I'll drink to that! That's exactly why I recrouted my friend Mirko to develop my speakers. At the time, he had a company doing that and had like 15 or so models at various levels under his belt, and they did him honor. I would not have dared to go for it myself, my knowledge of loudspeakers is far too limited. By myself, I could never have pulled it off this good, even with much luck.
Absolutely. I have seen some people dabble in speaker design and simply not put in the hard yards, as if designing speakers is a lesser discipline than what goes before it and feeds it and that it's just an art-form. I sometime also get the same impression about cable design, know what I mean? Everybody can solder (they think) wires together and voila! I am in the audio business.
I can only say that in the end the only way is to learn how to measure and model loudspeakers, and that is not for the fainthearted. I can construct crossovers before I build them with surprisingly great accuracy, not absolute mind you, but get me where I need to be, complete the loop, start in measurement, then modelling and then come back to measurement to confirm. That takes a LOT of applied discipline, let alone experience and having well sorted aims in mind.
But have you also noted that very few loudspeaker designers know much about amplifier design? There are few who design both. I think that is highly problematic when we speak of the marriage of power amps and speakers. There is still a lot to be learnt there.
One is that I am very interested in looking at this from a "current model" perspective rather than the "voltage model" which I believe has lead to lazy design. This also comes back to feedback where it seems to be the only way to create a low enough Z to become a voltage source. There is much work that needs to be done to understand what the current does, but it is so damn non-intuitive - hence we fall back on "voltage" look-see. I have done real blind tests (so blind that the person was even aware he was a subject) and proven that current contamination in crossovers are extremely audible. It shouldn't surprise since dynamic drivers/motors are current devices and it is current that makes them move. But current driving resonant systems can't be done (?) - you may be surprised, but there are techniques available.
Cheers, Joe
PS: Which is more important, the box design (make it stiff, solid and heavy) or crossover design? Guess which usually gets the vote - and wrong!
Last edited:
It can be true, and untrue as well. Sometimes, i have the feeling that what you name "science" is a new religion for naive believers...
We have to stay modest about what we know, (here what happens with transients reproduction in a speaker's membrane, room acoustic, and our ears+brain system).
Most of the speakers on the market, including very expensive "high end" ones, are very bad, even those with nice looking response curves.
You can provide me a lot of response curves on and off axis, polar curves, waterfall, square waves, distortion curves etc... it will tell-me less about a speaker than ten minutes of listening of several well known sources.
After a life dedicated in those questions (music recording and reproduction) i know one or two tips about what works better (anyway, in an imperfect way). And, at the end, even if i spend hours in intensive measurements, a good result can only be achieved by methods that looks more like cook recipes realized by a "chef" using his "taste" and smell, feelings and experience.
Last edited:
The mathematics and the formulas we use in astronomy, by example, are rough approximations of the reality. Because suns and planets are never spherical and of an uniform density. yes, we can succeed to land a robot on a comet at the other side of the solar system. But only after a lot of trajectory corrections. Groping and DIY ;-)
More than this, most of the scientific discovery are not the result of scientific reasoning. But by a mix of observation and intuition (from Newton to Einstein).
The last discoveries done by science in the last decades don't increase a lot our knowledge or understanding of the physical phenomenas, but just our evaluation of what we know and can observe: Most of the astrophysicists think we know and can observe something around 10 or 20% of the mass of the universe, and have no idea of the 80% remaining. (dark energy, dark matter, string theory...)
In terms of music, the laws of harmony are mathematical. Good musicians do not make calculations when they compose or improvise. And blues-men don't need to count in their mind the twelve bars. In recording studio, guitarists achieve a more accurate tuning of their instrument with their ears than with a guitar tuner or frequency meter. Because the strings and guitar necks are not perfect, and the pressure of the fingers not uniform, as loudspeakers are far from perfection.
More than this, most of the scientific discovery are not the result of scientific reasoning. But by a mix of observation and intuition (from Newton to Einstein).
The last discoveries done by science in the last decades don't increase a lot our knowledge or understanding of the physical phenomenas, but just our evaluation of what we know and can observe: Most of the astrophysicists think we know and can observe something around 10 or 20% of the mass of the universe, and have no idea of the 80% remaining. (dark energy, dark matter, string theory...)
In terms of music, the laws of harmony are mathematical. Good musicians do not make calculations when they compose or improvise. And blues-men don't need to count in their mind the twelve bars. In recording studio, guitarists achieve a more accurate tuning of their instrument with their ears than with a guitar tuner or frequency meter. Because the strings and guitar necks are not perfect, and the pressure of the fingers not uniform, as loudspeakers are far from perfection.
Last edited:
I have worked with some orbital mechanics guys. All PhD mathematicians. Yes its approximations, but damned good ones and all based very firmly in science. Are you saying that the people who worked out the window for Voyager launches were going on gut feel an intuition that the mutiple sling shots would work?
Flipping your statement around on speakers. I am sure no one disagrees that there is no substitute for experience in any field (the old 10,000 hours being the difference between a novice and an expert). but experience with no basic understanding of the fundamental principles is the brownian motion way of doing it unless every speaker is basically the same.
Take for example the andrew jones pioneer bookshelf speakers. An artisan would unlikely have managed to deliver that performance for that low cost. That is the combination of a good engineering mind, application of a lot of manufactuing disciplines and decades of experience to know which compromises would work.
Another flip. Why are most boutique UK turntables and speakers basically wood (MDF or ply). Cos you can knock them out in a garage.
Flipping your statement around on speakers. I am sure no one disagrees that there is no substitute for experience in any field (the old 10,000 hours being the difference between a novice and an expert). but experience with no basic understanding of the fundamental principles is the brownian motion way of doing it unless every speaker is basically the same.
Take for example the andrew jones pioneer bookshelf speakers. An artisan would unlikely have managed to deliver that performance for that low cost. That is the combination of a good engineering mind, application of a lot of manufactuing disciplines and decades of experience to know which compromises would work.
Another flip. Why are most boutique UK turntables and speakers basically wood (MDF or ply). Cos you can knock them out in a garage.
@Joe Rasmussen
In my view, the blame for mediocre amp-speaker interface lies on both sides.
Speaker manufacturers ASSUME the user will have a behemoth amplifier good enough to weld with it, so they just go out and do their thing the way it suits them and let the amp deal with it.
Amplifier manufacturers in he consumer sector are usually constricted with product end prices, and far too often tend to "save" money in just the wrong places. As a result, many amps are current starved if they face a low impednce complx load. The High End sector is in a much better position in that respect.
All this is hardly new. Over the last few months, I was able to renew my acquaintance with the venerable AR 3a Improved. I mention it not because I think it's the best speaker ever, but because it always had a well earned reputation of being bitch to drive. Use a typical commercial mid price amp and chances are the sound will collapse if it's pushed a little harder. Only a few (relative to the overall supply) of the better made amps we have seen over the decades were in fact capable of doing more loudness without sound degradation. You may remember the Infinity Reference speakers which were NEVER demonstrated without a Mark Levinson power amp because so few could do them justice.
Today, that has fortunately changed (somewhat). Current knowledge recognises the fact that a good amp muct be capable of handling low impedance complex loads at least in short time bursts of power, as normal music signals would demand assuming the amp is driven with pulsating music signals rather than laboratory sine waves as a constant signal source.
I believe, right or wrong, that it's the speaker manufacturers who generally lag behind, The current price wars have sadyl shortened the development time of speakers, it seems it's usually assumed that all you need do is to use quality drivers from people like ScanSpeak, Dynaudio, etc., slap the together and voila!, you have a credible product. Just have the copywriters harp on about your quality parts and cabinet manufacturing prowess and real your rewards. Surely an absurdity paradix, since speaker drivers have progressed more than any other field of audio over the last 40 years or so. Insufficnet development in "voicing" speakers results in many product falling short of their true potentials.
In my view, the blame for mediocre amp-speaker interface lies on both sides.
Speaker manufacturers ASSUME the user will have a behemoth amplifier good enough to weld with it, so they just go out and do their thing the way it suits them and let the amp deal with it.
Amplifier manufacturers in he consumer sector are usually constricted with product end prices, and far too often tend to "save" money in just the wrong places. As a result, many amps are current starved if they face a low impednce complx load. The High End sector is in a much better position in that respect.
All this is hardly new. Over the last few months, I was able to renew my acquaintance with the venerable AR 3a Improved. I mention it not because I think it's the best speaker ever, but because it always had a well earned reputation of being bitch to drive. Use a typical commercial mid price amp and chances are the sound will collapse if it's pushed a little harder. Only a few (relative to the overall supply) of the better made amps we have seen over the decades were in fact capable of doing more loudness without sound degradation. You may remember the Infinity Reference speakers which were NEVER demonstrated without a Mark Levinson power amp because so few could do them justice.
Today, that has fortunately changed (somewhat). Current knowledge recognises the fact that a good amp muct be capable of handling low impedance complex loads at least in short time bursts of power, as normal music signals would demand assuming the amp is driven with pulsating music signals rather than laboratory sine waves as a constant signal source.
I believe, right or wrong, that it's the speaker manufacturers who generally lag behind, The current price wars have sadyl shortened the development time of speakers, it seems it's usually assumed that all you need do is to use quality drivers from people like ScanSpeak, Dynaudio, etc., slap the together and voila!, you have a credible product. Just have the copywriters harp on about your quality parts and cabinet manufacturing prowess and real your rewards. Surely an absurdity paradix, since speaker drivers have progressed more than any other field of audio over the last 40 years or so. Insufficnet development in "voicing" speakers results in many product falling short of their true potentials.
Whether loudspeaker design is based on knowledge rather than science, that is what triggered it. My position is that it is based more on knowledge than science. The same can probably be said about amplifier design, although as I stated before, scientific knowledge does play its role.
Engineering is not science. Engineers rape and pillage science, use it as a cookbook, and mix it in with experience, ancient knowledge and rules of thumb to make stuff work.
The cryogenic hardening which came up earlier is a good example. It was applied for ages because it was known to work. Then NASA engineers started to ponder the effects of extreme temperatures on metals, and found out that cryogenic treatment homogenizes the hardness in metals. They came up with a protocol for cryogenic hardening, which subsequently became the departing point for all kinds of secret recipes. All this has the appearance of science, because you can repeat procedures, and measure repeatable results.
But, real understanding how it works in the scientific sense of the word is lacking.
Much of metallurgy is that way. The development of alloys is more like mixing perfumes than an exact science. And so it goes in other engineering disciplines.
There are many forms of Engineering. But in general scientists view their work as an end in itself. Engineers see it as a means to and end and work out what useful things they can make with the science.
In all areas there are exceptions. Look at CERN. Was Atlas built with rules of thumb and ancient knowledge or properly engineered by people with multiple disciplines at their fingertips?
Speaker design outside of a select few companies is hobby stuff.
Vacuphile,
I'll call my cousin and tell him that his PhD in metallurgy is just a piece of paper that lets him dabble in alchemy, he will appreciate that! He was one of those the Air Force would call when a plane went down and they had to piece that together and solve the mystery. It is BS that there is no science in speaker design, that is only the common misconception brought about by those who do indeed do nothing but rebox speakers designed by others, the majority of this industry.
To discount magnetic modeling, materials science, acoustic measurement and many other aspects of loudspeaker design is to turn this into a hobby as it is here mostly on this site. The endless combinations of existing commercial speakers is what we have here, very little scientific endeavor. Yes there is a part that is art, that has to do with some of the limitations of many current commercial materials that are used to make the individual components of a speaker, that does not mean it can't be a much more rigorous science that what you see commonly described in the many threads on this site.
I'll call my cousin and tell him that his PhD in metallurgy is just a piece of paper that lets him dabble in alchemy, he will appreciate that! He was one of those the Air Force would call when a plane went down and they had to piece that together and solve the mystery. It is BS that there is no science in speaker design, that is only the common misconception brought about by those who do indeed do nothing but rebox speakers designed by others, the majority of this industry.
To discount magnetic modeling, materials science, acoustic measurement and many other aspects of loudspeaker design is to turn this into a hobby as it is here mostly on this site. The endless combinations of existing commercial speakers is what we have here, very little scientific endeavor. Yes there is a part that is art, that has to do with some of the limitations of many current commercial materials that are used to make the individual components of a speaker, that does not mean it can't be a much more rigorous science that what you see commonly described in the many threads on this site.
quick update on the floating current generator approach
I finally got a clean-enough setup to get some decent data. To recall, I suggested that replacing the pullup and pulldown resistors in the simplified Vendetta-like preamp's folded-cascode circuit with a floating current generator could improve noise performance slightly, increase gain slightly, and reduce effects of the folded-cascode noise, without any deleterious effects except for added complexity.
Theory and experiment agree nicely. In this case the equivalent input noise voltage spectral density improved by 4.1%, going from 301 ohm resistors to batteries and a composite JFET current generator.
As further description is likely going to be lost in this crowded and ever-shifting thread, when I have time I will try to get round to a separate post. I think I will call the version the Vedanta preamp. Again, the aim is not to set people to building things that resemble it, but to demonstrate a technique that may be useful in general.
Brad
I finally got a clean-enough setup to get some decent data. To recall, I suggested that replacing the pullup and pulldown resistors in the simplified Vendetta-like preamp's folded-cascode circuit with a floating current generator could improve noise performance slightly, increase gain slightly, and reduce effects of the folded-cascode noise, without any deleterious effects except for added complexity.
Theory and experiment agree nicely. In this case the equivalent input noise voltage spectral density improved by 4.1%, going from 301 ohm resistors to batteries and a composite JFET current generator.
As further description is likely going to be lost in this crowded and ever-shifting thread, when I have time I will try to get round to a separate post. I think I will call the version the Vedanta preamp. Again, the aim is not to set people to building things that resemble it, but to demonstrate a technique that may be useful in general.
Brad
Bcarso, I think you did a unique modification to this low noise circuit. However, I think that just raising the voltage to +/- 30V would have given similar improvement, and it is much easier to implement. Still, you did find an effective way to add a current source instead of a resistive load.
Did you ever try to converge a 60's color TV while also swapping the three video output tubes to get the "best" picture? Was Muntz an engineer or a scientist?
One that always baffled me how did John Dahlquist design the DQ10, after all it did win lots of praise and sell.
It didn't exactly breach the realm of Beverage, etc., but I was interested in the design process hence my TV convergence analogy, once you have enough variables it becomes a random search.
BTW not a dream but a movie.
barriers to loudspeaker development
With electronics we have some very good instruments accessible to us, as well as simulation tools. With loudspeakers few have anechoic chambers at hand, and fewer still have really good simulation tools---they exist but are jealously guarded as proprietary. In both cases we have our ears and brains as the final arbiters.
Also, the effort involved in creating prototypes for electronics is significant, but pales compared to that required for the creation of loudspeaker prototypes.
Toole's book addresses some aspects of loudspeaker design, but for me the most interesting parts are about the loudspeaker-room interactions. It will be published as a third edition with some substantial changes (it's really the second edition, but a publisher snafu reprinted the first as a second ed.).
There will be an upcoming issue of the JAES featuring a lengthy invited paper from him, which I have yet to read but sounds promising.
With electronics we have some very good instruments accessible to us, as well as simulation tools. With loudspeakers few have anechoic chambers at hand, and fewer still have really good simulation tools---they exist but are jealously guarded as proprietary. In both cases we have our ears and brains as the final arbiters.
Also, the effort involved in creating prototypes for electronics is significant, but pales compared to that required for the creation of loudspeaker prototypes.
Toole's book addresses some aspects of loudspeaker design, but for me the most interesting parts are about the loudspeaker-room interactions. It will be published as a third edition with some substantial changes (it's really the second edition, but a publisher snafu reprinted the first as a second ed.).
There will be an upcoming issue of the JAES featuring a lengthy invited paper from him, which I have yet to read but sounds promising.
Without any form of requirements to put a stake in the ground yes it becomes random. Usually it seems boutique speaker designers have some sort of pet technology or configuration they like to push. a 'house fetish' .
The DL10 looks bonkers, but if you look at the linkwitz pages on the LX521 you can see the method in the madness. SL does admit he came up with his baffle shapes empirically!
The DL10 looks bonkers, but if you look at the linkwitz pages on the LX521 you can see the method in the madness. SL does admit he came up with his baffle shapes empirically!
- Status
- Not open for further replies.