Loudspeaker Myths: Separating the Scientific Facts from Science Fiction

[Swerd]

It is our understanding that Harman runs mono single speaker comparisons using positional substitution so that each time a speaker is switched, the competitor product is placed directly in the same location as the last speaker tested. This method is designed to remove the room and the stereo effect from the contest. According to Dr. Floyd Toole's research, a speaker that wins in mono ALWAYS win in stereo too.

Why not add the room and stereo effect to the contest?

I know only one speaker designer, and he always makes his designs while listening and testing a single speaker in mono. In fact, all the audio measurement techniques he uses while designing are done with a single speaker. I'm not sure what objective measurements could be done with a stereo pair, nor do I understand just how one could go about designing a speaker using non-objective listening techniques while listening in stereo.

Based on that limited experience, I readily believe what Floyd Toole says on that subject.

Never say never, but I'm not buying off on an audible break-in period for tweeters. There's a dedicated thread on this, where I bought a pair of Cambridge Aero-2 speakers, played one for 50 hours, and left the other unit mint.

I was one of those blind listeners , and I am definitely not the recording engineer. I heard no difference at all between the two speakers. I also witnessed the measurement session that went into the graphs that were posted:

 

[Swerd]

What I do know from talking to engineers and reading various articles about electronic equipment is that capacitors GENERALLY have a break in period and since there are various brands of crossover that can be used in a crossover, to me, it would sound plausible that that could be what is REALLY being broken in and possibly not the actual driver. Obviously there are a LOT of different brands/models and designs of speakers on the market so what works for one might not work for another.

When I first read the series of posts from this guy, my first thought was to refute him point by point. Then I read the above quote, and I decided a simple graphic comment was more appropriate:

After further reflection, that seems a bit harsh. This is more appropriate:

While it may be true that capacitors have an electronic break-in that is measurable, no one has ever demonstrated that capacitors of different construction but of the same capacitance make an audible difference is speaker crossovers. This qualifies as pseudo-scientific nonsense, and borders on misleading.

 

[markw]

When I first read the series of posts from this guy, my first thought was to refute him point by point. Then I read the above quote, and I decided a simple graphic comment was more appropriate:

After further reflection, that seems a bit harsh. This is more appropriate:

While it may be true that capacitors have an electronic break-in that is measurable, no one has ever demonstrated that capacitors of different construction but of the same capacitance make an audible difference is speaker crossovers. This qualifies as pseudo-scientific nonsense, and borders on misleading.

You are a very patient person. I came to this conclusion when he used MIT "cables" as undisputable proof of speaker cables having audiable difference.

 

[gene]

When I first read the series of posts from this guy, my first thought was to refute him point by point. Then I read the above quote, and I decided a simple graphic comment was more appropriate:

After further reflection, that seems a bit harsh. This is more appropriate:

While it may be true that capacitors have an electronic break-in that is measurable, no one has ever demonstrated that capacitors of different construction but of the same capacitance make an audible difference is speaker crossovers. This qualifies as pseudo-scientific nonsense, and borders on misleading.

Man I love these images and they could easily fit into my article. Where are you guys when I need images????

 

[gene]

While it may be true that capacitors have an electronic break-in that is measurable, no one has ever demonstrated that capacitors of different construction but of the same capacitance make an audible difference is speaker crossovers. This qualifies as pseudo-scientific nonsense, and borders on misleading.

You're wrong there. Electrolytics have much higher ESR and DF. This can easily be simulated and measured. Tolerance also plays a huge role in consistency.

I spent 3 years designing POTS filters in Telcom and studied this phenomenon with great detail. Be careful about oversimplification. Now some of the esoteric stuff? Well that's a diff story. If you change the response of the crossover network as a result, things can get audible depending on how severe and how good the design is.

 

[Swerd]

You're wrong there. Electrolytics have much higher ESR and DF. This can easily be simulated and measured. Tolerance also plays a huge role in consistency.

I spent 3 years designing POTS filters in Telcom and studied this phenomenon with great detail. Be careful about oversimplification. Now some of the esoteric stuff? Well that's a diff story. If you change the response of the crossover network as a result, things can get audible depending on how severe and how good the design is.

I may be generalizing some, but for capacitors, as they are used in passive audio crossover filters, the type of material does not make an audible difference. A number of years ago, I participated in a large (for audio) blind test of over 40 listeners, and no one could hear the difference among crossovers made with non-polar electrolytic, various metalized polypropylene types, or exotic silver/oil-filled caps. As long as the caps had the correct capacitance value, there was no audible difference. And yes, there were some careful efforts at scientific controls in this test. From those results, I'm convinced that the idea of different sound due capacitor material is debunked as myth.

Yes, tolerance in caps, especially the lack of it in many inexpensive electrolytic caps, is important.

 

[Swerd]

Man I love these images and they could easily fit into my article. Where are you guys when I need images????

Google Image is loads of fun – I searched BS meter and found dozens.

 

[Dennis Murphy]

Totally cool, Dennis. I know who you are and respect your opinion very much.

Mordaunt short is being liquidated everywhere now. Their Mezzo 2s are an excellent bookshelf speaker. I own them and experienced the subtle change in the tweeter. If anyone would like to test this theory out, they would be a great candidate. Run about 500$ online right now.

Its cool the room scenario is playing into some posts. Some rooms are better than others for sure. The differences we heard were not frequency response altering. They were simply smoother and not grating anymore in the high treble region. I was only exposed for minutes a day, so not an acclimation thing.

I should note that the "process" may have gone quicker had I ran them at higher levels. I never run them in at much louder than speaking levels. Too annoying for the rest of the house.

Hi I can't claim that my test with the Cambridge speakers can be generalized to other speakers. I picked the Aero's simply because there were so many posts claiming dramatic changes with break-in. I also measured harmonic distortion, and could find no differences before and after. If you heard less harshness with play time, and you don't think this was due to a change in frequency response per se, then it would pretty much have to be due to lower distortion. In any event, what was the list price of the Mezzo 2's before M-S went belly up?

 

[Paul_Apollonio]

You can't hear it is not the same as it's inaudible

I may be generalizing some, but for capacitors, as they are used in passive audio crossover filters, the type of material does not make an audible difference. A number of years ago, I participated in a large (for audio) blind test of over 40 listeners, and no one could hear the difference among crossovers made with non-polar electrolytic, various metalized polypropylene types, or exotic silver/oil-filled caps. As long as the caps had the correct capacitance value, there was no audible difference. And yes, there were some careful efforts at scientific controls in this test. From those results, I'm convinced that the idea of different sound due capacitor material is debunked as myth.

Yes, tolerance in caps, especially the lack of it in many inexpensive electrolytic caps, is important.

I remember the early 1980s and a company called DB systems that made preamps. Their claim to fame was having the lowest measured THD of anything on the market. I believe the owner was Dave Hadaway (best guess) who when asked to defend his practice of making the THD so low as something that would provide an audible benefit made this comment.

"There was never any shortage of people willing to claim any given level of distortion was inaudible".

Now forgetting for the time being that we are different in a number of ways including our sensitivity to distortion, it is extremely arrogant for one human being to declare categorically this or that is inaudible. It almost always means, 'well I didn't hear it, so if I didn't you can't, therefore it's inaudible'.

I guess that means the Edison phonograph is indistinguishable from the original. (There were a number of people who made this claim shortly after it was introduced.)

There are plenty of measurable effects one cannot hear without the correct excitation signal. The idea that you can simply dismiss audibility based on the failure of a group of people engaged in one experiment alone is short sighted. How do you know the experiment was not poorly designed? Because it produced no result? You can prove something is audible. Proving it is inaudible (by anyone) is quite another story. Suppose you test 100 people and nobody hears the effect. Then you test a person who hears it reliably over and over again.

Is the effect audible or inaudible?

Now if there were NO measurable metrics as to the phenomena of Dissipation factor (DF) or Equivalent Series Resistance (ESR) then perhaps you could from a purely theoretical standpoint make an argument that suggests if we can't measure something surely we have no difference to hear. Even that is highly questionable since advances consisting of eliminating defects nobody measured prior happen continuously. Before Dr Otala e.g., nobody was measuring T.I.M. distortion. Did that mean it did not exist or was inaudible? Of course not.

To suggest that a measurable effect is not audible to people who themselves have already heard the difference, and then follow it with "I may be generalizing some, but..." is a claim I take less than seriously.

 

[3db]

Man I love these images and they could easily fit into my article. Where are you guys when I need images????

Oh man, you don't want to know the images screaming through my mind.

 

[Hellcommute]

Hi I can't claim that my test with the Cambridge speakers can be generalized to other speakers. I picked the Aero's simply because there were so many posts claiming dramatic changes with break-in. I also measured harmonic distortion, and could find no differences before and after. If you heard less harshness with play time, and you don't think this was due to a change in frequency response per se, then it would pretty much have to be due to lower distortion. In any event, what was the list price of the Mezzo 2's before M-S went belly up?

The retail was 998$ before slashed for quick sale. MS was bought by the same parent that owns cambridge years ago, I was told anyway. They never rolled out another line as Cambridge became the focus.... Dont know how true it is though. I think they bought all the assets and intellectual property and gutted what was left.

 

[zieglj01]

The retail was 998$ before slashed for quick sale. MS was bought by the same parent that owns cambridge years ago, I was told anyway. They never rolled out another line as Cambridge became the focus.... Dont know how true it is though.

Audio Partnership owns Cambridge Audio, Mordaunt Short and Opus
Audio Partnership - Home

Still at Wild West Electronics
Bookshelf Speakers - Elac, Mordaunt Short

 

[Swerd]

"There was never any shortage of people willing to claim any given level of distortion was inaudible".

Nor is there a shortage of individuals who claim they can hear effects from more questionable features (such as wires, insulation, etc.) than the THD measured from a preamp. That is the purpose of audibility testing – to establish whether individual claims are or are not valid by testing a statistically significant larger population sample.

Does a measurable difference in physical parameters of an electronic component or a mechanical transducer translate into audibility by human listeners? This is the BIG QUESTION in audio, and few other than Toole & Olive have tackled it. This is as much a question of human physiology and behavior (sensory psychology) as it is a question of electronics and acoustics. But it is a testable question. Experimental answers will come as the results of listening tests. Measurements of physical parameters, no matter how carefully done, cannot answer this question.

Back to the specific question of whether different capacitors in loudspeaker passive crossover filters have an audible effect detectable by listeners. The fact remains that some tests were done that demonstrate their inaudibility. This is far from proof, but it remains the only demonstrated results that try to answer this question. No such tests have been done, as far as I know, that successfully demonstrate that capacitors do contribute to audible differences in speakers. The answer to this question incomplete, but the trend so far is clear.

“In theory there is no difference between theory and practice, but in practice there often is.”

 

[Irvrobinson]

“In theory there is no difference between theory and practice, but in practice there often is.”

Absolutely; it means the theory was flawed or incomplete, which I theorize happens a lot.

 

[Dennis Murphy]

Nor is there a shortage of individuals who claim they can hear effects from more questionable features (such as wires, insulation, etc.) than the THD measured from a preamp. That is the purpose of audibility testing – to establish whether individual claims are or are not valid by testing a statistically significant larger population sample.

Does a measurable difference in physical parameters of an electronic component or a mechanical transducer translate into audibility by human listeners? This is the BIG QUESTION in audio, and few other than Toole & Olive have tackled it. This is as much a question of human physiology and behavior (sensory psychology) as it is a question of electronics and acoustics. But it is a testable question. Experimental answers will come as the results listening tests. Measurements of physical parameters, no matter how carefully done, cannot answer this question.

Back to the specific question of whether different capacitors in loudspeaker passive crossover filters have an audible effect detectable by listeners. The fact remains that some tests were done that demonstrate their inaudibility. This is far from proof, but it remains the only demonstrated results that try to answer this question. No such tests have been done, as far as I know, that sucessfully demonstrate that capacitors do contribute to audible differences in speakers. The answer to this question incomplete, but the trend so far is clear.

“In theory there is no difference between theory and practice, but in practice there often is.”

I think that's a good statement of the current state of affairs. Gene and I had a private conversation about NPE's and passive crossovers recently, which led me to read up a little on the relevant literature. I must say that most seemingly informed opinions seem to agree with Gene--NPE's, particularly the kind often used in less expensive loudspeakers, are not thought to be appropriate for passive crossover networks, and they claim the negative effects can be readily measured. I'm not really clear on what all of these effects are--ESR (series resistance) is higher for NPE's, and the problem increases with increasing frequency, and may increase over time. However, there are NPE's around that are rated as low ESR. I just don't know how much lower, or how they compare with poly's. But assuming the NPE's specified as low are low enough not to be problematic, are there are issues with NPE's that would create non-linearity in crossover performance? Perhaps Gene can chime in here. I would be most concerned with performance in woofer circuits at cross points in the 300 - 600 Hz, range since that's where the real cost savings would lie. There are so many decent low-uf poly's out there at reasonable prices to make it not worthwhile taking a chance with NPE's in tweeter circuits.

 

[Swerd]

I think that's a good statement of the current state of affairs. Gene and I had a private conversation about NPE's and passive crossovers recently, which led me to read up a little on the relevant literature. I must say that most seemingly informed opinions seem to agree with Gene--NPE's, particularly the kind often used in less expensive loudspeakers, are not thought to be appropriate for passive crossover networks, and they claim the negative effects can be readily measured. I'm not really clear on what all of these effects are--ESR (series resistance) is higher for NPE's, and the problem increases with increasing frequency, and may increase over time. However, there are NPE's around that are rate as low ESR. I just don't know how much lower, or how they compare with poly's. But assuming the NPE's specified as low are low enough not to be problematic, are there are issues with NPE's that would create non-linearity in crossover performance? Perhpas Gene can chime in here. I would be most concerned with performance in woofer circuits at cross points in the 300 - 600 Hz, range since that's where the real cost savings would lie. There are so many decent low-uf poly's out there at reasonable prices to make it worthwhile taking a chance with NPE's in tweeter circuits.

Dennis – Thanks for clarifying that.

If the ESR of crossover caps has an audible effect, it could be tested by assembling crossovers with a variety of different caps with the same µF value but with different ESR, and listening as they are installed in identical speakers. I wonder if anyone has done that.

My background is in a field of science where nearly everyone does experimental science. If someone claimed to be a theoretical molecular biologist, he'd be laughed at, and have a short career. As a student, I was always told to avoid quoting theory and instead taught to pose all questions as a testable experiment. As a result, I still bristle at arguments made in the absence of experimental evidence – such as here.

 

[gene]

Dennis – Thanks for clarifying that.

If the ESR of crossover caps has an audible effect, it could be tested by assembling crossovers with a variety of different caps with the same µF value but with different ESR, and listening as they are installed in identical speakers. I wonder if anyone has done that.

My background is in a field of science where nearly everyone does experimental science. If someone claimed to be a theoretical molecular biologist, he'd be laughed at, and have a short career. As a student, I was always told to avoid quoting theory and instead taught to pose all questions as a testable experiment. As a result, I still bristle at arguments made in the absence of experimental evidence – such as here.

This is NOT rocket science. Data sheets are readily available comparing metrics of various capacitor types showing ESR and DF vs frequency. No legitimately good speaker designer is putting electrolytic caps in series with a tweeter anymore. Those that use electrolytics often bypass them with smaller value NPO caps to achieve better results.

Why would one want to make a less reliable, less precise design simply b/c their limited listening test, testing a limited sample size, on a limited audience may not have produced sonically verifiable results? As an engineer I always designed things to be accurate and consistent, using Monte Carlo analysis to ensure my designs always met min target goals.

Look at the impedance graphs in my Infinity P363 review and note how the 2 speakers don't measure exactly, esp around resonance.

http://www.audioholics.com/tower-speaker-reviews/primus-p363/P63_impedance_compare.JPG/image_view_fullscreen

Simply using better parts and tighter QC for a few bucks more would have solved that. Dumbing down the science to cheapen a product is not a good design choice IMO but I guess that is why there are so many choices on the market.

I prefer better parts when they are offered.

 

[Paul_Apollonio]

Back to the specific question of whether different capacitors in loudspeaker passive crossover filters have an audible effect detectable by listeners. The fact remains that some tests were done that demonstrate their inaudibility. This is far from proof, but it remains the only demonstrated results that try to answer this question. No such tests have been done, as far as I know, that successfully demonstrate that capacitors do contribute to audible differences in speakers. The answer to this question incomplete, but the trend so far is clear.
”

Demonstrate inaudibility? That means the test does not reveal detection. This is not proof something is inaudible, it's evidence that under certain conditions we can't hear "___"

As you pointed out properly, there is no shortage of people willing to proclaim they can hear bats singing. True, and I get that. If I'd not heard the differences myself I'd not take a side in this debate...(and make no mistake, relative to the differences between air and magnetic core inductors differences between good electrolytics and mylar capacitors are small ones. Small, but there nonetheless). This was pretty well settled by Jung and Marsh about 1980. Now show us a retraction of the claim by either of these men, and we can call that evidence (not proof) of inaudibility.

You're trying to prove a negative with evidence. Now with a mountain of evidence, you may bring me to your side. The problem as I see it is that you don't have a mountain. You've got a molehill, and you can't remember the name of the mole.

Yes, I would concede there are a great number of people claiming they can hear the effect who cannot for the sake of their egos. I don't have the mountain of evidence in my personal arsenal to display either. Since I was able to hear differences I found no need to seek it out.

What I do have is the evidence of my senses vs your argument that I should believe the effects of ESR and DF are not audible because someone ran an experiment where you and some other people didn't hear these differences. I believe you didn't hear differences. You don't believe I can. I'm ok with that. You can believe whatever you believe and it really doesn't matter to the rest of us because the evidence of our own senses must always take precedence over the evidence of someone elses senses. To deny that would be to place faith over reason.

BTW, almost all the time products fail with defective capacitors, the ones replaced are almost always electrolytics. Assuming for arguments sake alone the consumer cannot hear the difference, he/she still has good reason (reliability and MTBF) to prefer better film capacitors are used whenever possible. Now that is just common sense.

 

[Swerd]

This is NOT rocket science…

Gene – This seems like a classic case of miscommunication.

I do not advocate using non-polar electrolytic (NPE) capacitors in speaker crossovers. And I don’t believe I ever said that in this thread.

There are several practical reasons against using them; they are often made with capacitance values outside their ±10% tolerance values; their capacitance can drift over time; and now, if I understand correctly, their ESR and DF values also argue against their use. There are plenty of metalized polypropylene (MPP) caps available, at moderate cost, which appear to avoid all these issues. I see no reason not to use them in passive crossovers.

Although I never tried to make a case for using NPEs, I did point out that in a controlled listening test, no listeners could successfully distinguish speakers with any of the caps that were used in the test. It was noted after that test that using NPEs (specifically those sold at the time under the brand name Dayton) posed a problem because as many as 10 or 20 caps of the same nominal value had to be tested before a pair could be found with matching capacitance at the target value. It was cheaper and easier to buy a single pair of MPP caps. I don't recall if the ESR or DF values of any caps were considered for that test.

All the listeners at the test were amateur DIY speaker builders of varying background and speaker designing experience. There were at least 3 individuals present who were involved in designing the test who had advanced degrees, either PhD or MSEE. However, none of them were rocket scientists .

The purpose of this and other similar tests (comparing various kinds of resistors) was to try to determine if various types of capacitors or resistors could make an audible difference when used in speaker crossovers. If I recall, the cap test was repeated 3 times in different locations with different participants, all with similar results. I don’t remember details of the resistor test.

Both versions of these test ended up exploring just what is the smallest change in otherwise similar speakers that listeners can reliably hear. What is the lower limit of detection in such a test? It was recognized that we didn’t know that. If we did know, we could directly compare that to the results with the caps and resistors.

Gene & Paul – I don’t doubt what you say about the ESR and DF measurements of NPE caps. I’m not an EE so I can’t discuss the significance of those measurements with you. But I am well versed in the subject of what is and isn’t valid scientific evidence. And further, I know what kind of conclusions can be made with various kinds of evidence. If you want to claim that capacitors with certain ESR and DF metrics make poor sounding crossovers, you have to perform listening tests with them. Short of that, if you want to claim that those poor metrics make a convincing argument against ever using NPEs, I have no problem with that.

It comes down to differences in philosophy. Some people prefer “the best available” materials regardless of cost, and others want to use what is “good enough without being overkill”. Both are valid approaches.

 

[gene]

Gene – This seems like a classic case of miscommunication.

I do not advocate using non-polar electrolytic (NPE) capacitors in speaker crossovers. And I don’t believe I ever said that in this thread.

There are several practical reasons against using them; they are often made with capacitance values outside their ±10% tolerance values; their capacitance can drift over time; and now, if I understand correctly, their ESR and DF values also argue against their use. There are plenty of metalized polypropylene (MPP) caps available, at moderate cost, which appear to avoid all these issues. I see no reason not to use them in passive crossovers.

Although I never tried to make a case for using NPEs, I did point out that in a controlled listening test, no listeners could successfully distinguish speakers with any of the caps that were used in the test. It was noted after that test that using NPEs (specifically those sold at the time under the brand name Dayton) posed a problem because as many as 10 or 20 caps of the same nominal value had to be tested before a pair could be found with matching capacitance at the target value. It was cheaper and easier to buy a single pair of MPP caps. I don't recall if the ESR or DF values of any caps were considered for that test.

All the listeners at the test were amateur DIY speaker builders of varying background and speaker designing experience. There were at least 3 individuals present who were involved in designing the test who had advanced degrees, either PhD or MSEE. However, none of them were rocket scientists .

The purpose of this and other similar tests (comparing various kinds of resistors) was to try to determine if various types of capacitors or resistors could make an audible difference when used in speaker crossovers. If I recall, the cap test was repeated 3 times in different locations with different participants, all with similar results. I don’t remember details of the resistor test.

Both versions of these test ended up exploring just what is the smallest change in otherwise similar speakers that listeners can reliably hear. What is the lower limit of detection in such a test? It was recognized that we didn’t know that. If we did know, we could directly compare that to the results with the caps and resistors.

Gene & Paul – I don’t doubt what you say about the ESR and DF measurements of NPE caps. I’m not an EE so I can’t discuss the significance of those measurements with you. But I am well versed in the subject of what is and isn’t valid scientific evidence. And further, I know what kind of conclusions can be made with various kinds of evidence. If you want to claim that capacitors with certain ESR and DF metrics make poor sounding crossovers, you have to perform listening tests with them. Short of that, if you want to claim that those poor metrics make a convincing argument against ever using NPEs, I have no problem with that.

It comes down to differences in philosophy. Some people prefer “the best available” materials regardless of cost, and others want to use what is “good enough without being overkill”. Both are valid approaches.

I understand what you're saying and yes I believe there was a bit of miscommunication going on. No worries.

That being said, I always error on the side of caution and over design. That is the engineering nature in me.

I will tell you this, the very best loudspeakers I've listened to and measured have ALWAYS used the very highest caliber parts. The ones that were designed to be "good enough", especially regarding cabinet bracing, crossover topologies and drivers have never been what I considered state of the art sonically or especially how they measured.