differences between the caps. I could try to describe these different sounds here, but I don't know whether it was for real or just so much mental masturbation. (I also don't want to influence anyone who may be part of the tests in Dayton and elsewhere.)

After the tests were done, I was very interested in finding out what audible features other people perceived (that word includes heard and/or thought they heard) as being different between the caps. In talking to 3 others, all of whom were equally convinced that they could hear it as I was, I found that we all had focused on completely different features in the music when we thought we could hear it. We also got differing scores. So we were hearing, or believing we were hearing, quite different features. So, I'm not sure that using short repeated passages of music that were standard for everyone would have helped.

 

[WmAx]

Perhaps this was valid and perhaps not. About 3-6 people at a time took sat through the tests. Some groups did back up and listen to certain passages over and over, and others did not. People also often decided to skip certain types of music as being poor for detecting the very subtle differences said to come from different caps.

If you don't use the same signal to compare, then you are relying on the similar but non-exact patterns of the musical passage to extrapolate. Absolute sensativity is reduced without using the same signal to compare.

I was actually one of the 5 people whose score approached significance. I selected correctly 8 out of 12 times. This is 2 hits better than random guessing leading to 6 out of 12, and 2 hits worse than a convincing score of 10 out of 12. The one listener who scored 9 out of 12 really preferred the cheap caps. So his score was actually 3 out of 12, but odds of randomly getting 9 of 12 or 3 of 12 are really the same.

This is common deviance among large group of test subjects. As you increase the number of trials, the more likely random errors that appear as significant are to emerge in small samples. The solution is to retest subjects that score significantly in order to confirm the result was accurate or an error.

-Chris

 

[mtrycrafts]

-Rediculous point: he claims that relays may be masking the signals --
-Chris

One of his old and tired, straw man excuses with no evidence to support him that a switch is audible in any way, except perhaps when the point is corroded or something.
What did you expect from him?

 

[mtrycrafts]

I agree with you on all your points here. It was interesting to me that JR's comments, although lengthy as usual for him, were much more coherent and civil than his usual rants that I've read or tried to read in the past. Perhaps his new medication is working well .


Perhaps this was valid and perhaps not. About 3-6 people at a time took the tests. Some groups did back up and listen to certain passages over and over, and others did not. People also often decided to skip certain types of music as being poor for detecting the very subtle differences said to come from different caps. I'm not sure there was any agreement as to which music passages were the most useful to hear. Each was on his own to examine the music for features that could illuminate any differences between the caps. Because no one knew what exactly these subtle differences were, I thought it was better to allow listeners to try various musical passages for themselves, rather than for the test organizers to preselect music for the listeners. Afterwards, both Bob Cardell and Dennis Murphy, who administered the tests, admitted that they could not hear any differences at all between capacitors, and that is why they couldn't select music passages for standard use during the tests.

I was actually one of the 5 people whose score approached significance. I selected correctly 8 out of 12 times. (Note that I said "approached significance". That is the same as saying "not statistically significant".) This is 2 hits better than random guessing leading to 6 out of 12, and 2 hits worse than a convincing score of 10 out of 12. The one listener who scored 9 out of 12 really preferred the cheap caps. So his score was actually 3 out of 12, but odds of randomly getting 9 of 12 or 3 of 12 are really the same.

During the test, I thought that I had learned to discern very small and subtle differences between the caps. I could try to describe these different sounds here, but I don't know whether it was for real or just so much mental masturbation. (I also don't want to influence anyone who may be part of the tests in Dayton and elsewhere.)

After the tests were done, I was very interested in finding out what audible features other people perceived (that word includes heard and/or thought they heard) as being different between the caps. In talking to 3 others, all of whom were equally convinced that they could hear it as I was, I found that we all had focused on completely different features in the music when we thought we could hear it. We also got differing scores. So we were hearing, or believing we were hearing, quite different features. So, I'm not sure that using short repeated passages of music that were standard for everyone would have helped.

Thanks for this post

Would you not expect 1 person in 30 to have such a score, 9 of 12?

Caps are interesting to compare as its value can alter the crossover characteristics that may be audible. It must be carefully selected. The group did, obviously

Ceramic caps are sensitive to heat, demonstrated in DBT, as on one test, it was next to a heat source and its value skyrocketed.

 

[TJ_Jazz]

As someone in the ceramic, film and electrolytic capacitor industry, I really enjoyed reading this discussion.

Thanks to all for the postings!

- TJ

 

[mtrycrafts]

As someone in the ceramic, film and electrolytic capacitor industry, I really enjoyed reading this discussion.

Thanks to all for the postings!

- TJ

Since you are in the industry, are there any industry or pro group DBT listening test that you know of? Any discussion about audible differences in these caps?

 

[Unregistered]

Unbelieveable

So room acoustics (FR response), relay contacts (oops it depends on corrosion, of course who determines how much corrosion is necessary for sonic integrity to be compromised), component quality, and speaker quality is assumed not to matter in these tests. What blind faith!

Ok, then I could use an amp with an FR of 20hz to 100hz to compare caps and accurately determine if caps sound different. And, yes, I can setup a resonance at 30hz, some 20db elevation and still get an accurate DB test.

LOL, I am all over the floor on this one.

I cannot believe anyone would be gullible to "assume" so many variables make no sonic difference, without proof, and can come out with a proper conclusion.

So this is an accurate DB test with no assumptions made, and results are considered factual.

WOW, I hope others don't buy into this mumble jumble propaganda crap of a test.

Anonymous

 

[Rip Van Woofer]

Cross-post alert: As I said in another area, might I suggest we ignore Mr. or Ms. Anonymous until he/she sees fit to register and post under at least a handle? The person seems more intent on s***-stirring than constructive debate. Let's not encourage trolling and flame-baiting.

 

[WmAx]

Wow! Judging by your post you did not bother to read & consider the actual posts/replies in this thread. Foot in mouth disease is curable, though....

-Chris

So room acoustics (FR response), relay contacts (oops it depends on corrosion, of course who determines how much corrosion is necessary for sonic integrity to be compromised), component quality, and speaker quality is assumed not to matter in these tests. What blind faith!

Ok, then I could use an amp with an FR of 20hz to 100hz to compare caps and accurately determine if caps sound different. And, yes, I can setup a resonance at 30hz, some 20db elevation and still get an accurate DB test.

LOL, I am all over the floor on this one.

I cannot believe anyone would be gullible to "assume" so many variables make no sonic difference, without proof, and can come out with a proper conclusion.

So this is an accurate DB test with no assumptions made, and results are considered factual.

WOW, I hope others don't buy into this mumble jumble propaganda crap of a test.

Anonymous

 

[WmAx]

Cross-post alert: As I said in another area, might I suggest we ignore Mr. or Ms. Anonymous until he/she sees fit to register and post under at least a handle? The person seems more intent on s***-stirring than constructive debate. Let's not encourage trolling and flame-baiting.

Mr. moderator: why not REQUIRE registration to post? I never 'got' the anonymous posting permissions allowed on a forum such as Audioholics.

-Chris

 

[mtrycrafts]

Cross-post alert: As I said in another area, might I suggest we ignore Mr. or Ms. Anonymous until he/she sees fit to register and post under at least a handle? The person seems more intent on s***-stirring than constructive debate. Let's not encourage trolling and flame-baiting.

So many anonymous posters who knows whom to ignore

 

[Rob Babcock]

Mr. moderator: why not REQUIRE registration to post? I never 'got' the anonymous posting permissions allowed on a forum such as Audioholics.

-Chris

We have been discussing this, Chris. Gene originally set it up so anyone could post in an attempt to increase traffic. Oftimes casual forum surfers won't bother with registration, but will post if they can do so without any "added hassles." I've never liked that myself- I think a person should be accountable for their words, even if they uses a screen name. People will think twice about saying something if they can be called out on it. I've long lobbied Admin to change this, and perhaps it may change. We'll just have to wait and see.

 

[mtrycrafts]

We have been discussing this, Chris. Gene originally set it up so anyone could post in an attempt to increase traffic. Oftimes casual forum surfers won't bother with registration, but will post if they can do so without any "added hassles." I've never liked that myself- I think a person should be accountable for their words, even if they uses a screen name. People will think twice about saying something if they can be called out on it. I've long lobbied Admin to change this, and perhaps it may change. We'll just have to wait and see.

A consitent screen name is one thing A blanket anonymous is another. Difficult to know if one anonymous is the same as the next until later. Maybe a running counter if that could be linked to a person; anonymous 12

 

[Westrock2000]

But continue reading his posts and eventually he makes the extraordinary claim to be able to identify the 'type' of capacitor in a speaker crossover by sound. Uhm. Yeah. Sure. Proof? This sounds alot like those times he used to claim he could indentify the type of insulation on wire by sound. Wonder if he still makes that claim?

-Chris

I saw this happen to a particulary popular person on a particulary major audio message board, in about a years time they went from "so your saying that insulation types on wires make a difference?" to "I can easily tell the difference in material types on <equipment> feet".....very sad I won't mention that persons name, but they are pretty respected on that particular message board.

I just flat gave up on trying so hard....there are certain things I think/know make a difference...such as planar speakers versus cone speakers....but I fail too see how something like wire or a resistor can have an impact when you've got something as horrible as a speaker driver on the end. Drivers seem to be the most inaccurate part of a stereo by far....yet poeple compare them just the same a freakin solid state component...which has accuracy with the zero several places to the left!

I think blind tests are the best method. I mean the way that these poeple claim its the only way that it works, is completely different from real life....I mean I hope no one listens to a 10 second clip over and over again for enjoyment....normal listening is your best acid test.


Perhaps assinging < somewhat > unique names to IP addresses would help....like if some one kept logging in from 234.345.276.73 then they would get a auto name like ANON234BUBBLES....course I doubt the software has that kinda feature

 

[Audioholics]

I'll chat with Gene today - I'm leaning towards making it mandatory to register to post, at least in some forums. It's getting to that point where its not fair to our users.

 

[Unregistered]

Just make it mandatory to enter a username, not to register.