Why build speakers?

[Buckle-meister]

I'm unsure what your specific question is that you are asking.

What I was alluding to in post #8 was that one would expect the level of R&D to increase (broadly) in line with the cost of a commercially manufactured speaker. Thus, excepting 'kit' speakers, it must become increasingly difficult for the DIY enthusiast to create speakers of a quality exceeding or, now even matching the performance of similarly high priced speakers. Turning this around, this would suggest that DIYers maximise their investment with low to medium quality, i.e. priced, speakers.

Yes/no?

[speaker manufacturers] obviously also have their marketing and R&D overheads to cover (which wouldn't be a factor to a DIY builder).

My problem is in accepting that 90%+ of DIYers can match the knowledge, experiance and general know-how of commercial speaker manufacturers. Remember, I'm not talking about 'kit' speakers; that's a separate issue.

Although we may be preaching to the choir, together, we might convince Buckle-meister of the wisdom of our ways.

I don't consider you to be preaching. Your words are always welcome.

 

[WmAx]

My problem is in accepting that 90%+ of DIYers can match the knowledge, experiance and general know-how of commercial speaker manufacturers. Remember, I'm not talking about 'kit' speakers; that's a separate issue.

I think first of all, it's important to put all this into perspective. The true average DIYer builds a coffee table or a subwoofer box for his car(surely you have known a few people like this in person, I have), not hi-fi speakers of his own design. Consider the actual number of DIYers with sufficient ability to do any level of succesful design. I think that there are very few people that qualify. I believe that the majority of those that can, are concentrated on the DIY forums. It might seem like a lot of people superficially when you read forums like PESUPPORT, MADISOUND or DIYAUDIO, but then when you divide this among the population numbers, I think it will be an insignificant number. Even if you multiply the number of people that actually post by a factor of 10, it is still insignificant when divided among the population. How many people have you met in person by chance that have such ability? I have met exactly zero by chance, in person. Those I have met have been online, or those I have met in person, were a result of meeting online first.

As for commercial speakers, I expect it should not be difficult for an average speaker DIYer with some experience to end up with speakers that sound better as compared to the average offering in a mass retailer or a DIYer at a rather advanced level being able to match the standard offerings as found in a low to mid-level hi-fi shop. The offerings in the mass market unit are severely flawed in one or more areas in order to meet the very low price point requirement.

From looking around on the popular DIY forums such as DIYAUDIO.COM, MADISOUND.COM and PESUPPORT.COM, I have drawn the conclusion that the vast majority of people designing their own speakers have very little knowledge of the founding perceptual research required to design a very high quality loudspeaker, regardless of the technical knowledge they may have gathered --- but often even that is very questionable. A few of the DIYers demonstrate exceptional technical ability and knowledge concerning specific physical behaviour(s) of loudspeakers, but then these select few DIY people also seem to rarely have much knowledge of the perceptual research. To have a high probability of chance of consistently designing extremely high quality(I mean at or near the highest sound quality possible within the constraints of real rooms and recordings) loudspeakers for use in real rooms and for the purpose of playback enjoyment requires both areas of knowledge.

-Chris

 

[Buckle-meister]

Yes/no?

...I expect it should not be difficult for an average speaker DIYer with some experience to end up with speakers that sound better as compared to the average offering in a mass retailer or a DIYer at a rather advanced level being able to match the standard offerings as found in a low to mid-level hi-fi shop.

I'll interpret this as a 'yes' then. If it takes a DIYer at an advanced level to create a speaker equal to that found in a low to mid-level hi-fi shop, then logically it'd take a DIYer at an exceptional level to create something better.

 

[gmichael]

Didn't a few of the speaker makers out there start off by building their first pair at home. Then improvements to improvements. And they told two friends, and they told two friends, .......

 

[sploo]

...Turning this around, this would suggest that DIYers maximise their investment with low to medium quality, i.e. priced, speakers.

Yes/no?

That is assuming there's a direct link between quality and price - there's a lot of overpriced stuff out there. Having said that, I'm sure an experienced commercial organisation can produce something better than any DIYer. The question is - how much better? If you have, say, $100k to spend, then fine. But if that system has $10k in drivers, I do wonder how close you could get - especially considering you're saving $90k.

...I have drawn the conclusion that the vast majority of people designing their own speakers have very little knowledge of the founding perceptual research required to design a very high quality loudspeaker

Chris, I'm very interested in your thoughts here.

Do you consider (often copied) designs such as the ProAc Response 2.5 or these (http://www.seas.no/kits main page.htm) designs by Seas to be medium or high quality? How would you rate this (http://vonschweikert.com/vr9se.html)? Oh, and what's your thoughts on Siegfried Linkwitz?

Is the point you're making that a good speaker is one that's well designed in terms of driver selection, crossover design and box construction, but a great speaker is designed with an understanding of real world performance in an average room?

Either way, I'd be interested if you could recommend any publications or papers that would be useful in learning more about these techniques.

 

[Buckle-meister]

...I'm sure an experienced commercial organisation can produce something better than any DIYer. The question is - how much better?

I'm not sure it matters really. Better is better.

Even if the DIYer can buy the same (or better) parts of a speaker for less than the cost of that speaker from a commercial manufacturer who, as you rightly point out has added overheads, the DIYer still requires the knowledge of how to integrate them in order to obtain better performance than the commercially manufactured speaker.

http://vonschweikert.com/vr9se.html

It's like a Dalek.

 

[sploo]

I'm not sure it matters really. Better is better.

Yes, but $90k better? In the kind of rooms most people are likely to have, I expect there's a limit to what you can achieve in terms of sound quality. Of course, if you're loaded, then you've probably got a huge house, with an acoustically excellent room, and spending major cash on great speakers isn't a problem. Not me though!

It's like a Dalek.

Yea... that was exactly my first thought.

Having read on though, it looks like quite an interesting unit. Would love to hear a pair.

 

[WmAx]

Do you consider (often copied) designs such as the ProAc Response 2.5

The ProAc is not what I consider a well functioning speaker. Not only is the overall response poor, and off axis response average, it uses a cabinet that I suspect is highly resonant, thus distorting the timbre.

or these (http://www.seas.no/kits main page.htm) designs by Seas to be medium or high quality?

The Thor design has a superb crossover for it's design objective. The crossover was designed by one of the most qualified crossover engineers in the world, after all. The cabinet would appear to be your standard resonant offering, however. I personally do not favor the chosen off axis response objective.

How would you rate this (http://vonschweikert.com/vr9se.html)?

This speaker is a different league of engineering than the last two. A very substantial cabinet, that is [probably] not audibly resonant, this is one of the most important features usually not given much attention by manufacturers(or DIYers). I am not familiar with 3rd party measurements of this product, however, so far as crossover execution. The rear radiating tweeter is a big plus for standard stereo recordings, but it is a super tweeter; I would have used a wider bandwidth range rear firing driver. I also would have used a wider horiztonal radiation tweeter system.

Oh, and what's your thoughts on Siegfried Linkwitz?

Of course, he is an engineer of 1st class qualification. His executions of his design objectives are as ideal as is probably possible. But I do not tend to prefer his long held tradition of not radiating high frequencies to the rear. He recently changed this, and now his Orion design has been revised to use a rear-firing tweeter.

Is the point you're making that a good speaker is one that's well designed in terms of driver selection, crossover design and box construction, but a great speaker is designed with an understanding of real world performance in an average room?

Exactly. Though I should note that I rarely see a DIY or Hi-Fi speaker with what I believe is probably an inaudible cabinet, even if other attributes are desirable.

Either way, I'd be interested if you could recommend any publications or papers that would be useful in learning more about these techniques.

My knowledge on this subject is a result of years of reading credible published papers and articles, along with my application of said research, along with limited perceptual control testing on myself(usually via PC ABX) and test subjects(usually via single blind testing). There are certain narrow areas that I must also speculate, due to lack of credible perceptual research in that area, or my inability to measure for a specific artifact in a practical situation(some things require an anechoic chamber with a very broad bandwidth effective range to measure accurately) If you note a specific parameter, I can probably find the relevant perceptual research on that specific parameter and refer you to the relevant publication. But I can not recommend just a few specific papers in a general sense of how to make a high grade loudspeaker. The relevant parameters are not so easily distilled to just a few papers.

-Chris

 

[WmAx]

I'll interpret this as a 'yes' then. If it takes a DIYer at an advanced level to create a speaker equal to that found in a low to mid-level hi-fi shop, then logically it'd take a DIYer at an exceptional level to create something better.

To do so with any degree of reliability, yes. But there is always the [very] improbable circumstance where someone stumbles across such a design by dumb luck.

-Chris

 

[Swerd]

What I was alluding to in post #8 was that one would expect the level of R&D to increase (broadly) in line with the cost of a commercially manufactured speaker. Thus, excepting 'kit' speakers, it must become increasingly difficult for the DIY enthusiast to create speakers of a quality exceeding or, now even matching the performance of similarly high priced speakers. Turning this around, this would suggest that DIYers maximise their investment with low to medium quality, i.e. priced, speakers.

Yes/no?

My problem is in accepting that 90%+ of DIYers can match the knowledge, experiance and general know-how of commercial speaker manufacturers. Remember, I'm not talking about 'kit' speakers; that's a separate issue.

I’ll have to answer that with a qualified no.

I think it is too optimistic to assume that “the level of R&D increases in line with the cost of a commercially manufactured speaker” for all speaker manufacturers. It very well might be true for some manufacturers, but in my limited experience it is clearly not true for others. These commercial makers might know quite a lot more than amateurs about speaker design, but often something gets lost in the manufacturing, distribution, and marketing of their products.

About five years ago I was searching for new speakers, and was told by one dealer at the time that there were some 450 speaker manufacturers around. And although there were many expensive speakers available, many were plagued with clearly identifiable problems. I suspect that there are just as many, if not more, manufacturers around today. The competition among them is intense, especially considering that most buyers today would rather spend more money on a large screen TV and less on speakers, regardless of their quality. As a result, in many cases, marketing issues rule over engineering issues. Therefore, I believe that a DIY enthusiast, of sufficient experience and skill, could easily produce quality speakers at a significantly lower price than commercial speakers, regardless of the price range. Again, I remind you of the price range of my own direct experience. I don’t know of any amateurs who build speakers that resemble, in any way, commercial speakers selling in the $10,000 or higher price range.

 

[sploo]

Chris,

Some very interesting stuff, thanks.

Surely the issue of resonant cabinets can be resolved with a more sturdy box - thicker walls and more bracing? (i.e. it's not too much of a hard problem to solve)

I thought that off axis response was a function of the chosen drivers, or are there things you can do to improve this (or even screw it up)?

The benefits/use of rear radiating drivers are intruiging. It's not a subject I've seen in any DIY speaker literature. Do you have anything related to this?

When I posted the mail with Linkwitz's name I did spend a bit of time taking another look at his site, and had noticed his mod to the Orion system. I've not read up in detail what he claims the improvements are though.

Out of interest, what's your background? Are you a general audio/acoustics person, or do you design speakers for a living?

Is there any chance you could list what you think are the main differences/considerations between a good speaker and a great one? I'd be more than happy to spend time looking these things up to try and learn more (though I'm sure I'll pester you at some point for more info )

 

[WmAx]

Surely the issue of resonant cabinets can be resolved with a more sturdy box - thicker walls and more bracing? (i.e. it's not too much of a hard problem to solve)

It is simple when you do not consider the variables. If you add a brace at point 1 that is the central origin of resonance X, you effectively reduce the amount that this point can physically move, thus removing the central action of resonance X. But when you do this, you effectively have created additional resonance points centered between this brace and the nearest other fixed points. You end up with new resonance(s) at different frequencies. If you try to use the brute force method to reduce resonance(s) from what is probably an audible amount, would probably require a massively braced and high mass system. You are in effect mainly increasing mass to reduce the effective panel movement, since the driver outputs a practically fixed amount of energy. The much more efficient solution is a relatively low mass cabinet with a narrow, high amplitude resonance. This is one solution that results in mostly inaudible cabinet output, as is proven by Dr. Floyd Toole in his research, and was used at one point in Infinity's highest level speaker(which is no longer manufactured). But it is difficult for a DIYer to build a full size speaker with such a feature using conventional building methods. You can incorporate this method, but it is not nearly as simple as a standard box construction. You can also use a combination of lower mass, but still well braced systems, with large degrees damping, in other words, thermal conversion of mechanical energy. This is a costly prospect for a manufacturer, and a substantial added complexity for a DIYer. Another solution is a hybrid of various techniques. Witness B&Ws combination of single frequency centered resonance mid-range cabinet with high inherent self-damping, along with a suspension system to reduce the energy transmitted to the enclosure. Refer to B&W's top level speakers (N802, N801 and N800). All of these solutions are usually complex for a DIYER compared to the standard speaker cabinet system.

In a slight side issue, it is interesting to me that very few manufactures use internal acoustic damping material with a sufficiently wide bandwidth to remove the internal cabinet reflection(s) and resonance(s) related to wall spacing.

I thought that off axis response was a function of the chosen drivers, or are there things you can do to improve this (or even screw it up)?

It is a function of the driver in part, but it is also a function of the crossover. You can change the off axis response of a driver if you change it's effective radiation area or use a wave guide, or mount it in a unique orientation(refer to Mirage's upwards pointed speakers on their lower end OmniStats). Each solution has negative side effects that must be considered for a specific application.

The benefits/use of rear radiating drivers are intriguing. It's not a subject I've seen in any DIY speaker literature. Do you have anything related to this?

This comes down to psycho-acoustics. The human brain interprets certain delays(in specific bands) as different effects, dependant on the time delay vs. the primary signal. Using a rear radiating high frequency driver creates a phantom source reflected off of the back wall that returns the listener after the direct speaker signal. The delay time determines the effect. Make no mistake, this is a coloration, but it almost always results in substantially increased sound quality with almost all listeners(97%), regardless of musical preferences or background. The former head researcher at Mirage(Ian Paisley) demonstrated this at the NRC on thousands of test subjects in double-blinded listening trials. You can read an over-view of this research on Mirage's FAQ pages. I have also tried different radiation patterns, on myself, and test subjects, in as controlled a manner as possible(using absorption panels to control the radiation of a near omnipolar HF driver, but keeping all other variables constant). Of course, successfully using an omnipolar system in a normal room may be difficult, due to the required minimum distances from the rear walls, and ideally symmetrical reflection points on either side. In many standard arrangements, omnipolar response may not be preferred.

If you want specific example(s) and/or solution(s) to specific issue(s) as mentioned above, please PM me.

-Chris

 

[Swerd]

WmAx

I've never considered the whole issue of removing or minimizing cabinet resonances. How can one measure cabinet resonances? And is it possible to minimize them with notch filters built into a crossover network?

The much more efficient solution [by Floyd Toole] is a relatively low mass cabinet with a narrow, high amplitude resonance.

At what frequency or range of freqs was this resonance?

You can also use a combination of lower mass, but still well braced systems, with large degrees damping, in other words, thermal conversion of mechanical energy.

Besides glass fiber, what kind of materials do this thermal conversion?

 

[WmAx]

WmAx

I've never considered the whole issue of removing or minimizing cabinet resonances. How can one measure cabinet resonances?

Measuring the actual audio output ratio vs. main signal of speaker, of cabinet resonances, requires a broad bandwidth anechoic chamber. A very long gate for an impulse measurement. FFT calculation to derive the resonances to useful format such as a CSD plot. Standard gated measurements in a normal room are useless, because you have to measure at a distance and have high resolution at the frequencies of interest for cabinet resonances (100Hz-1000Hz). The other method is calculation of the output of the panel resonances. But the mathematics do to this accurately would be daunting, and I don't know of a consumer available software package that is designed to do such calculation(s). One possibly practical method to do in a normal room is not in any way simple or quick, but should work: measure response of a speaker from the listening position acoustically. At the same time, measure a fixed point on the panel with an accelerometer to note band response and magnitude. Remove the mid-range driver from the cabinet. Now replace with a special blank front plate that has a driver firing into the cabinet. Now re-measure and adjust the volume so that the magnitude on the accelerometer reading is equal to the previous measurement. Now use an equalization compensation to adjust the response band to equal that of the previous measurement. Now re-take an acoustical measurement with this blank front transducer, and you will get a response plot of the cabinet itself in the room. You overlay the acoustic response of the speaker system with the cabinet response and then correlate this with the established resonance audibility research [1]data from Dr. Floyd Toole. The most likely complication with this method is getting a small driver system to fire into the cabinet with the blank front that has substantial enough acoustic output to get reliable readings.

And is it possible to minimize them with notch filters built into a crossover network?

No. Doing so would cause substantial dips in the driver/system response.

At what frequency or range of freqs was this resonance?

I do not recall.

Besides glass fiber, what kind of materials do this thermal conversion?

I believe you are referring to acoustic wave reflected signal attenuation. This is not the same. To dampen the panel resonances, you need to use a high mass loss material. Dynamat(often used in car audio) is such a material. Several roofing materials share similar properties, and are at a much lower cost. There is an article in the AH technical archive that has a case study of using such roofing material for this purpose. But do not mistake this as an all-encompassing solution. This is merely one method to further reduce resonances, it will not make an already substantially resonant cabinet inaudible. Constrained layer damping will do substantially more, but adds more complexity.

-Chris

Footnotes
[1] "The Modification of Timbre by Resonances: Perception and Measurement", Floyd Toole, Sean Olive, JAES, Vol. 36, No. 3, 1988, March, pages 122-141

 

[jaxvon]

I thought I'd add some of my personal experience regarding omni-polar vs. mono-polar radiation. I did a direct A/B comparison with some nearly omni speakers, using one of my favorite songs, Deacon Blues by Steely Dan, from the Aja album. Everything was the same except the backwave and first reflection points were treated with absorption for the mono-polar test. The difference was astounding. The stereo image depth and, for lack of a better term, 'fullness', were completely destroyed. With the mono-polar radiation (what any standard speaker is), the sound was good, and the image had good width. With the omni radiation, the stereo image took on depth and clarity unheard in the other setup. The mono-polar setup presented the music; the omni setup transported me to the original performance.

Edit: Here are the links to the articles WmAx mentioned:

http://www.audioholics.com/techtips/audioprinciples/loudspeakers/mechanicalnoisefloor.php
http://www.audioholics.com/techtips/audioprinciples/loudspeakers/MechanicalNoiseLoudspeaker.php

 

[dave1490]

All,

Over time I've seen various threads about folk who're thinking of, or are in the process of building their own speakers.

I appreciate that some want to do so purely out of enjoyment of DIY and others to better understand how speakers work; I myself fall within the latter category.

These are fine reasons to build speakers. However, I can't quite fathom how anyone, barring those who truly know what they're doing, expects to make a speaker (assuming it's not simply a kit) of comparable (or better) sound quality to that made by a manufacturer. After all, a manufacturer does this for a living.

I'm pretty sure that there's more to building speakers than simply purchasing high quality drivers and fitting them within a cabinet. I've seen it written that crossover design forms an important part of, and influences the sound of the speakers. However, unless I'm mistaken, most people do not test their completed designs in an anechoic chamber, nor measure the completed speaker's performance.

Assuming this to be true (correct me if I'm wrong), isn't the building of speakers an iterative process of design, manufacture and measurement? Thus, in terms of sound quality, aren't these people wasting their time? If they don't measure their completed speakers performance, aren't they effectively relying on blind luck that the speaker they build will exhibit the desireable qualities of a quality loudspeaker?

i think this is true,not to mention that the equipment needed to test it.i allways stay with the same manufacturer when building speakers.you can always mount a heil amt with a 3in excurion woofer.number,s look good but i woudnt trust it.

 

[jonnythan]

I think some people put *way* too much faith into the "knowledge and experience" of the pro speaker design guys.

I've learned in more than one area that things you don't know how to do can seem like black magic, but once you get a handle on what's going on it becomes pretty straightforward.

 

[Swerd]

Chris & jaxvon

Thanks for the explanations and the links. I hadn't noticed those two AH articles by Mark Sanfilipo. I'll print them and read them in detail to see what I can absorb.

Now that crossover design is no longer such a "black art", reducing cabinet resonances is the next aspect for DIYers to tackle. Improved or simplified measurement methods will allow more amateurs to address this.

 

[sploo]

It is simple when you do not consider the variables. If you add a brace at point 1 that is the central origin of resonance X, you effectively reduce the amount that this point can physically move, thus removing the central action of resonance X. But when you do this, you effectively have created additional resonance points centered between this brace and the nearest other fixed points. You end up with new resonance(s) at different frequencies.

Yes, of course... I hadn't thought about that. It's not dissimilar to creating harmonics on a guitar string I guess.

The first thing that came to my mind was to use accelerometers on a prototype box in order to measure vibration. Having read more on your replies, and the two articles by Mark Sanfilipo (thanks jaxvon) it sounds possible.

Would it be worth playing different frequencies and measuring the accelerometer output at different points on a cabinet? My thinking is that you could add bracing in those areas with the largest movement. I accept you wouldn't be separating cabinet movement from speaker movement, or speaker movement from cabinet pressure.

The much more efficient solution is a relatively low mass cabinet with a narrow, high amplitude resonance.

Surely a narrow, high amplitude resonance would be very audible, or is the point that it's moved to a frequency above human hearing? (or instead that having a single spike is better than numerous frequencies being affected?)

Witness B&Ws combination of single frequency centered resonance mid-range cabinet with high inherent self-damping, along with a suspension system to reduce the energy transmitted to the enclosure. Refer to B&W's top level speakers (N802, N801 and N800). All of these solutions are usually complex for a DIYER compared to the standard speaker cabinet system.

Interesting. My 'specialization' is that I have a CNC router, CAD packages and plenty of will to cut MDF. My goal is to create curved and/or visually interesting cabinets - such that I believe I do have the ability to make things beyond the usual cuboid boxes. Whilst I don't pretend for a second that I'm going to get to the level of the experts, if I could determine some basic shapes/sizes/bracing designs that would aid overall sound quality I'm more than happy to have a go at cutting them.

It is a function of the driver in part, but it is also a function of the crossover. You can change the off axis response of a driver if you change it's effective radiation area or use a wave guide, or mount it in a unique orientation(refer to Mirage's upwards pointed speakers on their lower end OmniStats). Each solution has negative side effects that must be considered for a specific application.

Is that why some speakers have drivers on a sloping baffle? It always struck me as a little odd, as my understanding was always that the best sound came from the center of a driver, and (within reason) you don't want to be listening to off axis sound (either direct or reflected).

This comes down to psycho-acoustics....

Very interesting. I need to have a good read about Linkwitz's rear treble on the Orion, and the Mirage FAQ you mention.

Certainly my room (and most listening rooms of people I know) are too small for these techniques - given the figures I've read of recommended spacing. Nonetheless, it's tempting to chuck another treble in the back of a cabinet and just have a listen to the effect - even if that's a completely crude way to work.

If you want specific example(s) and/or solution(s) to specific issue(s) as mentioned above, please PM me.

Sounds like I need to work out what my questions are first! Many thanks.

 

[jaxvon]

Chris & jaxvon

Thanks for the explanations and the links. I hadn't noticed those two AH articles by Mark Sanfilipo. I'll print them and read them in detail to see what I can absorb.

Now that crossover design is no longer such a "black art", reducing cabinet resonances is the next aspect for DIYers to tackle. Improved or simplified measurement methods will allow more amateurs to address this.

I'm not sure what exact roofing product Mark was using, but I use Peel 'n Seal, a tar/aluminum foil product that comes in 6"x25' rolls and costs under $14 at Lowes for a roll. I've gone into basic treatment of existing speakers in more detail in this post:

It seems though that the best option for constructing a non-resonant cabinet is to use a constrained layer of damping material. I have not tried constructing a cabinet like this yet (mostly due to fund). However, since I know you're a DIY kind of guy, it might be a fun experiment.